Simple line layout: Add performance test with varying string and font-size to minimiz...
[WebKit-https.git] / PerformanceTests / ChangeLog
1 2017-02-05  Zalan Bujtas  <zalan@apple.com>
2
3         Simple line layout: Add performance test with varying string and font-size to minimize width-cache hit.
4         https://bugs.webkit.org/show_bug.cgi?id=167859
5         <rdar://problem/30368070>
6
7         Reviewed by Antti Koivisto.
8
9         This is in preparation for webkit.org/b/167843.
10
11         * Layout/simple-line-layout-with-varying-content.html: Added.
12
13 2017-01-16  Filip Pizlo  <fpizlo@apple.com>
14
15         Add framerate baselines to the splay-detail test
16         https://bugs.webkit.org/show_bug.cgi?id=167659
17
18         Reviewed by Geoffrey Garen.
19
20         * JetStream/Octane2/splay-detail.html:
21
22 2017-01-31  Carlos Alberto Lopez Perez  <clopez@igalia.com>
23
24         [EFL][GTK] Skip some IndexedDB tests that time out on the Perf bots.
25
26         Unreviewed performance test gardening.
27
28         * Skipped:
29
30 2017-01-30  Zalan Bujtas  <zalan@apple.com>
31
32         Simple line layout: Small tweaks to improve performance.
33         https://bugs.webkit.org/show_bug.cgi?id=167611
34         <rdar://problem/30274294>
35
36         Reviewed by Simon Fraser.
37
38         * Layout/simple-line-layout-non-repeating-text.html: Added.
39
40 2017-01-13  Said Abou-Hallawa  <sabouhallawa@apple.com>
41
42         26 MotionMark performance tests failing
43         https://bugs.webkit.org/show_bug.cgi?id=166854
44
45         Reviewed by Ryosuke Niwa.
46         
47         Skip running MotionMark on WebKit perf bots.
48
49         * Skipped:
50
51 2017-01-06  Said Abou-Hallawa  <sabouhallawa@apple.com>
52
53         Rename the directory and the files of MotionMark from Animometer to MotionMark
54         https://bugs.webkit.org/show_bug.cgi?id=166659
55
56         Reviewed by Jon Lee.
57         
58         Rename the directory and the files of the benchmark to its new name.
59
60         * MotionMark/developer.html: Renamed from PerformanceTests/Animometer/developer.html.
61         * MotionMark/index.html: Renamed from PerformanceTests/Animometer/index.html.
62         * MotionMark/resources/debug-runner/d3.min.js: Renamed from PerformanceTests/Animometer/resources/debug-runner/d3.min.js.
63         * MotionMark/resources/debug-runner/graph.js: Renamed from PerformanceTests/Animometer/resources/debug-runner/graph.js.
64         * MotionMark/resources/debug-runner/motionmark.css: Renamed from PerformanceTests/Animometer/resources/debug-runner/animometer.css.
65         * MotionMark/resources/debug-runner/motionmark.js: Renamed from PerformanceTests/Animometer/resources/debug-runner/animometer.js.
66         * MotionMark/resources/debug-runner/tests.js: Renamed from PerformanceTests/Animometer/resources/debug-runner/tests.js.
67         * MotionMark/resources/extensions.js: Renamed from PerformanceTests/Animometer/resources/extensions.js.
68         * MotionMark/resources/runner/benchmark-runner.js: Renamed from PerformanceTests/Animometer/resources/runner/benchmark-runner.js.
69         * MotionMark/resources/runner/crystal.svg: Renamed from PerformanceTests/Animometer/resources/runner/crystal.svg.
70         * MotionMark/resources/runner/lines.svg: Renamed from PerformanceTests/Animometer/resources/runner/lines.svg.
71         * MotionMark/resources/runner/logo.svg: Renamed from PerformanceTests/Animometer/resources/runner/logo.svg.
72         * MotionMark/resources/runner/motionmark.css: Renamed from PerformanceTests/Animometer/resources/runner/animometer.css.
73         * MotionMark/resources/runner/motionmark.js: Renamed from PerformanceTests/Animometer/resources/runner/animometer.js.
74         * MotionMark/resources/runner/tests.js: Renamed from PerformanceTests/Animometer/resources/runner/tests.js.
75         * MotionMark/resources/statistics.js: Renamed from PerformanceTests/Animometer/resources/statistics.js.
76         * MotionMark/resources/strings.js: Renamed from PerformanceTests/Animometer/resources/strings.js.
77         * MotionMark/tests/3d/resources/webgl.js: Renamed from PerformanceTests/Animometer/tests/3d/resources/webgl.js.
78         * MotionMark/tests/3d/webgl.html: Renamed from PerformanceTests/Animometer/tests/3d/webgl.html.
79         * MotionMark/tests/bouncing-particles/bouncing-canvas-images.html: Renamed from PerformanceTests/Animometer/tests/bouncing-particles/bouncing-canvas-images.html.
80         * MotionMark/tests/bouncing-particles/bouncing-canvas-shapes.html: Renamed from PerformanceTests/Animometer/tests/bouncing-particles/bouncing-canvas-shapes.html.
81         * MotionMark/tests/bouncing-particles/bouncing-css-images.html: Renamed from PerformanceTests/Animometer/tests/bouncing-particles/bouncing-css-images.html.
82         * MotionMark/tests/bouncing-particles/bouncing-css-shapes.html: Renamed from PerformanceTests/Animometer/tests/bouncing-particles/bouncing-css-shapes.html.
83         * MotionMark/tests/bouncing-particles/bouncing-svg-images.html: Renamed from PerformanceTests/Animometer/tests/bouncing-particles/bouncing-svg-images.html.
84         * MotionMark/tests/bouncing-particles/bouncing-svg-shapes.html: Renamed from PerformanceTests/Animometer/tests/bouncing-particles/bouncing-svg-shapes.html.
85         * MotionMark/tests/bouncing-particles/bouncing-tagged-images.html: Renamed from PerformanceTests/Animometer/tests/bouncing-particles/bouncing-tagged-images.html.
86         * MotionMark/tests/bouncing-particles/resources/bouncing-canvas-images.js: Renamed from PerformanceTests/Animometer/tests/bouncing-particles/resources/bouncing-canvas-images.js.
87         * MotionMark/tests/bouncing-particles/resources/bouncing-canvas-particles.js: Renamed from PerformanceTests/Animometer/tests/bouncing-particles/resources/bouncing-canvas-particles.js.
88         * MotionMark/tests/bouncing-particles/resources/bouncing-canvas-shapes.js: Renamed from PerformanceTests/Animometer/tests/bouncing-particles/resources/bouncing-canvas-shapes.js.
89         * MotionMark/tests/bouncing-particles/resources/bouncing-css-images.js: Renamed from PerformanceTests/Animometer/tests/bouncing-particles/resources/bouncing-css-images.js.
90         * MotionMark/tests/bouncing-particles/resources/bouncing-css-shapes.js: Renamed from PerformanceTests/Animometer/tests/bouncing-particles/resources/bouncing-css-shapes.js.
91         * MotionMark/tests/bouncing-particles/resources/bouncing-svg-images.js: Renamed from PerformanceTests/Animometer/tests/bouncing-particles/resources/bouncing-svg-images.js.
92         * MotionMark/tests/bouncing-particles/resources/bouncing-svg-particles.js: Renamed from PerformanceTests/Animometer/tests/bouncing-particles/resources/bouncing-svg-particles.js.
93         * MotionMark/tests/bouncing-particles/resources/bouncing-svg-shapes.js: Renamed from PerformanceTests/Animometer/tests/bouncing-particles/resources/bouncing-svg-shapes.js.
94         * MotionMark/tests/bouncing-particles/resources/bouncing-tagged-images.js: Renamed from PerformanceTests/Animometer/tests/bouncing-particles/resources/bouncing-tagged-images.js.
95         * MotionMark/tests/bouncing-particles/resources/image1.jpg: Renamed from PerformanceTests/Animometer/tests/bouncing-particles/resources/image1.jpg.
96         * MotionMark/tests/bouncing-particles/resources/image2.jpg: Renamed from PerformanceTests/Animometer/tests/bouncing-particles/resources/image2.jpg.
97         * MotionMark/tests/bouncing-particles/resources/image3.jpg: Renamed from PerformanceTests/Animometer/tests/bouncing-particles/resources/image3.jpg.
98         * MotionMark/tests/bouncing-particles/resources/image4.jpg: Renamed from PerformanceTests/Animometer/tests/bouncing-particles/resources/image4.jpg.
99         * MotionMark/tests/bouncing-particles/resources/image5.jpg: Renamed from PerformanceTests/Animometer/tests/bouncing-particles/resources/image5.jpg.
100         * MotionMark/tests/dom/compositing-transforms.html: Renamed from PerformanceTests/Animometer/tests/dom/compositing-transforms.html.
101         * MotionMark/tests/dom/focus.html: Renamed from PerformanceTests/Animometer/tests/dom/focus.html.
102         * MotionMark/tests/dom/leaves.html: Renamed from PerformanceTests/Animometer/tests/dom/leaves.html.
103         * MotionMark/tests/dom/particles.html: Renamed from PerformanceTests/Animometer/tests/dom/particles.html.
104         * MotionMark/tests/dom/resources/compositing-transforms.js: Renamed from PerformanceTests/Animometer/tests/dom/resources/compositing-transforms.js.
105         * MotionMark/tests/dom/resources/dom-particles.js: Renamed from PerformanceTests/Animometer/tests/dom/resources/dom-particles.js.
106         * MotionMark/tests/dom/resources/focus.js: Renamed from PerformanceTests/Animometer/tests/dom/resources/focus.js.
107         * MotionMark/tests/dom/resources/leaves.js: Renamed from PerformanceTests/Animometer/tests/dom/resources/leaves.js.
108         * MotionMark/tests/master/canvas-stage.html: Renamed from PerformanceTests/Animometer/tests/master/canvas-stage.html.
109         * MotionMark/tests/master/focus.html: Renamed from PerformanceTests/Animometer/tests/master/focus.html.
110         * MotionMark/tests/master/image-data.html: Renamed from PerformanceTests/Animometer/tests/master/image-data.html.
111         * MotionMark/tests/master/leaves.html: Renamed from PerformanceTests/Animometer/tests/master/leaves.html.
112         * MotionMark/tests/master/multiply.html: Renamed from PerformanceTests/Animometer/tests/master/multiply.html.
113         * MotionMark/tests/master/resources/canvas-stage.js: Renamed from PerformanceTests/Animometer/tests/master/resources/canvas-stage.js.
114         * MotionMark/tests/master/resources/canvas-tests.js: Renamed from PerformanceTests/Animometer/tests/master/resources/canvas-tests.js.
115         * MotionMark/tests/master/resources/compass.svg: Renamed from PerformanceTests/Animometer/tests/master/resources/compass.svg.
116         * MotionMark/tests/master/resources/compass100.png: Renamed from PerformanceTests/Animometer/tests/master/resources/compass100.png.
117         * MotionMark/tests/master/resources/console.svg: Renamed from PerformanceTests/Animometer/tests/master/resources/console.svg.
118         * MotionMark/tests/master/resources/console100.png: Renamed from PerformanceTests/Animometer/tests/master/resources/console100.png.
119         * MotionMark/tests/master/resources/contribute.svg: Renamed from PerformanceTests/Animometer/tests/master/resources/contribute.svg.
120         * MotionMark/tests/master/resources/contribute100.png: Renamed from PerformanceTests/Animometer/tests/master/resources/contribute100.png.
121         * MotionMark/tests/master/resources/debugger.svg: Renamed from PerformanceTests/Animometer/tests/master/resources/debugger.svg.
122         * MotionMark/tests/master/resources/debugger100.png: Renamed from PerformanceTests/Animometer/tests/master/resources/debugger100.png.
123         * MotionMark/tests/master/resources/focus.js: Renamed from PerformanceTests/Animometer/tests/master/resources/focus.js.
124         * MotionMark/tests/master/resources/image-data.js: Renamed from PerformanceTests/Animometer/tests/master/resources/image-data.js.
125         * MotionMark/tests/master/resources/inspector.svg: Renamed from PerformanceTests/Animometer/tests/master/resources/inspector.svg.
126         * MotionMark/tests/master/resources/inspector100.png: Renamed from PerformanceTests/Animometer/tests/master/resources/inspector100.png.
127         * MotionMark/tests/master/resources/layout.svg: Renamed from PerformanceTests/Animometer/tests/master/resources/layout.svg.
128         * MotionMark/tests/master/resources/layout100.png: Renamed from PerformanceTests/Animometer/tests/master/resources/layout100.png.
129         * MotionMark/tests/master/resources/leaves.js: Renamed from PerformanceTests/Animometer/tests/master/resources/leaves.js.
130         * MotionMark/tests/master/resources/multiply.js: Renamed from PerformanceTests/Animometer/tests/master/resources/multiply.js.
131         * MotionMark/tests/master/resources/particles.js: Renamed from PerformanceTests/Animometer/tests/master/resources/particles.js.
132         * MotionMark/tests/master/resources/performance.svg: Renamed from PerformanceTests/Animometer/tests/master/resources/performance.svg.
133         * MotionMark/tests/master/resources/performance100.png: Renamed from PerformanceTests/Animometer/tests/master/resources/performance100.png.
134         * MotionMark/tests/master/resources/script.svg: Renamed from PerformanceTests/Animometer/tests/master/resources/script.svg.
135         * MotionMark/tests/master/resources/script100.png: Renamed from PerformanceTests/Animometer/tests/master/resources/script100.png.
136         * MotionMark/tests/master/resources/shortcuts.svg: Renamed from PerformanceTests/Animometer/tests/master/resources/shortcuts.svg.
137         * MotionMark/tests/master/resources/shortcuts100.png: Renamed from PerformanceTests/Animometer/tests/master/resources/shortcuts100.png.
138         * MotionMark/tests/master/resources/standards.svg: Renamed from PerformanceTests/Animometer/tests/master/resources/standards.svg.
139         * MotionMark/tests/master/resources/standards100.png: Renamed from PerformanceTests/Animometer/tests/master/resources/standards100.png.
140         * MotionMark/tests/master/resources/storage.svg: Renamed from PerformanceTests/Animometer/tests/master/resources/storage.svg.
141         * MotionMark/tests/master/resources/storage100.png: Renamed from PerformanceTests/Animometer/tests/master/resources/storage100.png.
142         * MotionMark/tests/master/resources/styles.svg: Renamed from PerformanceTests/Animometer/tests/master/resources/styles.svg.
143         * MotionMark/tests/master/resources/styles100.png: Renamed from PerformanceTests/Animometer/tests/master/resources/styles100.png.
144         * MotionMark/tests/master/resources/svg-particles.js: Renamed from PerformanceTests/Animometer/tests/master/resources/svg-particles.js.
145         * MotionMark/tests/master/resources/text.js: Renamed from PerformanceTests/Animometer/tests/master/resources/text.js.
146         * MotionMark/tests/master/resources/timeline.svg: Renamed from PerformanceTests/Animometer/tests/master/resources/timeline.svg.
147         * MotionMark/tests/master/resources/timeline100.png: Renamed from PerformanceTests/Animometer/tests/master/resources/timeline100.png.
148         * MotionMark/tests/master/svg-particles.html: Renamed from PerformanceTests/Animometer/tests/master/svg-particles.html.
149         * MotionMark/tests/master/text.html: Renamed from PerformanceTests/Animometer/tests/master/text.html.
150         * MotionMark/tests/resources/main.js: Renamed from PerformanceTests/Animometer/tests/resources/main.js.
151         * MotionMark/tests/resources/math.js: Renamed from PerformanceTests/Animometer/tests/resources/math.js.
152         * MotionMark/tests/resources/stage.css: Renamed from PerformanceTests/Animometer/tests/resources/stage.css.
153         * MotionMark/tests/resources/star.svg: Renamed from PerformanceTests/Animometer/tests/resources/star.svg.
154         * MotionMark/tests/resources/yin-yang.png: Renamed from PerformanceTests/Animometer/tests/resources/yin-yang.png.
155         * MotionMark/tests/resources/yin-yang.svg: Renamed from PerformanceTests/Animometer/tests/resources/yin-yang.svg.
156         * MotionMark/tests/simple/resources/simple-canvas-paths.js: Renamed from PerformanceTests/Animometer/tests/simple/resources/simple-canvas-paths.js.
157         * MotionMark/tests/simple/resources/simple-canvas.js: Renamed from PerformanceTests/Animometer/tests/simple/resources/simple-canvas.js.
158         * MotionMark/tests/simple/resources/tiled-canvas-image.js: Renamed from PerformanceTests/Animometer/tests/simple/resources/tiled-canvas-image.js.
159         * MotionMark/tests/simple/simple-canvas-paths.html: Renamed from PerformanceTests/Animometer/tests/simple/simple-canvas-paths.html.
160         * MotionMark/tests/simple/tiled-canvas-image.html: Renamed from PerformanceTests/Animometer/tests/simple/tiled-canvas-image.html.
161         * MotionMark/tests/template/resources/template-canvas.js: Renamed from PerformanceTests/Animometer/tests/template/resources/template-canvas.js.
162         * MotionMark/tests/template/resources/template-css.js: Renamed from PerformanceTests/Animometer/tests/template/resources/template-css.js.
163         * MotionMark/tests/template/resources/template-svg.js: Renamed from PerformanceTests/Animometer/tests/template/resources/template-svg.js.
164         * MotionMark/tests/template/template-canvas.html: Renamed from PerformanceTests/Animometer/tests/template/template-canvas.html.
165         * MotionMark/tests/template/template-css.html: Renamed from PerformanceTests/Animometer/tests/template/template-css.html.
166         * MotionMark/tests/template/template-svg.html: Renamed from PerformanceTests/Animometer/tests/template/template-svg.html.
167
168 2016-12-13  Brady Eidson  <beidson@apple.com>
169
170         Add a PerfTest targeting IDBObjectStore.get().
171         https://bugs.webkit.org/show_bug.cgi?id=165816
172
173         Reviewed by Alex Christensen.
174
175         * IndexedDB/objectstore-get.html: Added.
176
177 2016-12-13  Brady Eidson  <beidson@apple.com>
178
179         Add a PerfTest targeting IDBIndex.get().
180         https://bugs.webkit.org/show_bug.cgi?id=165803
181
182         Reviewed by Alex Christensen.
183
184         * IndexedDB/index-get.html: Added.
185
186 2016-12-12  Brady Eidson  <beidson@apple.com>
187
188         More IndexedDB perf tests.
189         https://bugs.webkit.org/show_bug.cgi?id=165634
190
191         Reviewed by Sam Weinig.
192
193         * IndexedDB/index-multientry.html:
194         * IndexedDB/large-number-of-inserts-responsiveness.html: Added.
195         * IndexedDB/large-number-of-inserts.html: Added.
196         * IndexedDB/objectstore-cursor.html: Added. Runtime cut in 1/4th compared to r209672.
197
198         * resources/runner.js:
199
200 2016-12-12  Commit Queue  <commit-queue@webkit.org>
201
202         Unreviewed, rolling out r209672.
203         https://bugs.webkit.org/show_bug.cgi?id=165766
204
205         IndexedDB/objectstore-cursor.html perf test always times out
206         (Requested by ap on #webkit).
207
208         Reverted changeset:
209
210         "More IndexedDB perf tests."
211         https://bugs.webkit.org/show_bug.cgi?id=165634
212         http://trac.webkit.org/changeset/209672
213
214 2016-12-10  Brady Eidson  <beidson@apple.com>
215
216         More IndexedDB perf tests.
217         https://bugs.webkit.org/show_bug.cgi?id=165634
218
219         Reviewed by Sam Weinig.
220
221         * IndexedDB/index-multientry.html:
222         * IndexedDB/large-number-of-inserts-responsiveness.html: Added.
223         * IndexedDB/large-number-of-inserts.html: Added.
224         * IndexedDB/objectstore-cursor.html: Added.
225         
226         * resources/runner.js: Add "track responsiveness" functionality to PerfTestRunner.
227
228 2016-12-06  Filip Pizlo  <fpizlo@apple.com>
229
230         Concurrent GC should be stable enough to land enabled
231         https://bugs.webkit.org/show_bug.cgi?id=164990
232
233         Reviewed by Geoffrey Garen.
234         
235         Made CDjs more configurable and refined the "large.js" configuration. I was using that one and
236         the new "long.js" configuration to tune concurrent eden GCs.
237         
238         Added a new way of running Splay in browser, which using chartjs to plot the execution times of
239         2000 iterations. This includes the minified chartjs.
240
241         * JetStream/Octane2/splay-detail.html: Added.
242         * JetStream/cdjs/benchmark.js:
243         (benchmarkImpl):
244         (benchmark):
245         * JetStream/cdjs/long.js: Added.
246
247 2016-12-07  Brady Eidson  <beidson@apple.com>
248
249         Add IDB perf tests stressing key size.
250         https://bugs.webkit.org/show_bug.cgi?id=165567
251
252         Reviewed by Alex Christensen.
253
254         * IndexedDB/large-array-keys.html: Added.
255         * IndexedDB/large-binary-keys.html: Added.
256         * IndexedDB/large-string-keys.html: Added.
257
258 2016-12-05  Brady Eidson  <beidson@apple.com>
259
260         Add an IndexedDB perf test to PerformanceTests.
261         https://bugs.webkit.org/show_bug.cgi?id=165430
262
263         Reviewed by Alex Christensen.
264
265         * IndexedDB/index-multientry.html: Added.
266
267 2016-12-02  Filip Pizlo  <fpizlo@apple.com>
268
269         ES6SampleBench should report an average for Steady State so that all of the numbers are comparable
270         https://bugs.webkit.org/show_bug.cgi?id=165325
271
272         Reviewed by Saam Barati.
273
274         This makes all of the numbers that ES6SampleBench reports comparable to each other: they all speak of the time
275         it took to run an iteration of something.
276
277         * ES6SampleBench/results.js:
278         (Results.prototype.reportResult):
279
280 2016-11-18  Jer Noble  <jer.noble@apple.com>
281
282         Add some performance tests for various aspects of media loading and playback.
283         https://bugs.webkit.org/show_bug.cgi?id=164977
284
285         Reviewed by Darin Adler.
286
287         Add new tests for a few aspects of media playback; namely: how quickly  media elements fire
288         the "canplaythrough" event when all data is immediately available, how quickly HLS playback
289         switches resolutions, and at what playback rate MSE-backed video can play without dropping
290         frames. Skip these tests by default, as they're not runnable on all ports, and they require
291         a webserver (run-webkit-httpd) to complete on macOS.
292
293         * Media/HLSCanPlayThrough.html: Added.
294         * Media/HLSGearChange.html: Added.
295         * Media/MSECanPlayThrough.html: Added.
296         * Media/MSEPlaybackRate.html: Added.
297         * Media/StandardCanPlayThrough.html: Added.
298         * Media/hls/1080p/iframe_index.m3u8: Added.
299         * Media/hls/1080p/prog_index.m3u8: Added.
300         * Media/hls/1080p/test.ts: Added.
301         * Media/hls/480p/iframe_index.m3u8: Added.
302         * Media/hls/480p/prog_index.m3u8: Added.
303         * Media/hls/480p/test.ts: Added.
304         * Media/hls/720p/iframe_index.m3u8: Added.
305         * Media/hls/720p/prog_index.m3u8: Added.
306         * Media/hls/720p/test.ts: Added.
307         * Media/hls/720p/test.ts.back: Added.
308         * Media/hls/index.m3u8: Added.
309         * Media/media-source-loader.js: Added.
310         (MediaSourceLoader):
311         (MediaSourceLoader.prototype.loadManifest):
312         (MediaSourceLoader.prototype.loadManifestSucceeded):
313         (MediaSourceLoader.prototype.loadManifestFailed):
314         (MediaSourceLoader.prototype.loadMediaData):
315         (MediaSourceLoader.prototype.loadMediaDataSucceeded):
316         (MediaSourceLoader.prototype.loadMediaDataFailed):
317         (MediaSourceLoader.prototype.get type):
318         (MediaSourceLoader.prototype.get duration):
319         (MediaSourceLoader.prototype.get initSegment):
320         (MediaSourceLoader.prototype.get mediaSegmentsLength):
321         (MediaSourceLoader.prototype.mediaSegments):
322         (MediaSourceLoader.prototype.get everyMediaSegment):
323         * Media/test-fragmented-video.json: Added.
324         * Media/test-fragmented-video.mp4: Added.
325         * Media/test.mp4: Added.
326         * Skipped:
327
328 2016-11-18  Filip Pizlo  <fpizlo@apple.com>
329
330         Concurrent GC should be able to run splay in debug mode and earley/raytrace in release mode with no perf regression
331         https://bugs.webkit.org/show_bug.cgi?id=164282
332
333         Reviewed by Geoffrey Garen and Oliver Hunt.
334         
335         CDjs is a fun benchmark for stressing concurrent GCs, but to really give the GC a good
336         workout you need to increase the amount of work that the test does. This adds a second
337         configuration of the benchmark that has more aircraft. It uses much more memory and causes us
338         to do more GCs and those GCs take longer.
339
340         * JetStream/cdjs/benchmark.js:
341         (benchmarkImpl):
342         (benchmark):
343         * JetStream/cdjs/large.js: Added.
344
345 2016-11-14  Filip Pizlo  <fpizlo@apple.com>
346
347         Unreviewed, revert unintended change.
348
349         * ES6SampleBench/Air/benchmark.js:
350
351 2016-10-03  Saam Barati  <sbarati@apple.com>
352
353         MapHash should speculate on the type of its child node
354         https://bugs.webkit.org/show_bug.cgi?id=161922
355
356         Reviewed by Filip Pizlo.
357
358         I gave the main generator a name so it's easier to see what
359         it is when using the sampling profiler.
360
361         * ES6SampleBench/Basic/ast.js:
362         (Basic.Program):
363
364 2016-09-19  Sergio Villar Senin  <svillar@igalia.com>
365
366         [css-grid] Remove the x2 computation of row sizes with indefinite heights
367         https://bugs.webkit.org/show_bug.cgi?id=162150
368
369         Reviewed by Darin Adler.
370
371         Added a new test case which checks the layout performance of grids inside other grids, i.e,
372         grids acting both as grid container and grid item.
373
374         * Layout/nested-grid.html: Added.
375
376 2016-09-09  Simon Fraser  <simon.fraser@apple.com>
377
378         Perf test Animation/css-accelerated-animation.html failing
379         https://bugs.webkit.org/show_bug.cgi?id=161795
380
381         Skip Animation/css-accelerated-animation.html since this only works as a MobileSafari
382         perf test.
383
384         * Skipped:
385
386 2016-09-08  Simon Fraser  <simon.fraser@apple.com>
387
388         Add a content-animation test with accelerated CSS animations
389         https://bugs.webkit.org/show_bug.cgi?id=161776
390
391         Reviewed by Dean Jackson.
392
393         Add a test that measures the frame rate of accelerated CSS animations.
394
395         This is like css-animation.html (which animates 'left' and 'top') but animates
396         transforms instead. In order to get animation in X and Y we make two nested elements
397         and set transformX() on one, and transformY() on the other.
398
399         * Animation/css-accelerated-animation.html: Added.
400
401 2016-08-23  Saam Barati  <sbarati@apple.com>
402
403         It should be easy to run ES6SampleBench from the jsc shell
404         https://bugs.webkit.org/show_bug.cgi?id=161085
405
406         Reviewed by Yusuke Suzuki.
407
408         This patch makes ES6 sample bench runnable through the `jsc` shell.
409         To do that, you need to be in the PerformanceTests/ES6SampleBench
410         directory and run `jsc cli.js`. To make this work, the benchmark
411         is now aware of being run in two modes: via the browser, and via
412         the shell. Each entry point will set a variable `isInBrowser`,
413         and the benchmark will do different things based on if that
414         variable is true or false.
415
416         * ES6SampleBench/Air/benchmark.js:
417         * ES6SampleBench/Air/stress-test.js:
418         * ES6SampleBench/Basic/benchmark.js:
419         (runBenchmark):
420         * ES6SampleBench/Basic/stress-test.js:
421         * ES6SampleBench/air_benchmark.js:
422         * ES6SampleBench/basic_benchmark.js:
423         * ES6SampleBench/cli.js: Added.
424         (return.doRun):
425         (makeBenchmarkRunner):
426         * ES6SampleBench/driver.js:
427         (Driver):
428         (Driver.prototype._recomputeSummary):
429         (Driver.prototype._iterate.window.setTimeout):
430         (Driver.prototype._iterate):
431         (Driver.prototype._updateIterations):
432         * ES6SampleBench/glue.js:
433         (reportResult):
434         * ES6SampleBench/index.html:
435         * ES6SampleBench/results.js:
436         (Results):
437         (Results.prototype.reportRunning):
438         (Results.prototype.reportDone):
439         (Results.prototype.reportError):
440         * ES6SampleBench/stats.js:
441         (Stats):
442         (Stats.prototype.toString):
443         (Stats.prototype._update):
444
445 2016-08-01  Said Abou-Hallawa  <sabouhallawa@apple.com>
446
447         Add an option to run the MotionMark tests with classic tile size (512x512)
448         https://bugs.webkit.org/show_bug.cgi?id=160371
449
450         Reviewed by Darin Adler.
451
452         Setting the body to large size (3000x3000) and overflow to scroll forces
453         the classic tile size (512x512). This option does not affect the canvas
454         tests because the canvas is displayed on one tile regardless of its size.
455
456         * Animometer/developer.html:
457         * Animometer/resources/debug-runner/animometer.css:
458         (body.showing-test-container.tiles-big):
459         (body.showing-test-container.tiles-classic):
460         * Animometer/resources/debug-runner/animometer.js:
461         * Animometer/resources/runner/animometer.js:
462         (window.benchmarkController.startBenchmark):
463
464 2016-06-29  Filip Pizlo  <fpizlo@apple.com>
465
466         Generators violate bytecode liveness validation
467         https://bugs.webkit.org/show_bug.cgi?id=159279
468
469         Reviewed by Yusuke Suzuki.
470
471         Add Basic to our test harness.
472
473         Also made some cosmetic changes to the benchmark harness.
474
475         * ES6SampleBench/Basic/basic-tests.yaml: Added.
476         * ES6SampleBench/Basic/stress-test.js: Added.
477         (preciseTime):
478         * ES6SampleBench/driver.js:
479         (Driver):
480         (Driver.prototype.start):
481         (Driver.prototype.reportError):
482         * ES6SampleBench/glue.js:
483         * ES6SampleBench/index.html:
484
485 2016-06-28  Filip Pizlo  <fpizlo@apple.com>
486
487         ES6SampleBench should have a harness
488         https://bugs.webkit.org/show_bug.cgi?id=159246
489
490         Reviewed by Saam Barati.
491         
492         This adds a simple web harness for ES6SampleBench. It runs Air and Basic 10 times for 200
493         iterations each and reports three metrics:
494         
495         First iteration: the time it takes for the first iteration to run. This is the first
496         iteration after the benchmark is loaded into the iframe, so it's representative of what
497         would happen if one of these workloads only ran for a short time.
498         
499         Worst 2%: of the last 199 iterations, the average of the worst 2% iterations. If code like
500         any of these workloads was used in an important event handler, you'd want that code to run
501         well in the worst case in addition to having great throughput.
502         
503         Steady state: the total of the last 199 iterations. This is representative of what would
504         happen if you ran code like this for a long time.
505         
506         The total score is the geomean of the firstIteration/worstCase/steadyState numbers of the
507         two benchmarks.
508         
509         The harness does statistics using Student's T-distribution confidence intervals.
510
511         * ES6SampleBench/Basic/benchmark.js:
512         (Benchmark):
513         * ES6SampleBench/air_benchmark.js: Added.
514         * ES6SampleBench/basic_benchmark.js: Added.
515         * ES6SampleBench/driver.js: Added.
516         (Driver):
517         (Driver.prototype.addBenchmark):
518         (Driver.prototype.start):
519         (Driver.prototype.reportResult):
520         (Driver.prototype.reportError):
521         (Driver.prototype._recomputeSummary.Geomean):
522         (Driver.prototype._recomputeSummary.Geomean.prototype.add):
523         (Driver.prototype._recomputeSummary.Geomean.prototype.get result):
524         (Driver.prototype._recomputeSummary):
525         (Driver.prototype._iterate):
526         (Driver.prototype._updateIterations):
527         * ES6SampleBench/glue.js: Added.
528         * ES6SampleBench/index.html: Added.
529         * ES6SampleBench/results.js: Added.
530         (Results):
531         (Results.prototype.get benchmark):
532         (Results.prototype.reset):
533         (Results.prototype.reportRunning):
534         (Results.prototype.reportDone):
535         (Results.prototype.reportResult.averageAbovePercentile):
536         (Results.prototype.reportResult):
537         (Results.prototype.reportError):
538         * ES6SampleBench/stats.js: Added.
539         (Stats):
540         (Stats.prototype.reset):
541         (Stats.prototype.add):
542         (Stats.prototype.get numIterations):
543         (Stats.prototype.valueForIteration):
544         (Stats.get result.tDist):
545         (Stats.prototype.get result):
546         (Stats.prototype.toString):
547         (Stats.prototype._update):
548         * ES6SampleBench/style.css: Added.
549         (body):
550         (body, th, tr):
551         (h1, h2, h3, h4):
552         (h1):
553         (h2):
554         (h3):
555         (hr):
556         (address):
557         (img):
558         (.underline):
559         (ol.loweralpha):
560         (ol.upperalpha):
561         (ol.lowerroman):
562         (ol.upperroman):
563         (ol.arabic):
564         (.banner-link:link, .banner-link:visited):
565         (:link, :visited):
566         (h4 :link, h4 :visited, h5 :link, h5 :visited, h6 :link, h6 :visited):
567         (.anchor:link, .anchor:visited):
568         (* > .anchor:link, * > .anchor:visited):
569         (span:hover .anchor):
570         (a.forbidden, span.forbidden):
571         (a.missing:hover):
572         (a.closed:link, a.closed:visited, span.closed):
573         (pre):
574         (div.code):
575         (div.code pre):
576         (dt):
577         (dd):
578         (dd:last-child):
579         (.site-logo):
580         (.site-logo .tagline):
581         (table):
582         (#contents):
583         (p):
584         (p:last-child):
585         (th):
586         (td):
587
588 2016-06-28  Jon Lee  <jonlee@apple.com>
589
590         Update focus test
591         https://bugs.webkit.org/show_bug.cgi?id=159242
592         rdar://problem/27070007
593
594         Reviewed by Dean Jackson.
595         Provisionally reviewed by Said Abou-Hallawa.
596
597         Move previous test to dom suite, and update the test for better reporting of frame rate, although
598         it uses a different rendering path.
599
600         * Animometer/resources/debug-runner/tests.js: Add to dom suite.
601         * Animometer/tests/dom/focus.html: Copied from PerformanceTests/Animometer/tests/master/focus.html.
602         * Animometer/tests/dom/resources/focus.js: Copied from PerformanceTests/Animometer/tests/master/resources/focus.js.
603         * Animometer/tests/master/focus.html: Remove center element.
604         * Animometer/tests/master/resources/focus.js: Use narrower size range with smaller particles. Remove the
605         container elements. Inline getBlurValue and getOpacityValue since they are only called once.
606
607 2016-06-28  Filip Pizlo  <fpizlo@apple.com>
608
609         Move Air.js and Basic into ES6SampleBench
610
611         Rubber stamped by Geoffrey Garen.
612         
613         We want to group our ES6 benchmarks together, and eventually, we want to give them a
614         common harness.
615
616         * Air.js: Removed.
617         * Air.js/README.md: Removed.
618         * Air.js/airjs-tests.yaml: Removed.
619         * Air.js/all.js: Removed.
620         * Air.js/allocate_stack.js: Removed.
621         * Air.js/arg.js: Removed.
622         * Air.js/basic_block.js: Removed.
623         * Air.js/benchmark.js: Removed.
624         * Air.js/code.js: Removed.
625         * Air.js/custom.js: Removed.
626         * Air.js/frequented_block.js: Removed.
627         * Air.js/insertion_set.js: Removed.
628         * Air.js/inst.js: Removed.
629         * Air.js/liveness.js: Removed.
630         * Air.js/make_dist.sh: Removed.
631         * Air.js/opcode.js: Removed.
632         * Air.js/payload-airjs-ACLj8C.js: Removed.
633         * Air.js/payload-gbemu-executeIteration.js: Removed.
634         * Air.js/payload-imaging-gaussian-blur-gaussianBlur.js: Removed.
635         * Air.js/payload-typescript-scanIdentifier.js: Removed.
636         * Air.js/reg.js: Removed.
637         * Air.js/stack_slot.js: Removed.
638         * Air.js/stress-test.js: Removed.
639         * Air.js/strip-hash.rb: Removed.
640         * Air.js/symbols.js: Removed.
641         * Air.js/test.html: Removed.
642         * Air.js/test.js: Removed.
643         * Air.js/tmp.js: Removed.
644         * Air.js/tmp_base.js: Removed.
645         * Air.js/util.js: Removed.
646         * Basic: Removed.
647         * Basic/ast.js: Removed.
648         * Basic/basic.js: Removed.
649         * Basic/benchmark.js: Removed.
650         * Basic/caseless_map.js: Removed.
651         * Basic/lexer.js: Removed.
652         * Basic/number.js: Removed.
653         * Basic/parser.js: Removed.
654         * Basic/random.js: Removed.
655         * Basic/state.js: Removed.
656         * Basic/test.html: Removed.
657         * Basic/test.js: Removed.
658         * Basic/util.js: Removed.
659         * ES6SampleBench: Added.
660         * ES6SampleBench/Air: Copied from PerformanceTests/Air.js.
661         * ES6SampleBench/Basic: Copied from PerformanceTests/Basic.
662         * Skipped:
663
664 2016-06-24  Filip Pizlo  <fpizlo@apple.com>
665
666         Add a ES6 generator benchmark
667         https://bugs.webkit.org/show_bug.cgi?id=159101
668
669         Rubber stamped by Keith Miller.
670         
671         This adds a Basic interpreter loosely based on ECMA-55:
672         
673         http://www.ecma-international.org/publications/files/ECMA-ST-WITHDRAWN/ECMA-55,%201st%20Edition,%20January%201978.pdf
674         
675         It includes a lexer that is a generator, a parser that uses regular expressions, and an AST
676         walk interpreter where the walking functions for statements are generators that call each
677         other with yield*. This enables the interpreter to look like an AST walk even though it can
678         yield at INPUT and PRINT statements.
679         
680         This also uses for-of, classes, Map, and WeakMap. It also uses deprecated-but-awesome RegExp
681         features like RegExp.lastMatch and RegExp.rightContext. I did it that way because this is
682         how I've always written lexers in dynamic languages; see offlineasm's lex() method in
683         parser.rb for example.
684         
685         The benchmark runs a handful of simple Basic programs. The longest-running one computes
686         prime numbers.
687         
688         Includes a command-line and web harness. On my machine it runs in 2-3 seconds.
689
690         * Basic: Added.
691         * Basic/ast.js: Added.
692         (Basic.NumberApply):
693         (Basic.Variable):
694         (Basic.Const):
695         (Basic.NumberPow):
696         (Basic.NumberMul):
697         (Basic.NumberDiv):
698         (Basic.NumberNeg):
699         (Basic.NumberAdd):
700         (Basic.NumberSub):
701         (Basic.StringVar):
702         (Basic.Equals):
703         (Basic.NotEquals):
704         (Basic.LessThan):
705         (Basic.GreaterThan):
706         (Basic.LessEqual):
707         (Basic.GreaterEqual):
708         (Basic.GoTo):
709         (Basic.GoSub):
710         (Basic.Def):
711         (Basic.Let):
712         (Basic.If):
713         (Basic.Return):
714         (Basic.Stop):
715         (Basic.On):
716         (sideState.shouldStop):
717         (Basic.For):
718         (Basic.Next):
719         (Basic.Print):
720         (Basic.Input):
721         (Basic.Read):
722         (Basic.Restore):
723         (Basic.Dim):
724         (Basic.Randomize):
725         (Basic.End):
726         (Basic.Program):
727         * Basic/basic.js: Added.
728         (prepare):
729         (simulate):
730         * Basic/benchmark.js: Added.
731         (Benchmark):
732         (Benchmark.prototype.runIteration.expect):
733         (Benchmark.prototype.runIteration):
734         (runBenchmark):
735         * Basic/caseless_map.js: Added.
736         (CaselessMap):
737         * Basic/lexer.js: Added.
738         (lex.consumeWhitespace):
739         (lex.consume):
740         (lex):
741         * Basic/lexer_test.js: Added.
742         * Basic/number.js: Added.
743         (NumberValue):
744         (NumberValue.prototype.get value):
745         (NumberValue.prototype.apply):
746         (NumberValue.prototype.leftApply):
747         (NumberValue.prototype.assign):
748         (NumberArray.):
749         (NumberArray):
750         (NumberArray.prototype.apply):
751         (NumberArray.prototype.leftApply):
752         (NumberFunction):
753         (NumberFunction.prototype.apply):
754         (NumberFunction.prototype.leftApply):
755         (NativeFunction):
756         (NativeFunction.prototype.apply):
757         (NativeFunction.prototype.leftApply):
758         * Basic/parser.js: Added.
759         (parse):
760         (parse.pushToken):
761         (parse.peekToken):
762         (parse.consumeKind):
763         (parse.consumeToken):
764         (parse.parseVariable):
765         (parse.parseNumericExpression.parsePrimary):
766         (parse.parseNumericExpression.parseFactor):
767         (parse.parseNumericExpression.parseTerm):
768         (parse.parseNumericExpression):
769         (parse.parseConstant):
770         (parse.parseStringExpression):
771         (parse.isStringExpression):
772         (parse.parseRelationalExpression):
773         (parse.parseNonNegativeInteger):
774         (parse.parseGoToStatement):
775         (parse.parseGoSubStatement):
776         (parse.parseStatement):
777         (parse.parseStatements):
778         * Basic/random.js: Added.
779         (createRNG):
780         (createRNGWithFixedSeed):
781         (createRNGWithRandomSeed):
782         * Basic/state.js: Added.
783         (State):
784         (State.prototype.getValue):
785         (State.prototype.getSideState):
786         (State.prototype.abort):
787         (State.prototype.validate):
788         * Basic/test.html: Added.
789         * Basic/test.js: Added.
790         * Basic/util.js: Added.
791         (this.performance.performance.now.currentTime):
792         (else.this.preciseTime.currentTime):
793         (else.currentTime):
794         * Skipped: Make sure that we don't run Basic yet.
795
796 2016-06-21  Jon Lee  <jonlee@apple.com>
797
798         Update canvas size when benchmark begins
799         https://bugs.webkit.org/show_bug.cgi?id=159010
800
801         Reviewed by Dean Jackson.
802
803         Previously we would evaluate the media queries and assign the canvas size to the test
804         on body.onload. Instead, do it once the user starts the benchmark.
805
806         * Animometer/resources/debug-runner/animometer.js: Move benchmarkController.determineCanvasSize()
807         to benchmarkController._startBenchmark.
808         * Animometer/resources/runner/animometer.js: Ditto.
809
810 2016-06-21  Jon Lee  <jonlee@apple.com>
811
812         Add new timestamp option
813         https://bugs.webkit.org/show_bug.cgi?id=159006
814
815         Reviewed by Dean Jackson.
816
817         Add a new option to take timestamps from the rAF callback.
818
819         * Animometer/developer.html: Added option.
820         * Animometer/tests/resources/main.js:
821         (Benchmark): Fall back to using the rAF timestamp if performance.now()
822         is not available.
823         (Benchmark._animateLoop): Update how timestamp is set. Prefer to use
824         local var instead of accessing private var.
825
826 2016-06-21  Jon Lee  <jonlee@apple.com>
827
828         Improvements to Animometer benchmark
829         https://bugs.webkit.org/show_bug.cgi?id=157738
830
831         Reviewed by Dean Jackson.
832         Provisionally reviewed by Said Abou-Hallawa.
833
834         Update tests.
835
836         * Animometer/tests/master/text.html: Ensure only three text sizes for
837         the three canvases.
838         * Animometer/tests/master/focus.html: Reduce the text size to fit with smaller
839         particle sizes.
840         * Animometer/tests/master/resources/focus.js: Remove the quadratic distribution for
841         particle sizes, and make it linear. Reduce the size variance. Shuffle the math to
842         reduce some calculations per frame. Fix the placement of the particles which might
843         otherwise be culled.
844         * Animometer/tests/master/resources/image-data.js: Reduce the particle size
845         to encourage larger scores.
846
847 2016-06-21  Jon Lee  <jonlee@apple.com>
848
849         Improvements to Animometer benchmark
850         https://bugs.webkit.org/show_bug.cgi?id=157738
851
852         Reviewed by Dean Jackson.
853         Provisionally reviewed by Said Abou-Hallawa.
854
855         Include confidence interval for the final score, and store the canvas
856         size in the serialization so that it is accurately shown in results.
857
858         * Animometer/developer.html: Add a "confidence" div.
859         * Animometer/index.html: Ditto. Convert "mean" to "confidence".
860         * Animometer/resources/debug-runner/animometer.js: Look at options, and
861         if the configuration is included, update the body class based on it
862         (similar to what we do when we first load the page). That way, if you
863         drag-and-drop previous results in, that configuration is reflected in
864         the dashboard. Show the full confidence interval.
865
866         * Animometer/resources/debug-runner/animometer.css:
867         * Animometer/resources/debug-runner/animometer.js: Style update.
868         * Animometer/resources/runner/animometer.css:
869
870         * Animometer/resources/runner/animometer.js:
871         (_processData): Propagate the confidence interval values out and calculate
872         the lower and upper bounds. For now, shortcut the aggregate calculation,
873         since we only go through one iteration.
874         (this._processData.calculateScore): Propagate the confidence interval out
875         to be next to the score. Depending on the controller these values are
876         calculated differently.
877         (this._processData._getResultsProperty): Convenience function.
878         (this._processData.get score): Refactor.
879         (this._processData.get scoreLowerBound): Get the aggregate lower bound.
880         (this._processData.get scoreUpperBound): Get the aggregate upper bound.
881         (window.sectionsManager.setSectionScore):
882         (window.benchmarkController._startBenchmark): When the benchmark starts, note
883         the canvas size and add it to options. That way it will be automatically be
884         serialized.
885         (window.benchmarkController.showResults): Include the maximum deviation
886         percentage.
887         * Animometer/resources/runner/lines.svg: Make the background line up with the
888         skew.
889         * Animometer/resources/runner/tests.js:
890         (Headers.details.text): Refactor.
891         * Animometer/resources/statistics.js:
892         (Statistics.largestDeviationPercentage): Convenience function to calculate
893         the deviation percentage on either end and return the largest deviation.
894         * Animometer/resources/strings.js:
895
896         Allow specifying a regression profile to use instead of taking the one
897         with the lowest error.
898
899         Address an issue in the focus test when the regression calculated ends
900         up overestimating the change point, causing a cascade of tougher
901         ramps. The reason behind this is that at max complexity of an initial
902         ramp, the frame length is very high, and it influences the second
903         segment of the piecewise regression strongly, causing it to be very
904         steep. As a result, the first segment, expected to be flat, ends up
905         covering a higher range of complexity. That makes the change point
906         much higher than it should be. To avoid this, we will add a sanity
907         check on the maximum value of the ramp. If the regression's projected
908         value at the maximum complexity of the current ramp is very slow (less
909         than 20 fps), 1) reduce the maximum complexity by 20%, and 2) do not
910         include the regression's change point in the change point estimator.
911         That estimator is used as the midpoint of the next ramp, and including
912         the change point from a poor regression can bake in the error. The
913         controller already knows how to adjust for ramps that are too easy for
914         the system.
915
916         * Animometer/resources/runner/animometer.js:
917         (this._processData.findRegression): Takes a preferred profile and gives that to
918         Regression.
919         (this._processData.calculateScore): With the ramp controller, take the profile
920         of the ramp that was used the most when calculating the ramp's regression. That
921         profile is what is used for the test's score.
922         * Animometer/resources/statistics.js:
923         (Regression.Utilities.createClass): Update to take an options object which can
924         specify a profile to calculate with. Otherwise it will continue to use both and
925         select the one with the lower error.
926         (_calculateRegression): Fix an issue where we claim 0 error if the regression
927         calculation fails due to divide-by-zero. Instead reject that regression calculation
928         by giving it Number.MAX_VALUE.
929         * Animometer/resources/strings.js: New strings for marking a regression as flat
930         or slope.
931         * Animometer/tests/resources/main.js:
932         (RampController): Rename the thresholds for clarity. Add a threshold that, if
933         exceeded, will lower the maximum complexity of the next ramp.
934         (tune): Relax the cdf check to consider whether the interval definitely falls in
935         the desired frame length threshold.
936         (processSamples): Include the profile in the ramp.
937
938         Update ramp controller test. Increase the length of the test to 30 seconds, and extend
939         the interval to 120 ms during sampling. Improve the estimation of the ramp parameters.
940
941         * Animometer/developer.html: Change default to 30 seconds, and don't show the progress bar
942         by default.
943         * Animometer/resources/runner/animometer.js: Change default to 30 seconds.
944         * Animometer/tests/resources/main.js: A number of improvements to the ramp controller, in
945         the order in which they appear in the patch:
946
947         - With a longer test length use longer ramps with longer intervals to get more data at each
948         complexity. Keep the 100 ms interval length during the ramp up phase since we don't need to
949         spend more time there to find the right order of magnitude, but increase it during the
950         ramps to 120 ms.
951         - The ramp linearly interpolates the complexity to render based on its timestamp, but it would
952         never sample the minimum complexity. Instead of lerping max to min complexity from time
953         0 to t where t is the ramp length, instead lerp from 0 to (t - intervalSampleLength) so that
954         we can have at least one interval sample at the min complexity for that ramp.
955         - Some regression calculations only come out with one line segment rather than the two
956         we expect. This could be due to a noisy ramp or the ramp's range is too narrow. If that's the
957         case, influence the minimum complexity of the next ramp towards the lowest bound of 1, so that
958         we ensure that at least part of the ramp is covering a complexity range that the system can
959         handle at full 60.
960         - Remove an assignment to interpolatedFrameLength since that is never subsequently used.
961
962         Update the format used to serialize the results for analysis.
963
964         Each data point used to have named properties for fields like complexity and frame rate.
965         In addition the serialized numbers had rounding errors that took up many characters.
966         Update the format by introducing a new data container called SampleData, which contains a
967         field map. The map maps a string to an array index. Each data point is an array, so, to
968         get a stat, use the field map to get the array index into the data point. This allows future
969         versions to track other data, and reduces the size of the output string by two-thirds.
970
971         * Animometer/resources/extensions.js:
972         (Utilities.toFixedNumber): Add convenience function that truncates the number to a fixed
973         precision, and converts it back to a number.
974         (SampleData): New class that contains sample data and a field map that maps properties to
975         an array index.
976         (get length): Number of data points.
977         (addField): Add a field to the field map.
978         (push): Add a data point.
979         (sort): Sort the data.
980         (slice): Return new SampleData object with sliced data.
981         (forEach): Iterate over the data with provided function.
982         (createDatum):
983         (getFieldInDatum): Returns the data point associated with the field name by looking it up
984         in the field map in the datum provided, which can be the datum object itself (an array) or
985         an index into the data member variable.
986         (setFieldInDatum): Sets the data point associated with the field name.
987         (at): Returns the data point at the provided index.
988         (toArray): Serializes the data where the field map serves as property names for each point.
989
990         * Animometer/resources/debug-runner/graph.js:
991         (updateGraphData): Remove unused _testData. Convert the data to the old array format for the
992         graph to use, since the old format was much easier to work with when displaying the graphs.
993         (onGraphTypeChanged): For some controllers, no alternative score or mean is provided.
994         * Animometer/resources/runner/animometer.css:
995         * Animometer/resources/runner/animometer.js: Refactor to use SampleData. Update JSON output
996         to only go to 3 digits of precision for purposes of reducing the data size.
997         * Animometer/resources/strings.js: Add new strings to put into the field maps.
998         * Animometer/tests/resources/main.js: Refactor to use SampleData.
999
1000         * Animometer/developer.html:
1001         * Animometer/index.html: Restructure results table for both pages. Add charset attribute to
1002         tests.js include.
1003         * Animometer/resources/debug-runner/animometer.css: Clear out styles from release runner.
1004         * Animometer/resources/debug-runner/graph.js:
1005         (onGraphTypeChanged): Update score and mean if bootstrap results are available from the
1006         controller, since not all controllers do bootstrapping.
1007         * Animometer/resources/debug-runner/tests.js: Update header text.
1008         * Animometer/resources/runner/animometer.css: Include confidence interval in results.
1009         * Animometer/resources/runner/animometer.js:
1010         (ResultsTable._addHeader): Header contents can be HTML, so use innerHTML instead.
1011         (ResultsTable._addBody): Add tbody element.
1012         (ResultsTable._addTest): Allow a data cell to invoke a JS function to get its contents.
1013         (window.benchmarkController.showResults): Add table that includes tests' confidence intervals.
1014         * Animometer/resources/runner/tests.js:
1015         (Headers.details.text): Add new details table that includes bootstrap confidence interval.
1016         The interval can be asymmetric, but for simplicity, report the maximum deviation percentage
1017         on either side of the bootstrap median.
1018         * Animometer/resources/statistics.js:
1019         (bootstrap): Include the confidence percentage in the return object.
1020
1021         Report canvas size in results.
1022
1023         * Animometer/developer.html: Add markup to indicate whether a small, medium, or large
1024         canvas was used.
1025         * Animometer/index.html: Ditto.
1026         * Animometer/resources/debug-runner/animometer.js: Call determineCanvasSize().
1027         * Animometer/resources/runner/animometer.css: Update styles to set the canvas based on the
1028         body class size.
1029         * Animometer/resources/runner/animometer.js:
1030         (window.benchmarkController.initialize): Update styles to set the canvas based on the
1031         body class size.
1032         (window.benchmarkController.determineCanvasSize): Run various media queries and set the body
1033         class based on the size of the device.
1034
1035         * Animometer/developer.html: Refactor to include the main CSS file, and redo
1036         the layout so that it doesn't rely on flexbox.
1037         * Animometer/resources/debug-runner/animometer.css:
1038         * Animometer/resources/debug-runner/animometer.js:
1039         (updateDisplay): Since various parts of the script alter the body class, we can't
1040         replace the className directly. Instead, remove all display-based values and then add
1041         the one that was selected.
1042         * Animometer/resources/debug-runner/graph.js:
1043         (updateGraphData): To set the size of the graph, use window.innerHeight.
1044         * Animometer/resources/runner/animometer.js:
1045         (window.sectionsManager.showSection): Since various parts of the script alter the body
1046         class, we can't replace the className directly. Remove all of the section classes
1047         individually and then add the one desired.
1048         * Animometer/tests/resources/stage.css: Remove -apple-system as a font to use in the
1049         stage.
1050
1051 2016-06-12  Filip Pizlo  <fpizlo@apple.com>
1052
1053         Fix round-down goof in Air.js's ShuffleCustom.forEachArg
1054         https://bugs.webkit.org/show_bug.cgi?id=158674
1055
1056         Reviewed by Michael Saboff.
1057         
1058         x / 3 * 3 is not how you round down to multiples of 3 in JavaScript. You need to do
1059         Math.floor(x / 3) * 3 instead.
1060         
1061         This is a benign change, because having extra arguments to a Shuffle is not something we
1062         actually take advantage of yet. But I think it's best to match the original C++ code's
1063         intent.
1064
1065         * Air.js/custom.js:
1066         (const.ShuffleCustom.forEachArg):
1067
1068 2016-06-11  Filip Pizlo  <fpizlo@apple.com>
1069
1070         Unreviewed, skip Air.js. It got unskipped when I renamed it from JSAir.
1071
1072         * Skipped:
1073
1074 2016-06-10  Filip Pizlo  <fpizlo@apple.com>
1075
1076         Make it easy to package Air.js
1077         https://bugs.webkit.org/show_bug.cgi?id=158652
1078
1079         Reviewed by Benjamin Poulain.
1080         
1081         If you want to give Air.js to someone, you can now do:
1082         
1083         cd PerformanceTests/Air.js
1084         ./make_dist.sh
1085         
1086         This will create Air.js.tar.gz and a directory called Air.js. If you want to send someone
1087         the benchmark, just send either of those things and tell them to look at index.html.
1088         
1089         You may have to edit make_dist.sh for wherever you have Markdown.pl. You can get that from
1090         https://daringfireball.net/projects/markdown/
1091
1092         * Air.js/make_dist.sh: Added.
1093         * Air.js/test.html:
1094
1095 2016-06-10  Filip Pizlo  <fpizlo@apple.com>
1096
1097         Unreviewed, make the link to test.html into an actual link.
1098
1099         * Air.js/README.md:
1100
1101 2016-06-10  Filip Pizlo  <fpizlo@apple.com>
1102
1103         Fix some text in Air.js/README.md
1104         https://bugs.webkit.org/show_bug.cgi?id=158650
1105
1106         Reviewed by Benjamin Poulain.
1107         
1108         I read the text again and found bugs:
1109         
1110         - We never actually say how to run the benchmark. This change adds a blurb about how to run
1111           it.
1112
1113         - We both say that allocateStack is responsible for the bulk of the running time and that
1114           we haven't measured where the bulk of the time is spent. This changes the text to say that
1115           it was a goal to make allocateStack be the hottest part of the benchmark, but that we did
1116           not measure this.
1117
1118         * Air.js/README.md:
1119
1120 2016-06-10  Filip Pizlo  <fpizlo@apple.com>
1121
1122         Air.js should have some documentation
1123         https://bugs.webkit.org/show_bug.cgi?id=158648
1124
1125         Reviewed by Keith Miller.
1126         
1127         I want to be able to point people at a document if they want to know more about this
1128         benchmark.
1129
1130         * Air.js/README.md: Added.
1131
1132 2016-06-10  Filip Pizlo  <fpizlo@apple.com>
1133
1134         Rename JSAir to Air.js.
1135
1136         Rubber stamped by Mark Lam.
1137
1138         * Air.js: Copied from PerformanceTests/JSAir.
1139         * Air.js/airjs-tests.yaml: Copied from PerformanceTests/JSAir/jsair-tests.yaml.
1140         * Air.js/benchmark.js: Copied from PerformanceTests/JSAir/benchmark.js.
1141         (Benchmark):
1142         * Air.js/jsair-tests.yaml: Removed.
1143         * Air.js/payload-airjs-ACLj8C.js: Copied from PerformanceTests/JSAir/payload-jsair-ACLj8C.js.
1144         * Air.js/payload-jsair-ACLj8C.js: Removed.
1145         * Air.js/stress-test.js: Copied from PerformanceTests/JSAir/stress-test.js.
1146         * Air.js/test.html: Copied from PerformanceTests/JSAir/test.html.
1147         * Air.js/test.js: Copied from PerformanceTests/JSAir/test.js.
1148         * JSAir: Removed.
1149         * JSAir/all.js: Removed.
1150         * JSAir/allocate_stack.js: Removed.
1151         * JSAir/arg.js: Removed.
1152         * JSAir/basic_block.js: Removed.
1153         * JSAir/benchmark.js: Removed.
1154         * JSAir/code.js: Removed.
1155         * JSAir/custom.js: Removed.
1156         * JSAir/frequented_block.js: Removed.
1157         * JSAir/insertion_set.js: Removed.
1158         * JSAir/inst.js: Removed.
1159         * JSAir/jsair-tests.yaml: Removed.
1160         * JSAir/liveness.js: Removed.
1161         * JSAir/opcode.js: Removed.
1162         * JSAir/payload-gbemu-executeIteration.js: Removed.
1163         * JSAir/payload-imaging-gaussian-blur-gaussianBlur.js: Removed.
1164         * JSAir/payload-jsair-ACLj8C.js: Removed.
1165         * JSAir/payload-typescript-scanIdentifier.js: Removed.
1166         * JSAir/reg.js: Removed.
1167         * JSAir/stack_slot.js: Removed.
1168         * JSAir/stress-test.js: Removed.
1169         * JSAir/strip-hash.rb: Removed.
1170         * JSAir/symbols.js: Removed.
1171         * JSAir/test.html: Removed.
1172         * JSAir/test.js: Removed.
1173         * JSAir/tmp.js: Removed.
1174         * JSAir/tmp_base.js: Removed.
1175         * JSAir/util.js: Removed.
1176
1177 2016-06-10  Filip Pizlo  <fpizlo@apple.com>
1178
1179         JSC Stress Test failing: jsair-tests.yaml/test.js.ftl-eager-no-cjit
1180         https://bugs.webkit.org/show_bug.cgi?id=158571
1181
1182         Reviewed by Keith Miller.
1183         
1184         Introduce a different harness for run-jsc-stress-tests, which runs for a minimum of 10
1185         iterations and then tries to do any number of "bonus" iterations until it's been running for
1186         two seconds. Since this is the sort of test that isn't really meant to stress anything in
1187         particular, I think it's OK if it is time-limited in this way. The worst case is that some
1188         of its failures will be flaky, but I think that they would have been flaky anyway given the
1189         complexity of the test.
1190
1191         * JSAir/benchmark.js:
1192         (Benchmark):
1193         (Benchmark.prototype.runIteration):
1194         (benchmark): Deleted.
1195         * JSAir/jsair-tests.yaml:
1196         * JSAir/stress-test.js: Added.
1197         (preciseTime):
1198         * JSAir/test.html:
1199         * JSAir/test.js:
1200
1201 2016-06-09  Filip Pizlo  <fpizlo@apple.com>
1202
1203         Unreviewed, teach the perf bots not to run JSAir.
1204
1205         * Skipped:
1206
1207 2016-06-08  Filip Pizlo  <fpizlo@apple.com>
1208
1209         Use more ES6 features in JSAir
1210         https://bugs.webkit.org/show_bug.cgi?id=158497
1211
1212         Reviewed by Keith Miller.
1213         
1214         This improves JSAir with the following ES6 features suggested by JoePeck:
1215         
1216         - String interpolation.
1217         - Destructuring inside PatchCustom.
1218         - Default arguments.
1219         
1220         All of these things are on hot paths.
1221         
1222         Note that I didn't use string interpolation everywhere that I could, only in those places
1223         where it made the code more readable. In Ruby, I used the style that if the interpolation
1224         expression has any non-trivial stuff (like a ternary operator, a chain of calls, or embedded
1225         strings) then it's better to use regular strcat. I think that's what I carried over to here.
1226         
1227         Note that the previous change (Add result validation to JSAir) also made the Proxy code not
1228         dead, though it's not necessarily on the hot path. The Proxy isn't called into frequently
1229         but it's used from a function that is otherwise hot, so if calling into the Proxy prevents
1230         that function from being optimized then it will hurt so good.
1231         
1232         I also reenabled tail calls in a few places.
1233         
1234         This change doesn't seem to change the performance of the benchmark for us. That's expected
1235         since these ES6 features are cheap. Note that this claim doesn't include Proxy, which was
1236         added in a separate change and that change did make the benchmark overall more expensive.
1237
1238         * JSAir/allocate_stack.js:
1239         (allocateStack):
1240         * JSAir/arg.js:
1241         (Arg.createBitImm64):
1242         (Arg.createAddr):
1243         (Arg.createStack):
1244         (Arg.logScale):
1245         (Arg.createIndex):
1246         * JSAir/basic_block.js:
1247         (BasicBlock.get headerString):
1248         (BasicBlock.prototype.get if):
1249         (BasicBlock):
1250         * JSAir/benchmark.js:
1251         (benchmark):
1252         * JSAir/code.js:
1253         (Code):
1254         (Code.prototype.addBlock):
1255         (Code.prototype.addStackSlot):
1256         (Code.prototype.newTmp):
1257         (Code.prototype.get size):
1258         (Code.prototype.get blocks):
1259         (Code.prototype.get stackSlots):
1260         (Code.prototype.tmps):
1261         (Code.prototype.get callArgAreaSize):
1262         (Code.prototype.toString):
1263         * JSAir/custom.js:
1264         (const.PatchCustom.forEachArg):
1265         * JSAir/inst.js:
1266         (Inst):
1267         * JSAir/reg.js:
1268         (Reg.prototype.toString):
1269         * JSAir/util.js:
1270         (symbolName):
1271         (lowerSymbolName):
1272
1273 2016-06-07  Filip Pizlo  <fpizlo@apple.com>
1274
1275         Add result validation to JSAir
1276         https://bugs.webkit.org/show_bug.cgi?id=158493
1277
1278         Reviewed by Saam Barati.
1279         
1280         This adds the ability to hash a Code in a way that matches the C++ code's hashing of Code.
1281         This allows us to check if the Code that JSAir sees is the code that C++ saw. We use this to
1282         check the Code before and after allocateStack, and compare against hashes we got from C++.
1283         
1284         Doing this uncovered bugs. roundUpToMultipleOf wasn't doing anything. allocateStack was not
1285         allocating things correctly because I was concatting a Set to an Array, which doesn't really
1286         work. Now these bugs are fixed.
1287         
1288         The checking step adds to the running time so I reduced the number of iterations. The
1289         benchmark spends a decent amount of its time computing Code hashes; I think it's around 1/3
1290         total. This is probably OK. It's better to verify the results even if the running time is
1291         not all in the "core" of the algorithm.
1292         
1293         Also add a run-jsc-stress-tests yaml file to allow this to run as a test.
1294
1295         * JSAir/allocate_stack.js:
1296         * JSAir/arg.js:
1297         (Arg.createImm):
1298         (Arg.createBigImm):
1299         (Arg.createBitImm):
1300         (Arg.createBitImm64):
1301         (Arg.createWidth):
1302         (Arg.createSpecial):
1303         (Arg.prototype.get kind):
1304         (Arg.prototype.get isTmp):
1305         (Arg.prototype.get isImm):
1306         (Arg.prototype.get isSomeImm):
1307         (Arg.prototype.get isSomeBigImm):
1308         (Arg.prototype.get isCondition):
1309         (Arg.prototype.get isWidth):
1310         (Arg.prototype.get isSpecial):
1311         (Arg.prototype.get isAlive):
1312         (Arg.prototype.get tmp):
1313         (Arg.prototype.get value):
1314         (Arg.prototype.get lowValue):
1315         (Arg.prototype.get highValue):
1316         (Arg.prototype.get base):
1317         (Arg.prototype.get isGP):
1318         (Arg.prototype.get isFP):
1319         (Arg.prototype.isValidForm):
1320         (Arg.prototype.get isInvertible):
1321         (Arg.kindCode):
1322         (Arg.prototype.hash):
1323         (Arg.prototype.toString):
1324         (Arg):
1325         * JSAir/basic_block.js:
1326         (BasicBlock.get successorBlocks):
1327         * JSAir/benchmark.js:
1328         (benchmark):
1329         * JSAir/code.js:
1330         (Code.prototype.setFrameSize):
1331         (Code.prototype.hash):
1332         (Code.prototype.toString):
1333         (Code):
1334         * JSAir/inst.js:
1335         (Inst.prototype.get hasNonArgEffects):
1336         (Inst.prototype.hash):
1337         (Inst.prototype.toString):
1338         (Inst):
1339         * JSAir/jsair-tests.yaml: Added.
1340         * JSAir/opcode.js:
1341         (Inst_forEachArg):
1342         (Inst_hasNonArgEffects):
1343         (opcodeCode):
1344         * JSAir/payload-gbemu-executeIteration.js:
1345         (createPayloadGbemuExecuteIteration):
1346         * JSAir/payload-imaging-gaussian-blur-gaussianBlur.js:
1347         (createPayloadImagingGaussianBlurGaussianBlur):
1348         * JSAir/payload-jsair-ACLj8C.js:
1349         (createPayloadJSAirACLj8C):
1350         * JSAir/payload-typescript-scanIdentifier.js:
1351         (createPayloadTypescriptScanIdentifier):
1352         * JSAir/reg.js:
1353         (Reg.prototype.get isReg):
1354         (Reg.prototype.hash):
1355         (Reg.prototype.toString):
1356         * JSAir/stack_slot.js:
1357         (StackSlot.prototype.setOffsetFromFP):
1358         (StackSlot.prototype.hash):
1359         (StackSlot.prototype.toString):
1360         * JSAir/symbols.js:
1361         (relCondCode):
1362         (resCondCode):
1363         (doubleCondCode):
1364         * JSAir/test.html:
1365         * JSAir/tmp.js:
1366         (Tmp.prototype.get isReg):
1367         (Tmp.prototype.hash):
1368         (Tmp.prototype.toString):
1369         * JSAir/util.js:
1370         (roundUpToMultipleOf):
1371         (symbolName):
1372
1373 2016-06-07  Filip Pizlo  <fpizlo@apple.com>
1374
1375         Implement Air::allocateStack() in ES6 to see how much of a bad idea that is
1376         https://bugs.webkit.org/show_bug.cgi?id=158318
1377
1378         Reviewed by Saam Barati.
1379         
1380         This adds a new benchmark for us to play with called JSAir. It's a complete ES6
1381         implementation of Air's allocateStack() phase along with all of Air needed to run that
1382         phase. This includes things like stack slots, registers, temporaries, basic blocks,
1383         instructions, and all of the code for iterating over, inspecting, and modifying those
1384         things.
1385         
1386         To make this work, JSC can now dump Air just before allocateStack() in the form of JS code
1387         that creates a Code object that matches exactly what C++ Air saw. This benchmark comprises
1388         four Air IRs:
1389         
1390         - Octane/gbemu's largest function, executeIteration.
1391         - Kraken/imaging-gaussian-blur's largest function in OSR entry mode, gaussuanBlur.
1392         - Octane/typescript's largest function that is actually hot, scanIdentifier.
1393         - JSAir's largest hot function, which is anonymous, so we call it by its hash (ACLj8C).
1394         
1395         This runs in about 2 seconds on my machine in JSC trunk. It includes both a commandline
1396         harness and a web harness. JSAir is almost exactly 100x slower in ES6 in WebKit than the
1397         C++ Air::allocateStack() phase on which it is based.
1398         
1399         JSAir uses the following ES6 features:
1400         
1401         - Symbol.
1402         - for-of.
1403         - arrow functions.
1404         - Map/Set.
1405         - let/const.
1406         - classes.
1407         
1408         All of these things are used in anger and should end up on the hot path. There is also code
1409         that uses Proxies, but it ends up being dead.
1410         
1411         We can improve this even more:
1412         I still need to add result validation: https://bugs.webkit.org/show_bug.cgi?id=158493
1413         I want to make it use more ES6 features: https://bugs.webkit.org/show_bug.cgi?id=158497
1414         
1415         * JSAir: Added.
1416         * JSAir/all.js: Added.
1417         * JSAir/allocate_stack.js: Added.
1418         (allocateStack.attemptAssignment):
1419         (allocateStack.assign):
1420         (allocateStack.interfere):
1421         (allocateStack.):
1422         (allocateStack):
1423         * JSAir/arg.js: Added.
1424         (Arg):
1425         (Arg.isAnyUse):
1426         (Arg.isColdUse):
1427         (Arg.isWarmUse):
1428         (Arg.cooled):
1429         (Arg.isEarlyUse):
1430         (Arg.isLateUse):
1431         (Arg.isAnyDef):
1432         (Arg.isEarlyDef):
1433         (Arg.isLateDef):
1434         (Arg.isZDef):
1435         (Arg.typeForB3Type):
1436         (Arg.widthForB3Type):
1437         (Arg.conservativeWidth):
1438         (Arg.minimumWidth):
1439         (Arg.bytes):
1440         (Arg.widthForBytes):
1441         (Arg.createTmp):
1442         (Arg.fromReg):
1443         (Arg.createImm):
1444         (Arg.createBigImm):
1445         (Arg.createBitImm):
1446         (Arg.createBitImm64):
1447         (Arg.createAddr):
1448         (Arg.createStack):
1449         (Arg.createCallArg):
1450         (Arg.createStackAddr):
1451         (Arg.isValidScale):
1452         (Arg.logScale):
1453         (Arg.createIndex):
1454         (Arg.createRelCond):
1455         (Arg.createResCond):
1456         (Arg.createDoubleCond):
1457         (Arg.createWidth):
1458         (Arg.prototype.get kind):
1459         (Arg.prototype.get isTmp):
1460         (Arg.prototype.get isImm):
1461         (Arg.prototype.get isBigImm):
1462         (Arg.prototype.get isBitImm):
1463         (Arg.prototype.get isBitImm64):
1464         (Arg.prototype.get isSomeImm):
1465         (Arg.prototype.get isAddr):
1466         (Arg.prototype.get isStack):
1467         (Arg.prototype.get isCallArg):
1468         (Arg.prototype.get isIndex):
1469         (Arg.prototype.get isMemory):
1470         (Arg.prototype.get isStackMemory):
1471         (Arg.prototype.get isRelCond):
1472         (Arg.prototype.get isResCond):
1473         (Arg.prototype.get isDoubleCond):
1474         (Arg.prototype.get isCondition):
1475         (Arg.prototype.get isWidth):
1476         (Arg.prototype.get isAlive):
1477         (Arg.prototype.get tmp):
1478         (Arg.prototype.get value):
1479         (Arg.prototype.get base):
1480         (Arg.prototype.get hasOffset):
1481         (Arg.prototype.get offset):
1482         (Arg.prototype.get stackSlot):
1483         (Arg.prototype.get index):
1484         (Arg.prototype.get scale):
1485         (Arg.prototype.get logScale):
1486         (Arg.prototype.get width):
1487         (Arg.prototype.get isGPTmp):
1488         (Arg.prototype.get isFPTmp):
1489         (Arg.prototype.get isGP):
1490         (Arg.prototype.get isFP):
1491         (Arg.prototype.get hasType):
1492         (Arg.prototype.get type):
1493         (Arg.prototype.isType):
1494         (Arg.prototype.isCompatibleType):
1495         (Arg.prototype.get isGPR):
1496         (Arg.prototype.get gpr):
1497         (Arg.prototype.get isFPR):
1498         (Arg.prototype.get fpr):
1499         (Arg.prototype.get isReg):
1500         (Arg.prototype.get reg):
1501         (Arg.isValidImmForm):
1502         (Arg.isValidBitImmForm):
1503         (Arg.isValidBitImm64Form):
1504         (Arg.isValidAddrForm):
1505         (Arg.isValidIndexForm):
1506         (Arg.prototype.isValidForm):
1507         (Arg.prototype.forEachTmpFast):
1508         (Arg.prototype.usesTmp):
1509         (Arg.prototype.forEachTmp):
1510         (Arg.prototype.is):
1511         (Arg.prototype.as):
1512         (Arg.prototype.forEachFast):
1513         (Arg.prototype.forEach):
1514         (Arg.extract):
1515         (Arg.forEachFast):
1516         (Arg.forEach):
1517         (Arg.prototype.get condition):
1518         (Arg.prototype.get isInvertible):
1519         (Arg.prototype.toString):
1520         * JSAir/basic_block.js: Added.
1521         (BasicBlock):
1522         (BasicBlock.prototype.get index):
1523         (BasicBlock.prototype.get size):
1524         (BasicBlock.prototype.Symbol.iterator):
1525         (BasicBlock.prototype.at):
1526         (BasicBlock.get last):
1527         (BasicBlock.get insts):
1528         (BasicBlock.get numSuccessors):
1529         (BasicBlock.get successors):
1530         (BasicBlock.get successorBlocks.):
1531         (BasicBlock.get successorBlocks):
1532         (BasicBlock.set get numPredecessors):
1533         (BasicBlock.get predecessors):
1534         (BasicBlock.get frequency):
1535         (BasicBlock.get headerString):
1536         (BasicBlock.get footerString):
1537         * JSAir/benchmark.js: Added.
1538         (benchmark):
1539         * JSAir/code.js: Added.
1540         (Code):
1541         (Code.prototype.addBlock):
1542         (Code.prototype.addStackSlot):
1543         (Code.prototype.newTmp):
1544         (Code.prototype.get size):
1545         (Code.prototype.at):
1546         (Code.prototype.Symbol.iterator):
1547         (Code.prototype.get blocks):
1548         (Code.prototype.get stackSlots):
1549         (Code.prototype.tmps):
1550         (Code.prototype.get callArgAreaSize):
1551         (Code.prototype.requestCallArgAreaSize):
1552         (Code.prototype.get frameSize):
1553         (Code.prototype.setFrameSize):
1554         (Code.prototype.toString):
1555         * JSAir/custom.js: Added.
1556         (const.ShuffleCustom.forEachArg):
1557         (const.ShuffleCustom.hasNonArgNonControlEffects):
1558         (const.PatchCustom.forEachArg):
1559         (const.PatchCustom.hasNonArgNonControlEffects):
1560         (const.CCallCustom.forEachArg):
1561         (const.CCallCustom.hasNonArgNonControlEffects):
1562         (const.ColdCCallCustom.forEachArg):
1563         (const.ColdCCallCustom.hasNonArgNonControlEffects):
1564         * JSAir/frequented_block.js: Added.
1565         (FrequentedBlock):
1566         (FrequentedBlock.prototype.toString):
1567         * JSAir/insertion_set.js: Added.
1568         (Insertion):
1569         (Insertion.prototype.get index):
1570         (Insertion.prototype.get element):
1571         (Insertion.prototype.lessThan):
1572         (InsertionSet):
1573         (InsertionSet.prototype.appendInsertion):
1574         (InsertionSet.prototype.append):
1575         (InsertionSet.prototype.execute):
1576         * JSAir/inst.js: Added.
1577         (Inst):
1578         (Inst.prototype.append):
1579         (Inst.prototype.clear):
1580         (Inst.prototype.get opcode):
1581         (Inst.prototype.get args):
1582         (Inst.prototype.visitArg):
1583         (Inst.prototype.forEachTmpFast):
1584         (Inst.prototype.forEachArg):
1585         (Inst.prototype.forEachTmp):
1586         (Inst.prototype.forEach):
1587         (Inst.forEachDef):
1588         (Inst.forEachDefWithExtraClobberedRegs):
1589         (Inst.prototype.get hasNonArgEffects):
1590         (Inst.prototype.toString):
1591         * JSAir/liveness.js: Added.
1592         (Liveness):
1593         (Liveness.prototype.get thing):
1594         (Liveness.prototype.get code):
1595         (Liveness.prototype.get liveAtHead):
1596         (Liveness.prototype.get liveAtTail):
1597         (Liveness.prototype.localCalc.LocalCalc):
1598         (Liveness.prototype.localCalc.LocalCalc.prototype.get liveSet):
1599         (Liveness.prototype.localCalc.LocalCalc.prototype.execute):
1600         (Liveness.prototype.localCalc):
1601         * JSAir/opcode.js: Added.
1602         (Inst_forEachArg):
1603         (Inst_hasNonArgEffects):
1604         * JSAir/payload-gbemu-executeIteration.js: Added.
1605         (createPayloadGbemuExecuteIteration):
1606         * JSAir/payload-imaging-gaussian-blur-gaussianBlur.js: Added.
1607         (createPayloadImagingGaussianBlurGaussianBlur):
1608         * JSAir/payload-jsair-ACLj8C.js: Added.
1609         (createPayloadJSAirACLj8C):
1610         * JSAir/payload-typescript-scanIdentifier.js: Added.
1611         (createPayloadTypescriptScanIdentifier):
1612         * JSAir/reg.js: Added.
1613         (Reg):
1614         (Reg.fromReg):
1615         (Reg.prototype.get index):
1616         (Reg.prototype.get type):
1617         (Reg.prototype.get name):
1618         (Reg.prototype.get isCalleeSave):
1619         (Reg.prototype.get isReg):
1620         (Reg.prototype.toString):
1621         (Reg.extract):
1622         (Reg.forEachFast):
1623         (Reg.forEach):
1624         (newGPR):
1625         (Reg.gprs.Reg.fprs.Reg.calleeSaveGPRs.Reg.calleeSaveFPRs.Reg.calleeSaves):
1626         * JSAir/stack_slot.js: Added.
1627         (StackSlot):
1628         (StackSlot.prototype.get byteSize):
1629         (StackSlot.prototype.get kind):
1630         (StackSlot.prototype.get isLocked):
1631         (StackSlot.prototype.get isSpill):
1632         (StackSlot.prototype.get index):
1633         (StackSlot.prototype.ensureSize):
1634         (StackSlot.prototype.get alignment):
1635         (StackSlot.prototype.get offsetFromFP):
1636         (StackSlot.prototype.setOffsetFromFP):
1637         (StackSlot.prototype.toString):
1638         (StackSlot.extract):
1639         (StackSlot.forEachFast):
1640         (StackSlot.forEach):
1641         * JSAir/symbols.js: Added.
1642         * JSAir/test.html: Added.
1643         * JSAir/test.js: Added.
1644         * JSAir/tmp.js: Added.
1645         (Tmp):
1646         (Tmp.fromReg):
1647         (Tmp.prototype.get index):
1648         (Tmp.prototype.get type):
1649         (Tmp.prototype.get isReg):
1650         (Tmp.prototype.toString):
1651         (Tmp.extract):
1652         (Tmp.forEachFast):
1653         (Tmp.forEach):
1654         * JSAir/tmp_base.js: Added.
1655         (TmpBase.prototype.get isGP):
1656         (TmpBase.prototype.get isFP):
1657         (TmpBase.prototype.get isGPR):
1658         (TmpBase.prototype.get isFPR):
1659         (TmpBase.prototype.get reg):
1660         (TmpBase.prototype.get gpr):
1661         (TmpBase.prototype.get fpr):
1662         (TmpBase):
1663         * JSAir/util.js: Added.
1664         (isRepresentableAsInt32):
1665         (addIndexed):
1666         (roundUpToMultipleOf):
1667         (symbolName):
1668         (mergeIntoSet):
1669         (nonEmptyRangesOverlap):
1670         (rangesOverlap):
1671         (removeAllMatching):
1672         (swap):
1673         (bubble):
1674         (bubbleSort):
1675         (currentTime):
1676         (else.currentTime):
1677
1678 2016-05-26  Yusuke Suzuki  <utatane.tea@gmail.com>
1679
1680         [JSC] Allow JSBench to use precise time
1681         https://bugs.webkit.org/show_bug.cgi?id=158050
1682
1683         Reviewed by Geoffrey Garen.
1684
1685         * JSBench/amazon-chrome-win/urem.js:
1686         (else.window.performance.window.performance.now.currentTimeInMS):
1687         (else.else.typeof.preciseTime.string_appeared_here.currentTimeInMS):
1688         (else.else.currentTimeInMS):
1689         (onload.cb):
1690         (onload):
1691         * JSBench/amazon-chrome/urem.js:
1692         (else.window.performance.window.performance.now.currentTimeInMS):
1693         (else.else.typeof.preciseTime.string_appeared_here.currentTimeInMS):
1694         (else.else.currentTimeInMS):
1695         (onload.cb):
1696         (onload):
1697         * JSBench/amazon-firefox-win/urm.js:
1698         (else.window.performance.window.performance.now.currentTimeInMS):
1699         (else.else.typeof.preciseTime.string_appeared_here.currentTimeInMS):
1700         (else.else.currentTimeInMS):
1701         (onload.cb):
1702         (onload):
1703         * JSBench/amazon-firefox/urm.js:
1704         (else.window.performance.window.performance.now.currentTimeInMS):
1705         (else.else.typeof.preciseTime.string_appeared_here.currentTimeInMS):
1706         (else.else.currentTimeInMS):
1707         (onload.cb):
1708         (onload):
1709         * JSBench/amazon-safari/urem.js:
1710         (else.window.performance.window.performance.now.currentTimeInMS):
1711         (else.else.typeof.preciseTime.string_appeared_here.currentTimeInMS):
1712         (else.else.currentTimeInMS):
1713         (onload.cb):
1714         (onload):
1715         * JSBench/facebook-chrome-win/urem.js:
1716         (else.window.performance.window.performance.now.currentTimeInMS):
1717         (else.else.typeof.preciseTime.string_appeared_here.currentTimeInMS):
1718         (else.else.currentTimeInMS):
1719         (onload.cb):
1720         (onload):
1721         * JSBench/facebook-chrome/urem.js:
1722         (else.window.performance.window.performance.now.currentTimeInMS):
1723         (else.else.typeof.preciseTime.string_appeared_here.currentTimeInMS):
1724         (else.else.currentTimeInMS):
1725         (onload.cb):
1726         (onload):
1727         * JSBench/facebook-firefox-win/urem.js:
1728         (else.window.performance.window.performance.now.currentTimeInMS):
1729         (else.else.typeof.preciseTime.string_appeared_here.currentTimeInMS):
1730         (else.else.currentTimeInMS):
1731         (onload.cb):
1732         (onload):
1733         * JSBench/facebook-firefox/urem.js:
1734         (else.window.performance.window.performance.now.currentTimeInMS):
1735         (else.else.typeof.preciseTime.string_appeared_here.currentTimeInMS):
1736         (else.else.currentTimeInMS):
1737         (onload.cb):
1738         (onload):
1739         * JSBench/facebook-safari/urem.js:
1740         (else.window.performance.window.performance.now.currentTimeInMS):
1741         (else.else.typeof.preciseTime.string_appeared_here.currentTimeInMS):
1742         (else.else.currentTimeInMS):
1743         (onload.cb):
1744         (onload):
1745         * JSBench/google-chrome-win/urem.js:
1746         (else.window.performance.window.performance.now.currentTimeInMS):
1747         (else.else.typeof.preciseTime.string_appeared_here.currentTimeInMS):
1748         (else.else.currentTimeInMS):
1749         (onload.cb):
1750         (onload):
1751         * JSBench/google-chrome/urem.js:
1752         (else.window.performance.window.performance.now.currentTimeInMS):
1753         (else.else.typeof.preciseTime.string_appeared_here.currentTimeInMS):
1754         (else.else.currentTimeInMS):
1755         (onload.cb):
1756         (onload):
1757         * JSBench/google-firefox-win/urem.js:
1758         (else.window.performance.window.performance.now.currentTimeInMS):
1759         (else.else.typeof.preciseTime.string_appeared_here.currentTimeInMS):
1760         (else.else.currentTimeInMS):
1761         (onload.cb):
1762         (onload):
1763         * JSBench/google-firefox/uem.js:
1764         (else.window.performance.window.performance.now.currentTimeInMS):
1765         (else.else.typeof.preciseTime.string_appeared_here.currentTimeInMS):
1766         (else.else.currentTimeInMS):
1767         (onload.cb):
1768         (onload):
1769         * JSBench/google-safari/urem.js:
1770         (else.window.performance.window.performance.now.currentTimeInMS):
1771         (else.else.typeof.preciseTime.string_appeared_here.currentTimeInMS):
1772         (else.else.currentTimeInMS):
1773         (onload.cb):
1774         (onload):
1775         * JSBench/harness.js:
1776         (runBenchmark.window.currentTimeInMS):
1777         (runBenchmark.else.window.currentTimeInMS):
1778         * JSBench/twitter-chrome-win/rem.js:
1779         (else.window.performance.window.performance.now.currentTimeInMS):
1780         (else.else.typeof.preciseTime.string_appeared_here.currentTimeInMS):
1781         (else.else.currentTimeInMS):
1782         (onload.cb):
1783         (onload):
1784         * JSBench/twitter-chrome/urem.js:
1785         (else.window.performance.window.performance.now.currentTimeInMS):
1786         (else.else.typeof.preciseTime.string_appeared_here.currentTimeInMS):
1787         (else.else.currentTimeInMS):
1788         (onload.cb):
1789         (onload):
1790         * JSBench/twitter-firefox-win/urem.js:
1791         (else.window.performance.window.performance.now.currentTimeInMS):
1792         (else.else.typeof.preciseTime.string_appeared_here.currentTimeInMS):
1793         (else.else.currentTimeInMS):
1794         (onload.cb):
1795         (onload):
1796         * JSBench/twitter-firefox/urem.js:
1797         (else.window.performance.window.performance.now.currentTimeInMS):
1798         (else.else.typeof.preciseTime.string_appeared_here.currentTimeInMS):
1799         (else.else.currentTimeInMS):
1800         (onload.cb):
1801         (onload):
1802         * JSBench/twitter-safari/urem.js:
1803         (else.window.performance.window.performance.now.currentTimeInMS):
1804         (else.else.typeof.preciseTime.string_appeared_here.currentTimeInMS):
1805         (else.else.currentTimeInMS):
1806         (onload.cb):
1807         (onload):
1808         * JSBench/yahoo-chrome-win/urem.js:
1809         (else.window.performance.window.performance.now.currentTimeInMS):
1810         (else.else.typeof.preciseTime.string_appeared_here.currentTimeInMS):
1811         (else.else.currentTimeInMS):
1812         (onload.cb):
1813         (onload):
1814         * JSBench/yahoo-chrome/urem.js:
1815         (else.window.performance.window.performance.now.currentTimeInMS):
1816         (else.else.typeof.preciseTime.string_appeared_here.currentTimeInMS):
1817         (else.else.currentTimeInMS):
1818         (onload.cb):
1819         (onload):
1820         * JSBench/yahoo-firefox-win/urem.js:
1821         (else.window.performance.window.performance.now.currentTimeInMS):
1822         (else.else.typeof.preciseTime.string_appeared_here.currentTimeInMS):
1823         (else.else.currentTimeInMS):
1824         (onload.cb):
1825         (onload):
1826         * JSBench/yahoo-firefox/urem.js:
1827         (else.window.performance.window.performance.now.currentTimeInMS):
1828         (else.else.typeof.preciseTime.string_appeared_here.currentTimeInMS):
1829         (else.else.currentTimeInMS):
1830         (onload.cb):
1831         (onload):
1832         * JSBench/yahoo-safari/urem.js:
1833         (else.window.performance.window.performance.now.currentTimeInMS):
1834         (else.else.typeof.preciseTime.string_appeared_here.currentTimeInMS):
1835         (else.else.currentTimeInMS):
1836         (onload.cb):
1837         (onload):
1838
1839 2016-05-25  Geoffrey Garen  <ggaren@apple.com>
1840
1841         replaceable own properties seem to ignore replacement after property caching
1842         https://bugs.webkit.org/show_bug.cgi?id=158091
1843
1844         Reviewed by Darin Adler.
1845
1846         * MallocBench/MallocBench.xcodeproj/project.pbxproj:
1847         * MallocBench/MallocBench/Benchmark.cpp:
1848         * MallocBench/MallocBench/Interpreter.cpp:
1849         (Interpreter::doMallocOp):
1850         * MallocBench/MallocBench/Interpreter.h:
1851         * MallocBench/MallocBench/fastMallocLog.63316.ops: Added.
1852         * MallocBench/MallocBench/jetstream.cpp: Added.
1853         (benchmark_jetstream):
1854         * MallocBench/MallocBench/jetstream.h: Added.
1855
1856 2016-05-25  Keith Miller  <keith_miller@apple.com>
1857
1858         Unreviewed, add JSBench to the skipped list for now since it doesn't
1859         work currently.
1860
1861         * Skipped:
1862
1863 2016-05-17  Keith Miller  <keith_miller@apple.com>
1864         We should have JSBench in PerformanceTests
1865         https://bugs.webkit.org/show_bug.cgi?id=157952
1866
1867         Rubber-stamped by Saam Barati.
1868
1869         There are some slight changes to the layout of the test directory
1870         to make it work nicely with run-jsc-benchmarks. Before JSBench
1871         had each of the browser specific sub-tests in a sub-directory.
1872         These have been flattened e.g. amazon/safari/ has become
1873         amazon-safari/.
1874
1875         * JSBench/amazon-chrome-win/urem.html: Added.
1876         * JSBench/amazon-chrome-win/urem.js: Added.
1877         * JSBench/amazon-chrome/urem.html: Added.
1878         * JSBench/amazon-chrome/urem.js: Added.
1879         * JSBench/amazon-firefox-win/urm.html: Added.
1880         * JSBench/amazon-firefox-win/urm.js: Added.
1881         * JSBench/amazon-firefox/urm.html: Added.
1882         * JSBench/amazon-firefox/urm.js: Added.
1883         * JSBench/amazon-safari/urem.html: Added.
1884         * JSBench/amazon-safari/urem.js: Added.
1885         * JSBench/browsercheck.js: Added.
1886         * JSBench/facebook-chrome-win/urem.html: Added.
1887         * JSBench/facebook-chrome-win/urem.js: Added.
1888         * JSBench/facebook-chrome/urem.html: Added.
1889         * JSBench/facebook-chrome/urem.js: Added.
1890         * JSBench/facebook-firefox-win/urem.html: Added.
1891         * JSBench/facebook-firefox-win/urem.js: Added.
1892         * JSBench/facebook-firefox/urem.html: Added.
1893         * JSBench/facebook-firefox/urem.js: Added.
1894         * JSBench/facebook-safari/urem.html: Added.
1895         * JSBench/facebook-safari/urem.js: Added.
1896         * JSBench/google-chrome-win/urem.html: Added.
1897         * JSBench/google-chrome-win/urem.js: Added.
1898         * JSBench/google-chrome/urem.html: Added.
1899         * JSBench/google-chrome/urem.js: Added.
1900         * JSBench/google-firefox-win/urem.html: Added.
1901         * JSBench/google-firefox-win/urem.js: Added.
1902         * JSBench/google-firefox/uem.html: Added.
1903         * JSBench/google-firefox/uem.js: Added.
1904         * JSBench/google-safari/urem.html: Added.
1905         * JSBench/google-safari/urem.js: Added.
1906         * JSBench/harness.html: Added.
1907         * JSBench/harness.js: Added.
1908         * JSBench/harness.py: Added.
1909         * JSBench/index.html: Added.
1910         * JSBench/reload.html: Added.
1911         * JSBench/twitter-chrome-win/rem.html: Added.
1912         * JSBench/twitter-chrome-win/rem.js: Added.
1913         * JSBench/twitter-chrome/urem.html: Added.
1914         * JSBench/twitter-chrome/urem.js: Added.
1915         * JSBench/twitter-firefox-win/urem.html: Added.
1916         * JSBench/twitter-firefox-win/urem.js: Added.
1917         * JSBench/twitter-firefox/urem.html: Added.
1918         * JSBench/twitter-firefox/urem.js: Added.
1919         * JSBench/twitter-safari/urem.html: Added.
1920         * JSBench/twitter-safari/urem.js: Added.
1921         * JSBench/yahoo-chrome-win/urem.html: Added.
1922         * JSBench/yahoo-chrome-win/urem.js: Added.
1923         * JSBench/yahoo-chrome/urem.html: Added.
1924         * JSBench/yahoo-chrome/urem.js: Added.
1925         * JSBench/yahoo-firefox-win/urem.html: Added.
1926         * JSBench/yahoo-firefox-win/urem.js: Added.
1927         * JSBench/yahoo-firefox/urem.html: Added.
1928         * JSBench/yahoo-firefox/urem.js: Added.
1929         * JSBench/yahoo-safari/urem.html: Added.
1930         * JSBench/yahoo-safari/urem.js: Added.
1931
1932 2016-05-20  Ryosuke Niwa  <rniwa@webkit.org>
1933
1934         run-benchmark's results should contain Animometer's debug output
1935         https://bugs.webkit.org/show_bug.cgi?id=157941
1936
1937         Reviewed by Stephanie Lewis.
1938
1939         Made developer.html support the JSON generated by run-benchmark which stores everything under debugOutput.
1940
1941         * Animometer/resources/debug-runner/animometer.js:
1942         (Utilities.initialize): Unwrap debugOutput in the case run-benchmark's result JSON is used.
1943
1944 2016-05-18  Timothy Hatcher  <timothy@apple.com>
1945
1946         Make Animometer work in all browsers
1947         https://bugs.webkit.org/show_bug.cgi?id=157855
1948         rdar://problem/26338521
1949
1950         Reviewed by Jon Lee.
1951
1952         * Animometer/resources/runner/logo.svg: Use a symbol instead
1953         of putting the id on the root svg. This works in all browsers.
1954
1955 2016-05-07  Timothy Hatcher  <timothy@apple.com>
1956
1957         Update Animometer design
1958         https://bugs.webkit.org/show_bug.cgi?id=157449
1959
1960         Reviewed by Darin Adler.
1961
1962         * Animometer/developer.html: Stop including runner/animometer.css.
1963         * Animometer/index.html: Udpated elements to fit new styles.
1964
1965         * Animometer/resources/debug-runner/animometer.css:
1966         Copy the old runner/animometer.css to here so the debug runner is unaffected.
1967
1968         * Animometer/resources/runner/animometer.css:
1969         New styles for the design. Drop flex box for broad browser support.
1970
1971         * Animometer/resources/runner/animometer.js:
1972         (ResultsTable.clear): Use textContent.
1973         (window.sectionsManager.showSection): Add class to the body.
1974         (window.sectionsManager.setSectionScore): Use textContent, not innerHTML.
1975
1976         * Animometer/resources/runner/crystal.svg: Added.
1977         * Animometer/resources/runner/lines.svg: Added.
1978         * Animometer/resources/runner/logo.svg: Added.
1979
1980         * Animometer/resources/runner/tests.js: Shorten test name and title case them.
1981
1982 2016-05-06  Manuel Rego Casasnovas  <rego@igalia.com>
1983
1984         [css-grid] Unprefix CSS Grid Layout properties
1985         https://bugs.webkit.org/show_bug.cgi?id=157137
1986
1987         Reviewed by Simon Fraser.
1988
1989         Remove "-webkit" prefix from all the grid layout properties,
1990         including the display value.
1991         Update the source code to remove the prefix where it was used too.
1992
1993         * Layout/auto-grid-lots-of-data.html:
1994         * Layout/fixed-grid-lots-of-data.html:
1995         * Layout/fixed-grid-lots-of-stretched-data.html:
1996
1997 2016-04-25  Geoffrey Garen  <ggaren@apple.com>
1998
1999         bmalloc: Misc improvements to MallocBench
2000         https://bugs.webkit.org/show_bug.cgi?id=157004
2001
2002         Reviewed by Darin Adler.
2003
2004         * MallocBench/run-malloc-benchmarks: Added --memory and --memory_warning
2005         modes for focused memory testing.
2006
2007         * MallocBench/MallocBench/Benchmark.cpp:
2008         (Benchmark::printReport): Clarified output.
2009
2010         (Benchmark::currentMemoryBytes): Added compressed memory because top
2011         does the same. (It always happens to zero in the benchmarks we run. But
2012         this is good for sanity.)
2013
2014         * MallocBench/MallocBench/CommandLine.cpp: Moved up to 8 runs to reduce
2015         variance.
2016
2017         * MallocBench/MallocBench/alloc_free.cpp:
2018         (benchmark_alloc_free): Cycle a single allocation in order to stress
2019         the effect of merging on calls to madvise.
2020
2021         * MallocBench/MallocBench/big.cpp:
2022         (benchmark_big): Graduated to 8kB-128kB because medium tests up to 8 and
2023         our large allocator doesn't kick in until 64kB.
2024
2025         * MallocBench/MallocBench/medium.cpp:
2026         (benchmark_medium): Test all the way down to 1kB because our large
2027         allocator used to service 1kB allocations and 1kB is an interesting
2028         middle size where memory is unusually large but allocation throughput
2029         still matters.
2030
2031         * MallocBench/MallocBench/stress.cpp:
2032         (benchmark_stress): Reduced the churn count to match stress_aligned
2033         because this test was taking too long to complete.
2034
2035         * MallocBench/MallocBench/stress_aligned.cpp:
2036         (benchmark_stress_aligned): Our new large allocator can handle even
2037         more absurdly large values.
2038
2039 2016-04-25  Simon Fraser  <simon.fraser@apple.com>
2040
2041         Add a content animation test that uses SVG animation.
2042         https://bugs.webkit.org/show_bug.cgi?id=156827
2043
2044         * Animation/svg-animation.html: Added.
2045         * Skipped:
2046
2047 2016-04-22  Simon Fraser  <simon.fraser@apple.com>
2048
2049         Skip two content animation tests which are only meant for iOS testing.
2050
2051         * Animation/css-animation.html: Added.
2052         * Animation/raf-animation.html: Added.
2053
2054         * Skipped:
2055
2056 2016-04-20  Simon Fraser  <simon.fraser@apple.com>
2057
2058         Add content animation tests to benchmark_runner, and allow the runner to collect device data as part of the results
2059         https://bugs.webkit.org/show_bug.cgi?id=156827
2060
2061         Add two files missing from the previous commit.
2062
2063         * Animation/css-animation.html: Added.
2064         * Animation/raf-animation.html: Added.
2065
2066 2016-04-20  Simon Fraser  <simon.fraser@apple.com>
2067
2068         Add content animation tests to benchmark_runner, and allow the runner to collect device data as part of the results
2069         https://bugs.webkit.org/show_bug.cgi?id=156827
2070
2071         Reviewed by Stephanie Lewis.
2072
2073         Two new files for the content-animation suite.
2074
2075         After patching for use in the benchmark, these tests are a little different to other benchmarks.
2076         The test content itself does not collect performance data; the tests spit out a boilerplate JSON
2077         result with a placeholder for framerate, which is collected in native code.
2078
2079         When run as a benchmark, test completion involves a couple of bounces between the test and MobileSafari.
2080         Test completion is initiated by a setTimeout() in the test, at which point location.hash is set to "#done".
2081         MobileSafari detects that, and asynchronously dumps performance data. Once this is complete, MobileSafari
2082         changes the URL has to "#submit", which triggers the XHR to the test relay that indicates the test is
2083         complete.
2084
2085         * Animation/css-animation.html: Added.
2086         * Animation/raf-animation.html: Added.
2087
2088 2016-04-15  Jon Lee  <jonlee@apple.com>
2089
2090         Animometer test could report a NaN
2091         https://bugs.webkit.org/show_bug.cgi?id=156646
2092
2093         Reviewed by Darin Adler.
2094         Provisionally reviewed by Said Abou-Hallawa.
2095
2096         * Animometer/tests/resources/main.js:
2097         (didFinishInterval): The ramp controller has a first phase where it ramps up the complexity and
2098         reacts based on how well the system handles the load. The assumption was that it would handle at least
2099         1 particle easily. That is not always the case. As a result, an interpolation calculation could
2100         end up setting an upper bound of NaN.
2101
2102         This occurs because we never get out of the first tier, so this._lastTierComplexity is undefined.
2103         Now that we guarantee a minimum complexity of 1, modify the conditional to check for this._lastTierComplexity
2104         before interpolating the upper bound of the first ramp. In the case where the system struggles with
2105         1 particle, set it to currentComplexity.
2106
2107 2016-04-09  Jon Lee  <jonlee@apple.com>
2108
2109         Update Animometer to accommodate different screens
2110         https://bugs.webkit.org/show_bug.cgi?id=156449
2111
2112         Reviewed by Darin Adler.
2113         Provisionally reviewed by Said Abou-Hallawa.
2114
2115         * Animometer/index.html: Wrap button in a container to add padding at the bottom.
2116         * Animometer/resources/debug-runner/animometer.css:
2117         (@media screen and (min-device-width: 1800px)): Deleted.
2118         (@media screen and (min-width: 1800px)): Cannot use min-device-width since it may match incorrectly.
2119         (screen and (max-device-height: 414px) and (orientation: landscape)): Some devices swap device width
2120         and height with orientation change.
2121         * Animometer/resources/runner/animometer.css: Similar.
2122         (screen and (min-device-width: 1024px) and (orientation: landscape)):
2123         (screen and (max-device-height: 414px) and (orientation: landscape)):
2124         (.frame-container): On smaller iPhones, adding 1px prevents the navigation bars from appearing.
2125         (@media screen and (min-device-width: 768px) and (max-device-width: 1024px)): Deleted.
2126         (@media (min-device-height: 768px) and (max-device-height: 1024px)): Target iPad Airs and similar.
2127         (@media screen and (min-device-width: 1024px) and (max-device-width: 1366px)): Deleted.
2128         (@media screen and (max-device-width: 1024px) and (min-device-height: 1366px)): Target iPad Pro.
2129         (#results footer): Add padding below the button for testing again.
2130         * Animometer/tests/master/multiply.html: Remove the center text.
2131         * Animometer/tests/master/resources/text.js: Update the test so that in every frame the text moves.
2132         * Animometer/tests/master/text.html: Update the text sizing depending on the size of the device.
2133
2134 2016-04-08  Jon Lee  <jonlee@apple.com>
2135
2136         Have Animometer benchmark always start with complexity of 1
2137         https://bugs.webkit.org/show_bug.cgi?id=156432
2138
2139         Reviewed by Ryosuke Niwa.
2140
2141         * Animometer/tests/resources/main.js: Update the default Controller and RampController to
2142         set its minimum complexities to 1 instead of 0.
2143
2144 2016-04-08  Jon Lee  <jonlee@apple.com>
2145
2146         Fix SVG benchmark test
2147         https://bugs.webkit.org/show_bug.cgi?id=156410
2148
2149         Reviewed by Dean Jackson.
2150
2151         * Animometer/resources/extensions.js: Update Point.zero to be a static Point.
2152         * Animometer/tests/simple/resources/tiled-canvas-image.js:
2153         (Stage.call._setupTiles): Refactor.
2154         * Animometer/tests/master/resources/particles.js:
2155         (Particle.prototype.reset): Use Point.center.
2156         (complexity): We are not using a gradient background anymore, so remove the +1.
2157         * Animometer/tests/master/resources/svg-particles.js: Update to use SVG transform
2158         instead of CSS transform.
2159
2160 2016-04-06  Jon Lee  <jonlee@apple.com>
2161
2162         Update master benchmark with SVG test
2163         https://bugs.webkit.org/show_bug.cgi?id=156273
2164
2165         Reviewed by Dean Jackson.
2166         Provisionally reviewed by Said Abou-Hallawa.
2167
2168         Switch masks tests for SVG path test.
2169
2170         * Animometer/resources/debug-runner/tests.js: Move mask test here.
2171         * Animometer/resources/runner/tests.js: Add SVG test here.
2172         * Animometer/tests/dom/particles.html: Renamed from PerformanceTests/Animometer/tests/master/particles.html.
2173         * Animometer/tests/dom/resources/dom-particles.js: Renamed from PerformanceTests/Animometer/tests/master/resources/dom-particles.js.
2174
2175         * Animometer/tests/master/resources/particles.js: Add minPosition for bounds checking. Prevents particle from being
2176         partially obscured.
2177         * Animometer/tests/master/resources/svg-particles.js: Added.
2178         (Particle): The particle is either a path object or a rect using a path as a clip. The
2179         same path is used either way. For each particle create a linear gradient with a random
2180         rotation.
2181         (SVGParticlesStage): Look in #shapes to see how many different kinds of paths are available.
2182         This makes the test more generic in case other shapes need to be tested.
2183         * Animometer/tests/master/svg-particles.html: Added. Have two defs, one that houses each
2184         particle's gradient, and one that holds the shape templates.
2185
2186 2016-03-27  Hunseop Jeong  <hs85.jeong@samsung.com>
2187
2188         [EFL] REGRESSION(r188793): It made 200 layout tests and Bindings/event-target-wrapper.html performance test fail
2189         https://bugs.webkit.org/show_bug.cgi?id=148470 
2190
2191         Reviewed by Darin Adler.
2192
2193         * Skipped: Unskip the Bindings/event-target-wrapper test.
2194
2195 2016-03-23  Jon Lee  <jonlee@apple.com>
2196
2197         Update focus and leaves tests
2198         https://bugs.webkit.org/show_bug.cgi?id=155825
2199
2200         Reviewed by Simon Fraser.
2201
2202         Particles would improperly disappear, before they were fully off stage. Update the calculation.
2203         * Animometer/tests/dom/resources/leaves.js:
2204         * Animometer/tests/master/resources/leaves.js:
2205
2206         * Animometer/tests/master/focus.html: Update the center element styling.
2207         * Animometer/tests/master/resources/focus.js:
2208         (FocusElement.Utilities.createClass): Give enough space in the container for each particle for the blur
2209         to spread. It should be about 3x more space than the blur radius on each side.
2210         (Stage.call.initialize): Ditto for the center element.
2211
2212 2016-03-23  Jon Lee  <jonlee@apple.com>
2213
2214         Update support for other platforms
2215         https://bugs.webkit.org/show_bug.cgi?id=155824
2216
2217         Reviewed by Simon Fraser.
2218
2219         * Animometer/resources/debug-runner/animometer.css: Remove extraneous media query. All of it should
2220         be handled in the release stylesheet.
2221         * Animometer/resources/debug-runner/animometer.js: Add support for browsers that return an HTMLCollection
2222         when retrieving radio elements with a name, instead of the selected value.
2223         * Animometer/resources/runner/animometer.css: Adjust the queries for devices of varying widths.
2224         (.frame-container): Move overflow:hidden to #stage.
2225         * Animometer/tests/master/resources/stage.css: Consolidate into tests/resources/stage.css.
2226         * Animometer/tests/resources/stage.css:
2227         (body): Remove overflow:hidden.
2228         (#stage): Set overflow:hidden here.
2229         (#center-text): Moved from master/resources/stage.css.
2230
2231         Update stylesheet hrefs.
2232         * Animometer/tests/master/canvas-stage.html:
2233         * Animometer/tests/master/focus.html:
2234         * Animometer/tests/master/leaves.html:
2235         * Animometer/tests/master/multiply.html:
2236         * Animometer/tests/master/particles.html:
2237         * Animometer/tests/master/text.html: Adjust text size to fit in the device screen.
2238
2239 2016-03-23  Jon Lee  <jonlee@apple.com>
2240
2241         Make the benchmark require device to be in landscape orientation
2242         https://bugs.webkit.org/show_bug.cgi?id=155822
2243         rdar://problem/25258650
2244
2245         Reviewed by Ryosuke Niwa.
2246
2247         Check that the device is in landscape orientation. If not, disable the button to start it,
2248         and show a warning message.
2249
2250         * Animometer/developer.html: Add a message about how the browser should be set up. Include a
2251         warning message that appears if the orientation is incorrect on mobile devices.
2252         * Animometer/index.html: Ditto.
2253
2254         * Animometer/resources/debug-runner/animometer.css: Migrate the .hidden rule into the release
2255         stylesheet. Update the style to accommodate the new UI.
2256         * Animometer/resources/runner/animometer.css:
2257
2258         * Animometer/resources/runner/animometer.js:
2259         (window.benchmarkController.initialize): Add the orientation listener is needed.
2260         (window.benchmarkController.addOrientationListenerIfNecessary): Only mobile devices need this. Check
2261         to see for support of window.orientation.
2262         (window.benchmarkController._orientationChanged): Toggle the warning based on whether we match the
2263         landscape query. Set a state variable, which is needed for the debug harness. Call updateStartButtonState.
2264         (window.benchmarkController.updateStartButtonState):
2265         * Animometer/resources/debug-runner/animometer.js: Change _updateStartButtonState to return
2266         a boolean about whether at least one test is selected. That will be used in
2267         benchmarkController.updateStartButtonState(). Move the load event listener to the release version.
2268         (window.benchmarkController.updateStartButtonState): Override the release version, and also check
2269         that at least one test is selected.
2270
2271 2016-03-22  Geoffrey Garen  <ggaren@apple.com>
2272
2273         MallocBench: consolidate regression testing for aligned allocation
2274         https://bugs.webkit.org/show_bug.cgi?id=155762
2275
2276         Reviewed by Michael Saboff.
2277
2278         * MallocBench/MallocBench.xcodeproj/project.pbxproj:
2279         * MallocBench/MallocBench/Benchmark.cpp:
2280         * MallocBench/MallocBench/memalign.cpp: Removed.
2281         * MallocBench/MallocBench/memalign.h: Removed. The stress_aligned test
2282         covers this and much more.
2283
2284         * MallocBench/MallocBench/stress_aligned.cpp:
2285         (benchmark_stress_aligned): Include specific tests for extreme sizes
2286         and alignments.
2287
2288 2016-03-21  Jon Lee  <jonlee@apple.com>
2289
2290         Update benchmark tests
2291         https://bugs.webkit.org/show_bug.cgi?id=155723
2292
2293         Reviewed by Darin Adler.
2294         Provisionally reviewed by Said Abou-Hallawa.
2295
2296         Add quadratic and bezier segments to the canvas path test.
2297
2298         * Animometer/resources/runner/tests.js: Some of the query strings are unnecessary.
2299         Rename the test.
2300         * Animometer/tests/master/resources/canvas-stage.js:
2301         (tune): Update to be able to take an array of possible constructors. Choose one randomly.
2302         * Animometer/tests/master/resources/canvas-tests.js:
2303         (CanvasLinePoint.Utilities.createClass): Move the point selection out to a separate
2304         function called randomPoint() for reuse. Move X_LOOPS, Y_LOOPS, and offsets into the
2305         class definition.
2306         (randomPoint): Scale the grid down a little bit so that the lines along the edge of the
2307         canvas are not cut off when the stroke size is thick.
2308         (CanvasQuadraticSegment): Added.
2309         (CanvasBezierSegment): Added.
2310         (SimpleCanvasStage): Pass in an array of the different segment types. Since line segments
2311         are short compared to the curved ones, make it twice as likely to render a line segment.
2312         (SimpleCanvasStage.animate): Update the drawing code so that we render all line segments.
2313
2314         Add a helper method that selects a random element from an array.
2315
2316         * Animometer/tests/resources/main.js:
2317         (Stage.randomElementInArray): Select a random element from the provided array.
2318
2319         * Animometer/tests/bouncing-particles/resources/bouncing-tagged-images.js: Refactor.
2320         * Animometer/tests/dom/resources/leaves.js: Ditto.
2321         * Animometer/tests/master/resources/dom-particles.js: Ditto.
2322         * Animometer/tests/master/resources/image-data.js: Ditto.
2323         * Animometer/tests/master/resources/leaves.js: Ditto.
2324         * Animometer/tests/simple/resources/simple-canvas-paths.js: Ditto.
2325
2326         Add canvas tests that includes all stroke and fill paths. This makes it possible to avoid having to
2327         include the full simple canvas suite for perf testing.
2328
2329         * Animometer/resources/debug-runner/tests.js: Add new tests. Move the canvas test into the Canvas
2330         suite.
2331         * Animometer/tests/simple/resources/simple-canvas-paths.js: Add a CanvasStroke and CanvasFill particle
2332         that random selects an object to render.
2333
2334         * Animometer/resources/debug-runner/tests.js: Move 3D suite before basic canvas suite.
2335         When updating the perf bot script, we will include the suites up to this one, but exclude
2336         the basic canvas suite.
2337
2338         Move compositing transforms test to HTML suite and remove the empty Miscellaneous suite.
2339
2340         * Animometer/resources/debug-runner/tests.js:
2341         * Animometer/tests/dom/compositing-transforms.html: Renamed from PerformanceTests/Animometer/tests/misc/compositing-transforms.html.
2342         * Animometer/tests/dom/resources/compositing-transforms.js: Renamed from PerformanceTests/Animometer/tests/misc/resources/compositing-transforms.js.
2343
2344         Clean up miscellaneous test suite. Add a canvas ellipse test, and remove the other
2345         canvas tests.
2346
2347         * Animometer/resources/debug-runner/tests.js: Add ellipse tests to the simple suite.
2348         * Animometer/tests/misc/canvas-electrons.html: Removed.
2349         * Animometer/tests/misc/canvas-stars.html: Removed.
2350         * Animometer/tests/misc/resources/canvas-electrons.js: Removed.
2351         * Animometer/tests/misc/resources/canvas-stars.js: Removed.
2352         * Animometer/tests/simple/resources/simple-canvas-paths.js: Add ellipse primitives.
2353
2354         Merge text tests together into one.
2355
2356         * Animometer/resources/runner/tests.js: Remove international.html.
2357         * Animometer/tests/master/international.html: Removed.
2358         * Animometer/tests/master/resources/text.js:
2359         (animate): Update styling. Manually calculate gradients for the shadow particles.
2360         Reduce the step size for y direction to avoid cutting text off at the margins. Increase
2361         step for x since there will be a little more room.
2362         * Animometer/tests/master/text.html: Add more translations and lay it out in a table.
2363
2364 2016-03-20  Jon Lee  <jonlee@apple.com>
2365
2366         Add a link to show debug data
2367         https://bugs.webkit.org/show_bug.cgi?id=155724
2368
2369         Reviewed by Simon Fraser.
2370
2371         * Animometer/developer.html: Attach the onclick handler to the score.
2372         * Animometer/index.html: Ditto.
2373
2374 2016-03-18  Dean Jackson  <dino@apple.com>
2375
2376         Add a basic WebGL test to Animometer
2377         https://bugs.webkit.org/show_bug.cgi?id=155475
2378         <rdar://problem/25156860>
2379
2380         Reviewed by Simon Fraser and Jon Lee.
2381
2382         Add a pretty simple WebGL test to Animometer
2383         which tries to draw a number of animated triangles.
2384         The result still needs a bit of tuning.
2385
2386         * Animometer/resources/debug-runner/tests.js: Add a new
2387         Suite for "3D" tests.
2388         * Animometer/tests/3d/resources/webgl.js: Added.
2389         * Animometer/tests/3d/webgl.html: Added.
2390
2391 2016-03-18  Jon Lee  <jonlee@apple.com>
2392
2393         Add support for statically linking to a specific test
2394         https://bugs.webkit.org/show_bug.cgi?id=155631
2395
2396         Rubber-stamped by Darin Adler.
2397
2398         * Animometer/developer.html: Update to call restartBenchmark instead so that it works for
2399         both the preset and the manually set test sessions.
2400         * Animometer/resources/debug-runner/animometer.css:
2401         (.tree .link): Style the "link" UI.
2402         * Animometer/resources/debug-runner/animometer.js: Move updateDisplay from suitesManager
2403         to optionsManager since this is a visual option and has nothing to do with the suites
2404         settings. Update the construction of the test UI by including a "link" after each test.
2405         Clicking on that link brings up a JS prompt with a URL and a query string with the current
2406         parameters of the controller, and selected test. Pasting this URL into the location bar will
2407         automatically start running the selected test.
2408         (suitesManager.suitesFromQueryString): Iterate through the Suites and tests and find the one
2409         that matches the provided parameters. Returns an object similar to the form in
2410         suitesManager.updateLocalStorageFromUI.
2411         (benchmarkController.initialize): After settings up the events and options, try parsing the
2412         query string and running the benchmark immediately. Otherwise, fall back to the form.
2413         (benchmarkController.startBenchmark): Store the options and suites into member variables for
2414         reuse in restartBenchmark.
2415         (benchmarkController.startBenchmarkImmediatelyIfEncoded): Convert the query string to an object.
2416         If that's successful, find the suite and test referenced in the query string. Start the benchmark if
2417         the search for the test succeeded.
2418         * Animometer/resources/extensions.js:
2419         (Utilities.stripNonASCIICharacters): Helper method to convert the name of the suite and test
2420         into a query-string-friendly version.
2421         (Utilities.convertObjectToQueryString): Helper method to convert an object to query string
2422         format.
2423         (Utilities.convertQueryStringToObject): Helper method to convert query string into an object
2424         with properties and values.
2425
2426 2016-03-16  Jon Lee  <jonlee@apple.com>
2427
2428         Add a new benchmark test
2429         https://bugs.webkit.org/show_bug.cgi?id=155570
2430
2431         Reviewed by Simon Fraser.
2432
2433         New Leaves test includes various image sizes and opacity.
2434
2435         * Animometer/resources/debug-runner/tests.js: Add it to the HTML test suite.
2436         * Animometer/tests/dom/leaves.html: Added.
2437         * Animometer/tests/dom/resources/leaves.js: Added. Override the
2438         (Particle.call.reset): Uses a range of sizes, and opacity.
2439         (Particle.call.animate): Opacity goes up then down. When it hits 0, reset the particle.
2440         (Particle.call.move): Set transform and opacity.
2441         * Animometer/tests/master/resources/leaves.js: Get rid of the closure so that it
2442         can be used in this test. Update the relative path so that it works in both the master
2443         and dom test suite.
2444
2445 2016-03-15  Simon Fraser  <simon.fraser@apple.com>
2446
2447         Add developer Animometer test that bounces P3-tagged images
2448         https://bugs.webkit.org/show_bug.cgi?id=155511
2449
2450         Reviewed by Tim Horton.
2451         
2452         Add a test for rendering performance of tagged images. The 5 images are tagged
2453         with the Display P3 colorspace.
2454
2455         * Animometer/resources/debug-runner/tests.js:
2456         * Animometer/tests/bouncing-particles/bouncing-tagged-images.html: Added.
2457         * Animometer/tests/bouncing-particles/resources/bouncing-tagged-images.js: Added.
2458         * Animometer/tests/bouncing-particles/resources/image1.jpg: Added.
2459         * Animometer/tests/bouncing-particles/resources/image2.jpg: Added.
2460         * Animometer/tests/bouncing-particles/resources/image3.jpg: Added.
2461         * Animometer/tests/bouncing-particles/resources/image4.jpg: Added.
2462         * Animometer/tests/bouncing-particles/resources/image5.jpg: Added.
2463
2464 2016-03-09  Jon Lee  <jonlee@apple.com>
2465
2466         Enhance existing Animometer tests
2467         https://bugs.webkit.org/show_bug.cgi?id=155261
2468
2469         Reviewed by Simon Fraser.
2470
2471         * Animometer/tests/master/resources/canvas-tests.js: Add a gradient to
2472         the filled circles. To expose more of the gradient, add another circle.
2473         (SimpleCanvasStage.call.animate): For each frame, create a gradient with
2474         undulating stop points and colors. Fill the circles twice; once with
2475         the solid color, and once with the gradient.
2476
2477         * Animometer/tests/master/resources/dom-particles.js: Refactor the
2478         emission variables into a separate stage for this test. Add a colorOffset
2479         variable to make the colors of each particle slightly different, since
2480         the ramp controller can add large numbers of particles all at once, which
2481         would otherwise get all the same color.
2482         * Animometer/tests/master/resources/particles.js:
2483         (initialize): Remove the code specific to the SVG mask test.
2484
2485         * Animometer/tests/master/resources/image-data.js:
2486         (initialize): Remove unused local variable.
2487         * Animometer/tests/master/resources/multiply.js:
2488         (initialize): Make the test harder by adding more total particles.
2489
2490 2016-03-09  Jon Lee  <jonlee@apple.com>
2491
2492         Add text tests
2493         https://bugs.webkit.org/show_bug.cgi?id=155257
2494
2495         Reviewed by Simon Fraser.
2496
2497         * Animometer/resources/extensions.js:
2498         (UnitBezier.Utilities.createClass): Add a class that computes Bezier points
2499         assuming that two of the control points are at (0,0) and (1,1). Taken from
2500         WebCore/platform/graphics/UnitBezier.h
2501         * Animometer/resources/runner/tests.js: Add a test for English text and one
2502         for international text.
2503         * Animometer/tests/master/international.html: Added.
2504         * Animometer/tests/master/resources/text.js: Added. The test assumes there is
2505         #template div which it will copy. The copies are placed behind the template, and
2506         are set with different colors each frame. They are moved around with CSS transform.
2507         * Animometer/tests/master/text.html: Added.
2508
2509         Remove the other text tests, since these ones cover the same techniques.
2510         * Animometer/tests/text/layering-text.html: Removed.
2511         * Animometer/tests/text/resources/layering-text.js: Removed.
2512         * Animometer/tests/text/resources/text-boxes.js: Removed.
2513         * Animometer/tests/text/text-boxes.html: Removed.
2514         * Animometer/resources/debug-runner/tests.js:
2515
2516 2016-03-09  Jon Lee  <jonlee@apple.com>
2517
2518         Add a new image test
2519         https://bugs.webkit.org/show_bug.cgi?id=155232
2520
2521         Reviewed by Dean Jackson.
2522         Provisionally reviewed by Said Abou-Hallawa.
2523
2524         The image test renders PNGs and moves them with translate and rotate
2525         transforms. Each particle has a lifetime, and when the lifetime ends or
2526         the particle goes offscreen, it resets itself somewhere on the stage.
2527
2528         * Animometer/resources/debug-runner/tests.js: Remove the CSS bouncing PNG
2529         images test, because this one tests the same technique.
2530         * Animometer/resources/runner/tests.js: Added here as "Leaves".
2531         * Animometer/tests/master/leaves.html: Added.
2532         * Animometer/tests/master/resources/particles.js: Refactor out the parts
2533         specific to the DOM particles test. Consequently make velocity a public
2534         member.
2535         (initialize): The options parameter is never used, so remove it.
2536         (animate): Remove unused local variable.
2537         * Animometer/tests/master/resources/leaves.js: Maintains a focal point that
2538         moves back and forth across the canvas. That point affects the velocity
2539         of the particles. When the focal point is on the leftmost side, it is
2540         between [-6, -2], and set in reset(). When the focal point is on the rightmost
2541         side, it will be from [2, 6].
2542         * Animometer/tests/master/resources/dom-particles.js: Move JS specific to
2543         this test here from particles.js.
2544
2545         New images.
2546         * Animometer/tests/master/resources/compass100.png: Added.
2547         * Animometer/tests/master/resources/console100.png: Added.
2548         * Animometer/tests/master/resources/contribute100.png: Added.
2549         * Animometer/tests/master/resources/debugger100.png: Added.
2550         * Animometer/tests/master/resources/inspector100.png: Added.
2551         * Animometer/tests/master/resources/layout100.png: Added.
2552         * Animometer/tests/master/resources/performance100.png: Added.
2553         * Animometer/tests/master/resources/script100.png: Added.
2554         * Animometer/tests/master/resources/shortcuts100.png: Added.
2555         * Animometer/tests/master/resources/standards100.png: Added.
2556         * Animometer/tests/master/resources/storage100.png: Added.
2557         * Animometer/tests/master/resources/styles100.png: Added.
2558         * Animometer/tests/master/resources/timeline100.png: Added.
2559
2560 2016-03-03  Jon Lee  <jonlee@apple.com>
2561
2562         Add ability to retrieve raw data from release harness
2563         https://bugs.webkit.org/show_bug.cgi?id=155026
2564
2565         Reviewed by Simon Fraser.
2566
2567         * Animometer/developer.html: Remove the special UI in the debug
2568         harness.
2569         * Animometer/resources/runner/animometer.css: Add styles for the
2570         overlay.
2571         * Animometer/resources/runner/animometer.js: Let 'j' show the JSON
2572         results, but only if the overlay doesn't exist. Add 'esc' key to dismiss
2573         the overlay.
2574         (window.benchmarkController.selectResults): Cycle the cases around
2575         so that the first 's' press selects both the benchmark score and
2576         the individual test scores.
2577         * Animometer/resources/debug-runner/animometer.css: Remove
2578         unneeded rules.
2579         * Animometer/resources/debug-runner/animometer.js: Make the same
2580         call to handleKeyPress.
2581
2582 2016-03-03  Jon Lee  <jonlee@apple.com>
2583
2584         Make sure multiply test particles have at least some opacity
2585         https://bugs.webkit.org/show_bug.cgi?id=155027
2586
2587         Reviewed by Simon Fraser.
2588
2589         * Animometer/tests/master/resources/multiply.js:
2590         (initialize): Have the elements spin a little faster to make it
2591         more obvious when the system is being stressed.
2592         (_addTile):
2593         (animate): Make sure each element being animated has at least 1%
2594         opacity.
2595
2596 2016-03-03  Jon Lee  <jonlee@apple.com>
2597
2598         Update image test
2599         https://bugs.webkit.org/show_bug.cgi?id=154962
2600
2601         Rubber-stamped by Darin Adler.
2602
2603         * Animometer/tests/master/image-data.html: Make each canvas have a compositing layer.
2604         * Animometer/tests/master/resources/image-data.js:
2605         (initialize): Update to load multiple images serially.
2606         (_loadImage):
2607         (tune): Use display instead of visibility. When showing an element anew, move it somewhere else.
2608         (_createTestElement): Set up the element, refactor out placement of element to _refreshElement.
2609         (_refreshElement): Place element in a tile grid.
2610         (animate):
2611         (_getRandomNeighboringPixelIndex): Use a more conservative distribution to make the effect last a little longer.
2612         * Animometer/tests/master/resources/compass.svg: Added.
2613         * Animometer/tests/master/resources/console.svg: Added.
2614         * Animometer/tests/master/resources/contribute.svg: Added.
2615         * Animometer/tests/master/resources/debugger.svg: Added.
2616         * Animometer/tests/master/resources/inspector.svg: Added.
2617         * Animometer/tests/master/resources/layout.svg: Added.
2618         * Animometer/tests/master/resources/performance.svg: Added.
2619         * Animometer/tests/master/resources/script.svg: Added.
2620         * Animometer/tests/master/resources/shortcuts.svg: Added.
2621         * Animometer/tests/master/resources/standards.svg: Added.
2622         * Animometer/tests/master/resources/storage.svg: Added.
2623         * Animometer/tests/master/resources/styles.svg: Added.
2624         * Animometer/tests/master/resources/timeline.svg: Added.
2625
2626 2016-03-02  Jon Lee  <jonlee@apple.com>
2627
2628         Add some new controllers, and refine tests
2629         https://bugs.webkit.org/show_bug.cgi?id=154914
2630
2631         Reviewed by Simon Fraser.
2632
2633         Improve tests.
2634
2635         * Animometer/tests/master/focus.html: Move each particle into a layer. Put the particle
2636         in a container element. Expand the container element by the max blur radius, and clip
2637         overflow. This way, when the blur is applied, it avoids causing layer resizes.
2638         * Animometer/tests/master/resources/focus.js:
2639         (FocusElement.Utilities.createClass): Create a container element, and put the particle
2640         inside.
2641         (hide): Set display:none.
2642         (show): Set display:block.
2643         (animate): Apply filters to the container element.
2644         (FocusStage.call.initialize): Instead of inserting and removing elements from the DOM,
2645         keep them in the stage, but set the display style instead. Use this._offsetIndex to
2646         keep track of which elements are displayed.
2647         (FocusStage.call.animate): Move some calculations around to avoid doing unneeded math.
2648         (FocusStage.call.getBlurValue): Make sure elements are always blurred.
2649         (FocusStage.call.getOpacityValue): Make sure elements have some opacity.
2650         * Animometer/tests/master/resources/multiply.js: Have the particles on the edge of the
2651         stage appear less black when the complexity gets large enough.
2652         (tune): Have this._distanceFactor calculate the factor. Avoid calculating square root each
2653         frame.
2654
2655 2016-03-02  Jon Lee  <jonlee@apple.com>
2656
2657         Add some new controllers, and refine tests
2658         https://bugs.webkit.org/show_bug.cgi?id=154914
2659
2660         Reviewed by Simon Fraser.
2661
2662         Add a controller that centers around 30 fps instead of 60 fps.
2663
2664         * Animometer/developer.html: Add a new option.
2665         * Animometer/resources/debug-runner/animometer.js:
2666         * Animometer/resources/runner/animometer.js:
2667         (this._processData.findRegression): When calculating the complexity-frameLength regression,
2668         check the controller, and use a 30 fps baseline if needed.
2669         * Animometer/resources/statistics.js:
2670         (Regression.Utilities.createClass): Update to allow clients to specify the baseline
2671         frame length.
2672         * Animometer/tests/resources/main.js:
2673         (tune): Override some of the constants in RampController. Move those constants out for
2674         easier reading.
2675
2676         Add a fixed controller, with no step.
2677
2678         * Animometer/developer.html: Add a controller that takes no step.
2679         * Animometer/resources/debug-runner/animometer.js:
2680         * Animometer/tests/resources/main.js:
2681         (Rotater.Utilities.createClass):
2682
2683         Switch to ramp controller as default.
2684
2685         * Animometer/developer.html: Increase the test length to 20 seconds.
2686         * Animometer/resources/debug-runner/animometer.js: Rename the "adjustment" field to "controller"
2687         since that is a more accurate description.
2688         * Animometer/resources/debug-runner/graph.js:
2689         * Animometer/resources/runner/animometer.js: Update preferences for release suite.
2690
2691 2016-02-26  Jon Lee  <jonlee@apple.com>
2692
2693         Address Dean's comments in 154673.
2694
2695         * Animometer/developer.html:
2696         * Animometer/resources/debug-runner/animometer.js:
2697         * Animometer/resources/debug-runner/graph.js:
2698
2699 2016-02-25  Jon Lee  <jonlee@apple.com>
2700
2701         Update animation benchmark and tests
2702         https://bugs.webkit.org/show_bug.cgi?id=154673
2703
2704         Reviewed by Dean Jackson.
2705
2706         Update test visuals.
2707
2708         * Animometer/tests/master/resources/canvas-tests.js:
2709         (CanvasLineSegment.Utilities.createClass): Line segments near the edge of the stage get clipped.
2710         Update circle position and radius to minimize impact.
2711         * Animometer/tests/master/resources/particles.js: Get rid of rotating gradient background, and
2712         have 3 locations for emitting particles.
2713         * Animometer/tests/master/resources/multiply.js: Update distance metric so that fringe tiles get
2714         more color.
2715         (initialize): Fix some of the math for laying out the tiles.
2716         (animate): When a tile isn't used, set visibility: hidden.
2717
2718 2016-02-24  Jon Lee  <jonlee@apple.com>
2719
2720         Update animation benchmark and tests
2721         https://bugs.webkit.org/show_bug.cgi?id=154673
2722
2723         Reviewed by Dean Jackson.
2724
2725         Update the ramp controller.
2726
2727         The controller refines the complexity interval to test across.
2728
2729         * Animometer/resources/statistics.js: Add functions that estimate cumulative distribution function.
2730         (Regression): For the flat regression, force the first segment to be at 60 fps.
2731         (valueAt): Add convenience function to return interpolated value based on the regression used.
2732         (_calculateRegression): Include the number of points included for both segments, and the piecewise
2733         errors.
2734         * Animometer/tests/resources/math.js: Make the Kalman estimator subclass Experiment, and allow it
2735         to be reset.
2736
2737         * Animometer/tests/resources/main.js: Initialize the tier such that it starts at 10^0 = 1.
2738         Increase the number of ramps. Maintain three FPS thresholds-- the frame rate of interest, a limit
2739         on the lowest FPS we care to go for later interpolation, and a minimum FPS threshold we want to
2740         aim for each ramp. Also keep three estimators: a running average of the change point, a minimum
2741         boundary for each ramp, and an estimator for all the frames within an interval. The first two
2742         are used to determine the parameters of the next ramp, and the latter allows us to refine the
2743         parameters.
2744         (update): During the tier phase, it is possible that the highest complexity possible for a test
2745         won't stress the system enough to trigger stopping the tier phase and transitioning to the ramps.
2746         If the complexity doesn't change when going to the next tier, we've maxed the test out, and move
2747         on. When the tier phase completed, turn off Controller.frameLengthEstimator, which estimates the
2748         FPS at each tier.
2749         (tune): At each interval, look at the confidence distribution of being on the 60 FPS side or the
2750         slow side. If the slowest FPS we achieve at the ramp's maximum complexity is not at least
2751         _fpsRampSlowThreshold, then increase the maximum complexity. If we ever achieve 60 FPS, increase
2752         the ramp's minimum complexity to that level. If, at an even lower complexity, a glitch causes the
2753         FPS to drop, we reset the minimum complexity.
2754
2755         Have the bootstrap calculation occur between tests. Clean up harness.
2756
2757         * Animometer/resources/debug-runner/animometer.js: Run bootstrap after a test has
2758         completed to avoid doing all of it at the end before showing the results. Clean up
2759         parameters being passed around.
2760         * Animometer/resources/debug-runner/tests.js:
2761         (text):
2762         * Animometer/resources/runner/animometer.js:
2763         (this._processData.calculateScore): Save the results to the same object holding the data.
2764         (this._processData._processData): In the case where a file is dragged, calculate the score
2765         serially. Grab the results object and move it to the results variable and remove it from
2766         the data object. This avoids serializing the results into the JSON.
2767         (this._processData.findRegression): Include the samples used for bootstrapping. Reduce the
2768         resample size to shorten the wait.
2769         * Animometer/resources/runner/benchmark-runner.js:
2770         * Animometer/resources/statistics.js:
2771         (bootstrap): Update how bootstrapData is sorted. In some regression results the mix of
2772         floats and integers causes an alphabetical sort to occur.
2773         * Animometer/resources/strings.js:
2774
2775         Add meta charset so that encodings between harness and test match.
2776
2777         * Animometer/tests/bouncing-particles/bouncing-canvas-images.html:
2778         * Animometer/tests/bouncing-particles/bouncing-canvas-shapes.html:
2779         * Animometer/tests/bouncing-particles/bouncing-css-images.html:
2780         * Animometer/tests/bouncing-particles/bouncing-css-shapes.html:
2781         * Animometer/tests/bouncing-particles/bouncing-svg-images.html:
2782         * Animometer/tests/bouncing-particles/bouncing-svg-shapes.html:
2783         * Animometer/tests/master/canvas-stage.html:
2784         * Animometer/tests/master/focus.html:
2785         * Animometer/tests/master/image-data.html:
2786         * Animometer/tests/master/multiply.html:
2787         * Animometer/tests/master/particles.html:
2788         * Animometer/tests/misc/canvas-electrons.html:
2789         * Animometer/tests/misc/canvas-stars.html:
2790         * Animometer/tests/misc/compositing-transforms.html:
2791         * Animometer/tests/simple/simple-canvas-paths.html:
2792         * Animometer/tests/simple/tiled-canvas-image.html:
2793         * Animometer/tests/template/template-canvas.html:
2794         * Animometer/tests/template/template-css.html:
2795         * Animometer/tests/template/template-svg.html:
2796         * Animometer/tests/text/layering-text.html:
2797         * Animometer/tests/text/text-boxes.html:
2798
2799         Update test harness reporting.
2800
2801         * Animometer/developer.html: Add missing meta charset.
2802         * Animometer/index.html: Remove unnecessary utf-8 declaration.
2803         * Animometer/resources/debug-runner/animometer.css: Add convenience classes for
2804         formatting the results table.
2805         * Animometer/resources/debug-runner/animometer.js: Adjust which stats are shown.
2806         * Animometer/resources/debug-runner/tests.js: Display bootstrapping statistics.
2807         * Animometer/resources/strings.js: Move strings not used by the release harness.
2808
2809         Switch to a pseudo-random number generator.
2810
2811         * Animometer/resources/statistics.js: Add a Pseudo class, with a simple
2812         pseudo-random number generator.
2813         (_calculateRegression): Reset the generator before running bootstrap.
2814         (bootstrap): Deleted.
2815
2816         Replace Math.random with Pseudo.random.
2817         * Animometer/tests/master/resources/canvas-tests.js:
2818         * Animometer/tests/master/resources/focus.js:
2819         * Animometer/tests/master/resources/particles.js:
2820         * Animometer/tests/resources/main.js:
2821
2822         Use bootstrapping to get confidence interval in the breakpoint.
2823
2824         For the ramp controller, calculate the piecewise regression, and then use
2825         bootstrapping in order to find the 95% confidence interval. Use the raw data.
2826
2827         * Animometer/developer.html: Default to the complexity graph. Add a legend
2828         checkbox to toggle visibility of the bootstrap score and histogram.
2829         * Animometer/resources/debug-runner/animometer.css: Make some more space to show
2830         the old raw and average scores in the legend. Add new styles for the data.
2831         * Animometer/resources/debug-runner/graph.js:
2832         (_addRegressionLine): Allow passing an array for the variance bar tied to the
2833         regression line. Now |stdev| is |range|.
2834         (createComplexityGraph): Add bootstrap median, and overlay a histogram of
2835         the bootstrap samples. Switch raw samples from circles to X's.
2836         (onComplexityGraphOptionsChanged): Allow toggling of the bootstrap data.
2837         (onGraphTypeChanged): Move the regressions for the raw and average samples to the
2838         legend. In the subtitle use the bootstrap median, and include the 95% confidence
2839         interval.
2840         * Animometer/resources/runner/animometer.js:
2841         (this._processData.findRegression): Factor out the code that determines which
2842         samples to include when calculating the piecewise regression. For series that have
2843         many samples, or a wider range of recorded complexities, throw away the 2.5%
2844         lowest and highest samples before calculating the regression. Keep all samples
2845         if the number of samples to regress is small or the range of complexities is
2846         narrow.
2847         (this._processData._calculateScore): Factor out regression calculation to
2848         findRegression(). Bootstrap the change point of the regression. The score is the
2849         median.
2850         * Animometer/resources/statistics.js:
2851         (_calculateRegression): Correct an issue in the calculation of the regression, where
2852         the denominator can be 0.
2853         (bootstrap): Template for bootstrapping. Create a bootstrap sample array, Create
2854         re-samples by random selection with replacement. Return the 95% confidence samples,
2855         the bootstrap median, mean, and the data itself.
2856         * Animometer/resources/strings.js: Add bootstrap.
2857         * Animometer/tests/resources/main.js:
2858         (processSamples): Don't prematurely cut the sample data.
2859
2860         Fix graph drawing.
2861
2862         * Animometer/resources/debug-runner/animometer.js: Add spacing in the JSON output.
2863         Multiple tests output a lot of JSON and can hang when selecting JSON with no whitespace.
2864         * Animometer/resources/debug-runner/animometer.css:
2865         (#complexity-graph .series.raw circle): Update the color.
2866         * Animometer/resources/debug-runner/graph.js: Use the FPS axis instead of the
2867         complexity axis, which can vary in domain. For determining the complexity domain,
2868         only use samples after samplingTimeOffset.
2869
2870         Allow dropping results JSON.
2871
2872         * Animometer/developer.html: Add a button.
2873         * Animometer/resources/debug-runner/animometer.css:
2874         * Animometer/resources/debug-runner/animometer.js: Read the data and go straight
2875         to the dashboard. With JSON output, write out only the options and the raw data.
2876
2877         Teach the harness to evaluate the samples and determine the test score.
2878
2879         This will allow us to update how the score is calculated separately from the samples recorded.
2880         This also prepares the harness to be able to accept JSON of prior runs.
2881
2882         * Animometer/resources/strings.js: Clean up and remove unneeded strings and reduce some of the
2883         hierarchy.
2884         * Animometer/resources/debug-runner/tests.js: Update to use the new strings.
2885
2886         * Animometer/tests/resources/main.js: Allow all controllers to show a complexity-FPS graph.
2887         (_processComplexitySamples): Factor out some of the sample processing done in the ramp
2888         controller for the benefit of the other controllers. |complexitySamples| contains a list of
2889         samples. Sort the samples by complexity. Optionally remove the top x% of samples.
2890         Group them, and calculate distribution of samples within the same complexity, and add those as
2891         new entries into |complexityAverageSamples|.
2892         (Controller.processSamples): Move the code responsible for determining the complexity and FPS
2893         scores out to ResultsDashboard. The structure of the data returned by the controller is:
2894
2895         {
2896             controller: [time-regression, time-regression, ...], // optional, data specific to controller
2897             marks: [...],
2898             samples: {                    // all of the sample data
2899                 controller: [...],
2900                 complexity: [...],        // processed from controller samples
2901                 complexityAverage: [...], // processed from complexity samples
2902             }
2903         }
2904
2905         (AdaptiveController.processSamples): Adding the target frame length is no longer necessary; we
2906         now pass the test options to the graph.
2907         (Regression): Move to statistics.js.
2908         * Animometer/resources/statistics.js: Move Regression to here. Add a check if the sampling range
2909         only contains one sample, since we cannot calculate a regression from one sample point.
2910
2911         Teach the test harness to evaluate the data.
2912         * Animometer/resources/runner/animometer.js:
2913         (ResultsDashboard): Store the options used to run the test and the computed results/score separately
2914         from the data. The results are stored as:
2915
2916         {
2917             score: /* geomean of iteration score */,
2918             iterationsResults: [
2919                 {
2920                     score: /* geomean of tests */,
2921                     testsResults: {
2922                         suiteName: {
2923                             testName: {
2924                                 controller: {
2925                                     average:
2926                                     concern:
2927                                     stdev:
2928                                     percent:
2929                                 },
2930                                 frameLength: { ... },
2931                                 complexity: {
2932                                     complexity:
2933                                     stdev:
2934                                     segment1:
2935                                     segment2:
2936                                 },
2937                                 complexityAverage: { ... }
2938                             },
2939                             testName: { ... },
2940                         },
2941                         ... next suite ...
2942                     }
2943                 },
2944                 { ...next iteration... }
2945             ]
2946         }
2947
2948         * Animometer/resources/debug-runner/animometer.js: Pass options around instead of relying
2949         on what was selected in the form. This will later allow for dropping previous results, and
2950         using those runs' options when calculating scores.
2951         (ResultsTable._addGraphButton): Simplify button action by using attached test data.
2952         * Animometer/resources/debug-runner/graph.js: Refactor to use the data.
2953
2954         Consolidate JS files, and move statistics out to a separate JS.
2955
2956         Preparation for having the Controller only handle recording and storage of the samples,
2957         and leave the evaluation of the test score out to the harness. Move Experiment to
2958         a new statistics.js, where Regression will also eventually go. Get rid of algorithm.js
2959         and move it to utilities.js since the Heap is used only for Experiments.
2960
2961         * Animometer/tests/resources/algorithm.js: Removed. Heap is in utilities.js.
2962         * Animometer/tests/resources/sampler.js: Removed. Experiment is in statistics.js,
2963         Sampler in main.js.
2964         * Animometer/tests/resources/main.js: Move Sampler here.
2965         * Animometer/resources/statistics.js: Added. Move Statistics and Experiment here.
2966         * Animometer/resources/extensions.js: Move Heap here. Attach static method to create
2967         a max or min heap to Heap, instead of a new Algorithm object.
2968
2969         Update JS files.
2970         * Animometer/developer.html:
2971         * Animometer/index.html:
2972         * Animometer/tests/bouncing-particles/bouncing-canvas-images.html:
2973         * Animometer/tests/bouncing-particles/bouncing-canvas-shapes.html:
2974         * Animometer/tests/bouncing-particles/bouncing-css-images.html:
2975         * Animometer/tests/bouncing-particles/bouncing-css-shapes.html:
2976         * Animometer/tests/bouncing-particles/bouncing-svg-images.html:
2977         * Animometer/tests/bouncing-particles/bouncing-svg-shapes.html:
2978         * Animometer/tests/master/canvas-stage.html:
2979         * Animometer/tests/master/focus.html:
2980         * Animometer/tests/master/image-data.html:
2981         * Animometer/tests/master/multiply.html:
2982         * Animometer/tests/master/particles.html:
2983         * Animometer/tests/misc/canvas-electrons.html:
2984         * Animometer/tests/misc/canvas-stars.html:
2985         * Animometer/tests/misc/compositing-transforms.html:
2986         * Animometer/tests/simple/simple-canvas-paths.html:
2987         * Animometer/tests/simple/tiled-canvas-image.html:
2988         * Animometer/tests/template/template-canvas.html:
2989         * Animometer/tests/template/template-css.html:
2990         * Animometer/tests/template/template-svg.html:
2991         * Animometer/tests/text/layering-text.html:
2992         * Animometer/tests/text/text-boxes.html:
2993
2994         Fix the cursor in the graph analysis when the min
2995         complexity is not 0.
2996
2997         * Animometer/resources/debug-runner/graph.js:
2998         (_addRegression):
2999         (createComplexityGraph):
3000
3001 2016-02-23  Geoffrey Garen  <ggaren@apple.com>
3002
3003         Fix some issues in MallocBench
3004         https://bugs.webkit.org/show_bug.cgi?id=154600
3005
3006         Reviewed by Oliver Hunt.
3007
3008         * MallocBench/MallocBench.xcodeproj/project.pbxproj: Make nimlang.ops
3009         a part of the build so that it copies to the right place for execution.
3010
3011         * MallocBench/MallocBench/Interpreter.cpp:
3012         (Interpreter::Interpreter): Don't try to open .ops files for writing
3013         because we're only going to read and we might need extra permissions to
3014         write.
3015
3016         (Interpreter::~Interpreter): Give more context when opening a file fails
3017         to help with debugging.
3018
3019         * MallocBench/MallocBench/stress.cpp:
3020         (benchmark_stress): Reduce iterations to complete in less than 10 seconds.
3021
3022 2016-02-23  Michael Saboff  <msaboff@apple.com>
3023
3024         Unreviewed change to revert extraneous changes made part of change set 196955.
3025
3026         * MallocBench/MallocBench/Interpreter.cpp:
3027         (Interpreter::doMallocOp):
3028         (Interpreter::Thread::switchTo):
3029         (writeData): Deleted.
3030
3031 2016-02-19  Michael Saboff  <msaboff@apple.com>
3032
3033         MallocBench: Added recording for nimlang website, new recording details and added new options
3034         https://bugs.webkit.org/show_bug.cgi?id=154485
3035
3036         Reviewed by Geoff Garen.
3037
3038         Added new capabilities to MallocBench.  These include:
3039             Added a recording of http://nim-lang.org/docs/lib.html.
3040             Added thread id to the recording and the ability to playback switching threads in MallocBench
3041             Added aligned allocations to recordings and the ability to playback
3042             Added --use-thread-id option to honor recorded thread ids
3043             Added --detailed-report to output remaining allocations by size after playback
3044             Added --no-warmup to not run the warm up iteration
3045
3046         Changed the way that options are passed down to the benchmarks.  Instead of passing individual
3047         boolean or numeric option values, just pass a reference the CommandLine itself.  Each benchmark
3048         can access the options that are appropriate.  The Benchmark class also uses the options for
3049         is parallel, run counts and warm up.
3050
3051         Added thread id and aligned malloc to the Op by noticing that structure padding and Opcode allowed
3052         for another 32 bits of data.  Breaking that unused 32 bits into a 16 bit thread id value and a
3053         16 bit log base 2 of the alignment for aligned malloc allowed for existing recordings to playback
3054         without any incompatibilities.
3055
3056         Threaded operation is simulated by creating threads as needed.  As long as the next Op's thread id
3057         is the same as the last, operation continues as normal.  When the next Op has a different thread id,
3058         we switch to that thread using the shared Op stream to continue playing back.  There is a mutex to
3059         assure that only one thread is really running at a time and a condition variable used to wait
3060         that the current thread id matches each block thread's thread id.  This doesn't simulate true
3061         concurrent threading, but is instead plays back Ops recorded for multiple thread faithfully.
3062
3063         * MallocBench/MallocBench.xcodeproj/project.pbxproj:
3064         * MallocBench/MallocBench/Benchmark.cpp:
3065         (deallocateHeap):
3066         (Benchmark::Benchmark):
3067         (Benchmark::runOnce):
3068         (Benchmark::run):
3069         * MallocBench/MallocBench/Benchmark.h:
3070         (Benchmark::isValid):
3071         * MallocBench/MallocBench/CommandLine.cpp:
3072         (CommandLine::printUsage):
3073         * MallocBench/MallocBench/CommandLine.h:
3074         (CommandLine::isValid):
3075         (CommandLine::benchmarkName):
3076         (CommandLine::isParallel):
3077         (CommandLine::useThreadID):
3078         (CommandLine::detailedReport):
3079         (CommandLine::warmUp):
3080         (CommandLine::heapSize):
3081         (CommandLine::runs):
3082         * MallocBench/MallocBench/Interpreter.cpp:
3083         (Interpreter::Interpreter):
3084         (Interpreter::run):
3085         (Interpreter::readOps):
3086         (Interpreter::doOnSameThread):
3087         (Interpreter::switchToThread):
3088         (Interpreter::detailedReport):
3089         (compute2toPower):
3090         (writeData):
3091         (Interpreter::doMallocOp):
3092         (Interpreter::Thread::Thread):
3093         (Interpreter::Thread::stop):
3094         (Interpreter::Thread::~Thread):
3095         (Interpreter::Thread::runThread):
3096         (Interpreter::Thread::waitToRun):
3097         (Interpreter::Thread::switchTo):
3098         * MallocBench/MallocBench/Interpreter.h:
3099         (Interpreter::Thread::isMainThread):
3100         * MallocBench/MallocBench/alloc_free.cpp: Added.
3101         (benchmark_alloc_free):
3102         * MallocBench/MallocBench/alloc_free.h: Added.
3103         * MallocBench/MallocBench/balloon.cpp:
3104         (benchmark_balloon):
3105         * MallocBench/MallocBench/balloon.h:
3106         * MallocBench/MallocBench/big.cpp:
3107         (benchmark_big):
3108         * MallocBench/MallocBench/big.h:
3109         * MallocBench/MallocBench/churn.cpp:
3110         (benchmark_churn):
3111         * MallocBench/MallocBench/churn.h:
3112         * MallocBench/MallocBench/facebook.cpp:
3113         (benchmark_facebook):
3114         * MallocBench/MallocBench/facebook.h:
3115         * MallocBench/MallocBench/flickr.cpp:
3116         (benchmark_flickr):
3117         (benchmark_flickr_memory_warning):
3118         * MallocBench/MallocBench/flickr.h:
3119         * MallocBench/MallocBench/fragment.cpp:
3120         (validate):
3121         (benchmark_fragment):
3122         (benchmark_fragment_iterate):
3123         * MallocBench/MallocBench/fragment.h:
3124         * MallocBench/MallocBench/list.cpp:
3125         (benchmark_list_allocate):
3126         (benchmark_list_traverse):
3127         * MallocBench/MallocBench/list.h:
3128         * MallocBench/MallocBench/main.cpp:
3129         (main):
3130         * MallocBench/MallocBench/medium.cpp:
3131         (benchmark_medium):
3132         * MallocBench/MallocBench/medium.h:
3133         * MallocBench/MallocBench/memalign.cpp:
3134         (test):
3135         (benchmark_memalign):
3136         * MallocBench/MallocBench/memalign.h:
3137         * MallocBench/MallocBench/message.cpp:
3138         (benchmark_message_one):
3139         (benchmark_message_many):
3140         * MallocBench/MallocBench/message.h:
3141         * MallocBench/MallocBench/nimlang.cpp: Added.
3142         (benchmark_nimlang):
3143         * MallocBench/MallocBench/nimlang.h: Added.
3144         * MallocBench/MallocBench/nimlang.ops: Added.
3145         * MallocBench/MallocBench/realloc.cpp:
3146         (benchmark_realloc):
3147         * MallocBench/MallocBench/realloc.h:
3148         * MallocBench/MallocBench/reddit.cpp:
3149         (benchmark_reddit):
3150         (benchmark_reddit_memory_warning):
3151         * MallocBench/MallocBench/reddit.h:
3152         * MallocBench/MallocBench/stress.cpp:
3153         (deallocate):
3154         (benchmark_stress):
3155         * MallocBench/MallocBench/stress.h:
3156         * MallocBench/MallocBench/stress_aligned.cpp:
3157         (benchmark_stress_aligned):
3158         * MallocBench/MallocBench/stress_aligned.h:
3159         * MallocBench/MallocBench/theverge.cpp:
3160         (benchmark_theverge):
3161         (benchmark_theverge_memory_warning):
3162         * MallocBench/MallocBench/theverge.h:
3163         * MallocBench/MallocBench/tree.cpp:
3164         (benchmark_tree_allocate):
3165         (benchmark_tree_traverse):
3166         (benchmark_tree_churn):
3167         * MallocBench/MallocBench/tree.h:
3168         * MallocBench/run-malloc-benchmarks:
3169
3170 2016-02-11  Jon Lee  <jonlee@apple.com>
3171
3172         Fix a missing refactoring.
3173
3174         * Animometer/tests/master/resources/multiply.js:
3175         (animate): Move to Utilities.lerp.
3176         (_lerp): Deleted.
3177
3178 2016-02-10  Jon Lee  <jonlee@apple.com>
3179
3180         Add new benchmark tests.
3181         https://bugs.webkit.org/show_bug.cgi?id=154063
3182
3183         Provisionally reviewed by Said Abou-Hallawa.
3184
3185         Add tests for get/put image data, filters, opacity, and css transforms.
3186
3187         * Animometer/resources/runner/benchmark-runner.js:
3188         (_runBenchmarkAndRecordResults): Update the body background color to match that of
3189         the stage.
3190         (this._runNextIteration): Clear the background color style for the results page.
3191         * Animometer/resources/runner/tests.js:
3192         * Animometer/tests/master/focus.html: Added.
3193         * Animometer/tests/master/image-data.html: Added.
3194         * Animometer/tests/master/multiply.html: Added.
3195         * Animometer/tests/master/resources/focus.js: Added.
3196         * Animometer/tests/master/resources/image-data.js: Added.
3197         * Animometer/tests/master/resources/multiply.js: Added.
3198         * Animometer/tests/master/resources/stage.css: Move common styles out.
3199         * Animometer/tests/resources/main.js: Update Stage.randomBool to use Math.random.
3200         Add Stage.randomSign for randomly setting a direction. Add the notion of the
3201         current timestamp of the test to Benchmark, since some animations cycle through
3202         colors and rely on an incremental counter like the time.
3203
3204 2016-02-09  Said Abou-Hallawa  <sabouhallawa@apple.com>
3205
3206         Add internal benchmark tests for CSS mix-blend-modes and filters
3207         https://bugs.webkit.org/show_bug.cgi?id=154058
3208
3209         Provisionally reviewed by Jon Lee.
3210
3211         * Animometer/resources/debug-runner/tests.js: Include the new tests in the
3212         "HTML suite" of the debug runner.
3213
3214         * Animometer/resources/extensions.js:
3215         (Utilities.browserPrefix):
3216         (Utilities.setElementPrefixedProperty): Utility functions to allow setting
3217         prefixed style properties.
3218
3219         * Animometer/tests/bouncing-particles/resources/bouncing-css-shapes.js:
3220         Set the mix-blend-mode and the filter to some random values if the options
3221         of the test requested that.
3222
3223         * Animometer/tests/bouncing-particles/resources/bouncing-particles.js:
3224         (parseShapeParameters): Parse the url options "blend" and "filter" and set
3225         the corresponding flags.
3226
3227         * Animometer/tests/resources/main.js:
3228         (randomStyleMixBlendMode):
3229         (randomStyleFilter): Return random mix-blend-mode and filter.
3230
3231 2016-02-08  Jon Lee  <jonlee@apple.com>
3232
3233         Add a ramp controller
3234         https://bugs.webkit.org/show_bug.cgi?id=154028
3235
3236         Provisionally reviewed by Said Abou-Hallawa.
3237
3238         Enhance the graph to include a complexity-fps graph, in addition
3239         to the time graph.
3240
3241         * Animometer/developer.html: Add a ramp option.
3242         * Animometer/resources/debug-runner/animometer.css: Update the style.
3243         * Animometer/resources/strings.js: Flatten the Strings.text constants.
3244         * Animometer/resources/debug-runner/animometer.js:
3245         (ResultsTable.call._addGraphButton): Refactor.
3246         (ResultsTable.call._addTest): Add regression data.
3247         (benchmarkController): Add a form that allows the user to switch between the two forms,
3248         Add a form that allows the user to toggle different data. Hide certain header columns
3249         depending on the selected controller.
3250         * Animometer/resources/debug-runner/graph.js: Add the complexity regressions.
3251         * Animometer/resources/debug-runner/tests.js: Add headers for the ramp results.
3252         * Animometer/resources/runner/animometer.js:
3253         (ResultsTable): If a header is disabled don't include them in _flattenedHeaders.
3254         * Animometer/tests/resources/main.js:
3255         (Controller): Allow options to specify the capacity for sample arrays.
3256         (Regression): A piecewise regression that tries to fit a slope and a flat profile.
3257         (_calculateRegression): Options can fix the slope and bias when calculating the minimal
3258         error. Sweep across the samples in time (which could be backward depending on the controller)
3259         and calculate the intersection point.
3260         (RampController): This controller assumes that the target frame rate is below
3261         58 FPS. It runs in two stages. The first stage quickly determines the order of
3262         magnitude of objects needed to stress the system by the setting the complexity
3263         to increasingly difficult tiers. Perform a series of ramps descending from a
3264         high-water mark of complexity. The complexity needed to reach the target frame
3265         length is done by performing a piecewise regression on each ramp, and track a
3266         running average of these values. For the next ramp, make that running average
3267         the center of the ramp. With a minimum complexity of 0, the high-water mark is
3268         twice that average. The score is based on the highest complexity that can
3269         reach 60 fps.
3270
3271 2016-02-08  Jon Lee  <jonlee@apple.com>
3272
3273         Address Said's comments on the benchmark, and do some clean up.
3274
3275         * Animometer/developer.html:
3276         * Animometer/resources/debug-runner/animometer.css: Add styles for averages.
3277         * Animometer/resources/debug-runner/animometer.js: Use the right
3278         Strings constants.
3279         * Animometer/resources/debug-runner/graph.js:
3280         (_addRegressionLine): Add missing code to draw the line and standard
3281         deviation highlight.
3282         (onGraphTypeChanged): Remove unneeded variables
3283         (onTimeGraphOptionsChanged):
3284         * Animometer/resources/runner/benchmark-runner.js:
3285         (_runBenchmarkAndRecordResults): Rename samplers to suiteResults and
3286         _suitesSamplers to _suitesResults.
3287         * Animometer/tests/resources/main.js:
3288         (results): Call processSamples().
3289         (update): Change sampling timestamp comparison.
3290         (_animateLoop): Move shouldStop call to before the update.
3291         * Animometer/tests/resources/sampler.js:
3292         (process): Rename to processSamples().
3293
3294 2016-02-07  Jon Lee  <jonlee@apple.com>
3295
3296         Teach Controller to measure intervals, and turn off the frame length estimator.
3297
3298         * Animometer/tests/resources/main.js: Default interval length is 100 ms.
3299         (start): Set the first interval.
3300         (_measureAndResetInterval): Reports the average frame length of the interval that just
3301         completed, and sets up the next interval.
3302         (update): If there is no length, then just use the estimator per frame, otherwise the
3303         estimator measures per interval. Add a didFinishInterval for subclasses to process
3304         prior to recording the sample. Update tune() to include whether an interval had
3305         finished.
3306         (StepController): Step controllers don't measure on an interval basis.
3307
3308 2016-02-07  Jon Lee  <jonlee@apple.com>
3309
3310         Minor refactoring. Rename Controller._estimator to Controller._frameLengthEstimator
3311         and switch the parameters for start(), update(), and tune(), so that the timestamp
3312         is first and stage is second.
3313
3314         * Animometer/tests/resources/main.js:
3315
3316 2016-02-07  Jon Lee  <jonlee@apple.com>
3317
3318         Move ResultsTable functionality not needed for release tests out.
3319         Move reporting of score and mean to selection of the time-based graph.
3320
3321         * Animometer/developer.html: Rename graph-options to time-graph-options.
3322         * Animometer/resources/debug-runner/animometer.js:
3323         (DeveloperResultsTable): Moved from runner/animometer.js. Switch from mean
3324         values to "average" objects which can hold stdev. Move graph button and
3325         calculation of noisy measurements here. Sophisticated header processing
3326         is not needed in release suite.
3327         (populateTable): Use DeveloperResultsTable.
3328         * Animometer/resources/debug-runner/graph.js: Pull time graph creation to
3329         its own function, and add a new onGraphTypeChanged handler in preparation
3330         of a complexity graph to be added later.
3331         * Animometer/resources/runner/animometer.js:
3332         (ResultsTable): Simplify to just handle test names and scores.
3333
3334 2016-02-07  Jon Lee  <jonlee@apple.com>
3335
3336         Tests: reuse objects already made.
3337
3338         Avoid thrash of object creation and removal by maintaining an index that
3339         moves along the array as the adjust values change. If the tune value
3340         requires more objects than the maximum size of the object array, then create
3341         new objects. This means that the object array size never decreases.
3342
3343         * Animometer/tests/master/resources/canvas-stage.js: Maintain a separate
3344         offsetIndex. For these tests, we want to avoid drawing the oldest objects,
3345         so the scene will draw the object at offsetIndex to the end of the array.
3346         (tune): Reverse the logic since "removal" of objects is much simpler and
3347         involves simply changing the offsetIndex.
3348         (animate): Update the for loop to draw from offsetIndex to the end.
3349         (complexity): Update the definition.
3350         * Animometer/tests/master/resources/canvas-tests.js: Maintain a separate
3351         offsetIndex. For these tests, we want to avoid drawing the newest objects,
3352         so the scene will draw the object at index 0 to the object at offsetIndex.
3353         (SimpleCanvasStage.animate): Fly-by removal of local stage variable,
3354         which is unneeded. Update the for loop to draw from offsetIndex to the end.
3355         * Animometer/tests/simple/resources/simple-canvas-paths.js:
3356         (SimpleCanvasStage.animate): Update the for loop to draw from 0 to
3357         offsetIndex.
3358         * Animometer/tests/simple/resources/simple-canvas.js:
3359         (tune): Update logic. Here, offsetIndex represents the boundary of the last
3360         index to render.
3361         (animate): Update the for loop to draw from 0 to offsetIndex.
3362         (complexity): Update the definition.
3363
3364 2016-02-07  Jon Lee  <jonlee@apple.com>
3365
3366         Tests: refactor and update styles.
3367
3368         * Animometer/tests/resources/main.js: Add helper methods that return
3369         a color that hue rotates based on the date, and a counter value that
3370         increases based on the date. Fix randomInt() to not bias against the min and
3371         max values.
3372
3373         * Animometer/tests/master/resources/canvas-tests.js: Use new helper methods.
3374         * Animometer/tests/master/resources/dom-particles.js: Ditto.
3375         * Animometer/tests/master/resources/particles.js: Ditto.
3376         * Animometer/tests/simple/resources/simple-canvas-paths.js: Refactor to
3377         use a rotating color instead of a random color. The fast switching of color
3378         is too vivid to watch.
3379
3380         * Animometer/tests/bouncing-particles/resources/bouncing-svg-shapes.js:
3381         (BouncingSvgParticlesStage.call.createGradient): Fix the gradient so
3382         that the last stop is located at the end.
3383
3384 2016-02-07  Jon Lee  <jonlee@apple.com>
3385
3386         Refactor tune() to not return the complexity of the scene.
3387
3388         We have stage.complexity() now, so returning the complexity through tune
3389         is unnecessary.
3390
3391         * Animometer/tests/bouncing-particles/resources/bouncing-particles.js:
3392         * Animometer/tests/master/resources/canvas-stage.js:
3393         * Animometer/tests/master/resources/particles.js:
3394         * Animometer/tests/misc/resources/canvas-electrons.js:
3395         * Animometer/tests/misc/resources/canvas-stars.js:
3396         * Animometer/tests/resources/main.js:
3397         * Animometer/tests/simple/resources/simple-canvas.js:
3398         * Animometer/tests/simple/resources/tiled-canvas-image.js:
3399         * Animometer/tests/template/resources/template-canvas.js:
3400         * Animometer/tests/template/resources/template-css.js:
3401         * Animometer/tests/template/resources/template-svg.js:
3402         * Animometer/tests/text/resources/layering-text.js:
3403
3404 2016-02-07  Jon Lee  <jonlee@apple.com>
3405
3406         Make the fixed controller a step controller instead. Halfway through the test
3407         it will bump up the complexity 4-fold. Calculate the step timestamp using options
3408         instead of a separate parameter to the Controller constructor.
3409
3410         * Animometer/developer.html: Change value to "step"
3411         * Animometer/resources/debug-runner/animometer.js:
3412         (window.suitesManager.updateEditsElementsState): Show number inputs when set to "step".
3413         * Animometer/tests/resources/main.js:
3414         (update): Provide a hook for subclasses to tune.
3415         (StepController): Maintain a flag determining whether we've stepped, and the time
3416         we should step.
3417         (Benchmark): Use the new StepController.
3418
3419 2016-02-07  Jon Lee  <jonlee@apple.com>
3420
3421         Adjust the FPS graph scale.
3422
3423         Instead of making the FPS graph linearly scale, scale it based on the frame length,
3424         but show the data in terms of FPS. Because it is inversely proportional, and most
3425         of the data never gets below 20, concentrate the axis from 20-60 FPS, since otherwise
3426         over half of the available graph space ends up blank.
3427
3428         This means we should convert all of the FPS data to frame length data.
3429
3430         * Animometer/resources/debug-runner/graph.js: Update the domain to be based on
3431         frame length in milliseconds instead of FPS. Update the cursor to consider all of the
3432         values being shown, and then pick the min and max values to represent the length of the
3433         cursor.
3434         * Animometer/resources/runner/animometer.js:
3435         * Animometer/resources/strings.js:
3436         * Animometer/tests/resources/main.js:
3437         (processSamples): Add the ability to only sample a range of the data instead of everything
3438         after an offset index. Update sampler to record the frame lengths instead of the frame
3439         rate.
3440
3441 2016-02-07  Jon Lee  <jonlee@apple.com>
3442
3443         Add option to use different methods for retrieving a timestamp.
3444
3445         * Animometer/developer.html: Add performance.now and Date.now options.
3446         * Animometer/resources/runner/animometer.js: Default to performance.now.
3447         (window.benchmarkController.startBenchmark):
3448         * Animometer/tests/resources/main.js: Tie the desired method to _getTimestamp.
3449         (run): Use _getTimestamp.
3450         (_animateLoop): Ditto.
3451
3452 2016-02-07  Jon Lee  <jonlee@apple.com>
3453
3454         Allow adding any number of markers to the graph. The markers can be labeled
3455         and contain timestamp and sample index data. Make it a part of the controller
3456         rather than keeping it in the sampler.
3457
3458         * Animometer/resources/debug-runner/animometer.css: Add styles for markers
3459         * Animometer/resources/debug-runner/graph.js: Create the markers and add
3460         text labels.
3461         * Animometer/resources/runner/animometer.js: Assume the samplingTimeOffset
3462         is just one of the marks provided.
3463         * Animometer/resources/strings.js: Add Strings.json.marks.
3464         * Animometer/tests/resources/main.js:
3465         (Controller): Keep marks here. They are keyed by the marker name, so no two
3466         markers should have the same name.
3467         (recordFirstSample): Refactor to use mark.
3468         (mark): Allows for arbitrary data if needed later. The timestamp maintained
3469         is relative to the absolute start timestamp.
3470         (containsMark): Checks whether a mark with a specific comment exists.
3471         (processSamples): Removes the _startTimestamp offset from the marks before
3472         setting it in results.
3473         * Animometer/tests/resources/sampler.js: Remove marks.
3474
3475 2016-02-07  Jon Lee  <jonlee@apple.com>
3476
3477         Get rid of options member variable in Benchmark.
3478
3479         Options are only needed when initializing the stage or benchmark, so there's no
3480         need to also keep a reference to it.
3481
3482         * Animometer/tests/resources/main.js: Get rid of options variable in Benchmark.
3483         Pass options to Controllers and Stages.
3484         (Controller.Utilities.createClass):
3485         (Benchmark.Utilities.createClass):
3486         (get options): Deleted.
3487
3488         * Animometer/tests/bouncing-particles/resources/bouncing-canvas-images.js:
3489         * Animometer/tests/bouncing-particles/resources/bouncing-canvas-particles.js:
3490         * Animometer/tests/bouncing-particles/resources/bouncing-canvas-shapes.js:
3491         * Animometer/tests/bouncing-particles/resources/bouncing-css-images.js:
3492         * Animometer/tests/bouncing-particles/resources/bouncing-css-shapes.js:
3493         * Animometer/tests/bouncing-particles/resources/bouncing-particles.js:
3494         * Animometer/tests/bouncing-particles/resources/bouncing-svg-images.js:
3495         * Animometer/tests/bouncing-particles/resources/bouncing-svg-shapes.js:
3496         * Animometer/tests/master/resources/canvas-stage.js:
3497         * Animometer/tests/master/resources/canvas-tests.js:
3498         * Animometer/tests/master/resources/particles.js:
3499         * Animometer/tests/misc/resources/canvas-electrons.js:
3500         * Animometer/tests/misc/resources/canvas-stars.js:
3501         * Animometer/tests/misc/resources/compositing-transforms.js:
3502         * Animometer/tests/simple/resources/simple-canvas-paths.js:
3503         * Animometer/tests/simple/resources/tiled-canvas-image.js:
3504         * Animometer/tests/template/resources/template-canvas.js:
3505         * Animometer/tests/template/resources/template-css.js:
3506         * Animometer/tests/template/resources/template-svg.js:
3507         * Animometer/tests/text/resources/layering-text.js:
3508
3509 2016-02-07  Jon Lee  <jonlee@apple.com>
3510
3511         Update how the benchmark is run
3512         https://bugs.webkit.org/show_bug.cgi?id=153960
3513
3514         Provisionally reviewed by Said Abou-Hallawa.
3515
3516         Introduce the notion of a Controller. It is responsible for recording, updating,
3517         and processing the statistics and complexity of the benchmark. This allows
3518         plugging in different Controllers.
3519
3520         This strips most of the functionality from Animator and BenchmarkState, so fold
3521         what's left into Benchmark. Now, Benchmarks only own a stage and a controller, but
3522         are responsible for driving the animation loop.
3523
3524         Rewrite Animator._shouldRequestAnotherFrame into two different Controllers. One
3525         maintains a fixed complexity, and the other adapts the complexity to meet a
3526         fixed FPS.
3527
3528         Fix the Kalman estimator to be modeled on a scalar variable with no model.
3529
3530         * Animometer/tests/resources/main.js: Remove BenchmarkState and Animator, and
3531         replace it with a Controller. Add a FixedController and refactor the previous controller
3532         to an AdaptiveController.
3533
3534         (Controller): Controllers own the estimator and the sampler. When a new frame is
3535         displayed, the animation loop calls update(). The estimator and sampler record
3536         stats, then tune. Samplers can track multiple series of data. The basic controller
3537         tracks timestamp, complexity, and estimated frame rate.
3538                 The Kalman estimation is based on the frame length rather than the frame
3539         rate. Because FPS is inverse proportional to frame length, in the case where the measured
3540         frame length is very small, the FPS ends up being a wildly large number (in the order of
3541         600-1000 "FPS"), and it pulls the estimator up drastically enough that it takes a while
3542         for it to settle back down. Using frame length reduces the impact of these spikes.
3543                 Converging the estimation takes enough time to avoid initializing it immediately
3544         when the benchmark starts. Instead, the benchmark runs for a brief period of time (100ms)
3545         before running it in earnest. Allow controllers an opportunity to set the complexity
3546         before starting recording.
3547                 When the benchmark is complete, the controller has an opportunity to process
3548         the samples. The default implementation calculates the raw FPS based on the time
3549         difference of the samples, and calculates the complexity score. This is moved from
3550         Benchmark.processSamples.
3551
3552         (Controller): Initialize timestamps. These are at first relative to the start of the
3553         benchmark, but are offset by the absolute start time during start(). By default maintain
3554         3 data series, but subclasses can override.
3555         (start): Calls recordFirstSample() for subclasses to override if needed.
3556         (recordFirstSample): For basic controller, start sampling at the beginning.
3557         (update): Update the frame length estimator and sample.
3558         (shouldStop): Checks that the time is before _endTimestamp.
3559         (results): Returns the processed samples.
3560         (processSamples): Iterate through the sample data and collate them. Include scores.
3561
3562         (FixedComplexityController): Controller that tunes the stage to the desired complexity
3563         prior to starting, and keeps it at that complexity.
3564
3565         (AdaptiveController): Have the estimator estimate the interval frame rate instead of the
3566         raw frame rate.
3567                 The previous version of this controller ignored the frame that came after the
3568         adjustment. The raw FPS show that whatever noise the scene change adds is negligible
3569         compared to the noise of the system overall. Stop ignoring that frame and include all
3570         frames in the measurements.
3571
3572         (Benchmark): Remove dependency on animator, and instantiate a runner based on what is
3573         selected. Most of the loop's functionality is in Controller, so remove here.
3574         (Benchmark.run): Remove start() since it is only called from run(), and fold it in here.
3575         (Benchmark._animateLoop): Fold in from Animator.animateLoop. Let the benchmark run for
3576         a brief period before calling Controller.start().
3577
3578         * Animometer/tests/resources/math.js: Fix the Kalman estimator. The filter estimates
3579         a scalar variable, and makes basic assumptions regarding the model. As a result
3580         none of the linear algebra classes are needed, so remove Matrix, Vector3, and Matrix3.
3581         (SimpleKalmanEstimator): Calculate the gain based on the provided process and
3582         measurement errors.
3583         (KalmanEstimator): Deleted.
3584         (IdentityEstimator): Deleted.
3585         (PIDController): Refactor to use the Utilities.createClass() helper.
3586
3587         The Kalman filter algorithm is explained here http://greg.czerniak.info/guides/kalman1/.
3588         The state, represented by a scalar, is the estimated frame length. There is no user
3589         transition of the state, and the state is the same as the measurement. With this model,
3590         the estimation error converges, so calculate the gain ahead of time.
3591
3592         * Animometer/developer.html: Remove fixed-after-warmup since it is not useful.
3593         Replace the option to toggle the estimator, and make it possible to customize the
3594         estimator's error parameters. Show raw FPS by default, and remove interval FPS,
3595         which will be shown instead of the filtered raw FPS.
3596         * Animometer/resources/debug-runner/animometer.css: Put the header behind the graph.
3597         Remove #intervalFPS rules; move the color to #filteredFPS.
3598         * Animometer/resources/debug-runner/graph.js:
3599         (updateGraphData): Update the hr style to force the layout to be calculated
3600         correctly. Change the tick format to be in terms of seconds, since the timestamps
3601         are in milliseconds. Remove interval data.
3602         * Animometer/resources/runner/animometer.js:
3603         (window.benchmarkController.startBenchmark): Set Kalman parameters.
3604         * Animometer/resources/runner/benchmark-runner.js:
3605         (_runBenchmarkAndRecordResults): When a benchmark completes, expect it to return
3606         the final data, rather than passing a sampler from the controller. This avoids
3607         needing to expose the sampler variable in the benchmark.
3608         * Animometer/tests/resources/sampler.js:
3609         (process): Move the setting of the target frame rate to AdaptiveController.
3610
3611 2016-02-06  Jon Lee  <jonlee@apple.com>
3612
3613         Code clean up: Move Rotater function closer to Stage static methods.
3614         The Rotater is used together with those methods; keep them close.
3615
3616         * Animometer/tests/resources/main.js:
3617
3618 2016-02-06  Jon Lee  <jonlee@apple.com>
3619
3620         Update the JS includes due to ResultsTable move.
3621
3622         * Animometer/developer.html:
3623         * Animometer/index.html:
3624
3625 2016-02-06  Jon Lee  <jonlee@apple.com>
3626
3627         Move createElement and createSVGElement to Utilities.
3628
3629         * Animometer/resources/extensions.js:
3630         (Utilities.createElement): Added.
3631         (Utilities.createSVGElement): Added.
3632         (DocumentExtension.createElement): Deleted.
3633         (DocumentExtension.createSvgElement): Deleted.
3634
3635         * Animometer/resources/debug-runner/animometer.js:
3636         * Animometer/resources/runner/animometer.js:
3637         * Animometer/tests/bouncing-particles/resources/bouncing-svg-images.js:
3638         * Animometer/tests/bouncing-particles/resources/bouncing-svg-particles.js:
3639         * Animometer/tests/bouncing-particles/resources/bouncing-svg-shapes.js:
3640
3641 2016-02-06  Jon Lee  <jonlee@apple.com>
3642
3643         Add a convenience function for creating a class.
3644
3645         The pattern for creating a class is common enough to add as a Utilities
3646         helper function. It also makes it easy to collapse class definitions when
3647         editing.
3648
3649         * Animometer/resources/debug-runner/animometer.js: Move ProgressBar definition,
3650         since it is only used here.
3651         * Animometer/resources/runner/animometer.js: Move ResultsDashboard and
3652         ResultsTable definition, since it is only used here.
3653         * Animometer/resources/extensions.js: Move Utilities definition to the top. Convert
3654         Point, Insets, SimplePromise.
3655         (ProgressBar): Moved to animometer.js.
3656         (ResultsDashboard): Moved to animometer.js.
3657         (ResultsTable): Moved to animometer.js.
3658         * Animometer/resources/runner/benchmark-runner.js: Convert BenchmarkRunnerState,
3659         BenchmarkRunner.
3660         * Animometer/tests/resources/main.js: Convert Rotater, Stage, Animator, Benchmark.
3661         * Animometer/tests/resources/sampler.js: Convert Experiment, Sampler.
3662
3663         Convert test primitives.
3664         * Animometer/tests/master/resources/canvas-tests.js: Convert CanvasLineSegment,
3665         CanvasArc, CanvasLinePoint.
3666         * Animometer/tests/simple/resources/simple-canvas-paths.js: Convert CanvasLineSegment,
3667         CanvasLinePoint, CanvasQuadraticSegment, CanvasQuadraticPoint, CanvasBezierSegment,
3668         CanvasBezierPoint, CanvasArcToSegment, CanvasArcToSegmentFill, CanvasArcSegment,
3669         CanvasArcSegmentFill, CanvasRect, CanvasRectFill.
3670         * Animometer/tests/simple/resources/tiled-canvas-image.js: Convert CanvasImageTile.
3671
3672 2016-02-06  Jon Lee  <jonlee@apple.com>
3673
3674         Minor improvements to debug harness.
3675
3676         * Animometer/developer.html:
3677         * Animometer/resources/debug-runner/animometer.css:
3678         (#suites): Put the complexity text boxes closer to the test names.
3679         (#options):
3680         (#rawFPS circle): Make the interval FPS appear as a separate data series, with a line.
3681         (#intervalFPS path):
3682         (#intervalFPS circle):
3683         * Animometer/resources/debug-runner/animometer.js:
3684         (window.optionsManager.updateLocalStorageFromUI): Convert number inputs from text.
3685         (window.suitesManager._onChangeTestCheckbox): Refactor to take a checkbox.
3686         (window.suitesManager._createTestElement): Enhance such that typing into the complexity
3687         input will automatically select that test for running.
3688         (window.suitesManager.updateLocalStorageFromJSON): Make the harness work for private
3689         browsing.
3690         * Animometer/resources/debug-runner/graph.js: Separate the intervalFPS data, and show
3691         more accuracy in timestamps.
3692
3693 2016-02-06  Jon Lee  <jonlee@apple.com>
3694
3695         Refactor helper methods for getting random values for a stage.
3696
3697         Instead of requiring a Stage instance, just attach it to the Stage object.
3698
3699         * Animometer/tests/bouncing-particles/resources/bouncing-canvas-shapes.js:
3700         * Animometer/tests/bouncing-particles/resources/bouncing-css-shapes.js:
3701         * Animometer/tests/bouncing-particles/resources/bouncing-particles.js:
3702         * Animometer/tests/bouncing-particles/resources/bouncing-svg-shapes.js:
3703         * Animometer/tests/master/resources/canvas-tests.js:
3704         * Animometer/tests/master/resources/particles.js:
3705         * Animometer/tests/misc/resources/canvas-electrons.js:
3706         * Animometer/tests/misc/resources/canvas-stars.js:
3707         * Animometer/tests/misc/resources/compositing-transforms.js:
3708         * Animometer/tests/resources/main.js:
3709         * Animometer/tests/simple/resources/simple-canvas-paths.js:
3710
3711 2016-02-05  Said Abou-Hallawa  <sabouhallawa@apple.com>
3712
3713         Add a new graphics test for CanvasRenderingContext2D functions: getImageData and putImageData
3714         https://bugs.webkit.org/show_bug.cgi?id=151716
3715
3716         Reviewed by Darin Adler.
3717
3718         The purpose of this test is to measure the performance of getImageData
3719         and putImageData functions. This test draws a background on the canvas
3720         and then gets some random tiles from this background and draw them in
3721         destinations different from their original sources.
3722
3723         * Animometer/resources/debug-runner/tests.js: Adding the new test to the canvas simple tests suite.
3724
3725         * Animometer/resources/extensions.js:
3726         (Array.prototype.shuffle): Shuffles the elements of an array.
3727
3728         (Point.zero): Returns a new Point object whose x and y are equal zero.
3729         (Point.prototype.str): Used for debugging the Point object.
3730
3731         * Animometer/tests/simple/resources/tiled-canvas-image.js: Added.
3732         (CanvasImageTile):
3733         (CanvasImageTile.prototype.getImageData):
3734         (CanvasImageTile.prototype.putImageData):
3735         (Stage.call.initialize):
3736         (Stage.call._createTiles):
3737         (Stage.call._nextTilePosition):
3738         (Stage.call.tune):
3739         (Stage.call._drawBackground):
3740         (Stage.call.animate):
3741         (Stage.call.complexity):
3742         (Stage.call):
3743         * Animometer/tests/simple/tiled-canvas-image.html: Added.
3744
3745 2016-01-07  Jon Lee  <jonlee@apple.com>
3746
3747         Fix new test.
3748
3749         * Animometer/resources/runner/tests.js: Wrong URL from an
3750         old patch.
3751         * Animometer/tests/master/particles.html:
3752         * Animometer/tests/master/resources/dom-particles.js:
3753         (Particle.call.reset): Figured out a simpler way to set up
3754         the particles.
3755         (this.move.reset): Deleted.
3756         (this.move._applyAttributes): Deleted.
3757         * Animometer/tests/master/resources/particles.js:
3758         (Particle): Call move() after reset().
3759
3760 2016-01-07  Jon Lee  <jonlee@apple.com>
3761
3762         Update benchmark test suite
3763         https://bugs.webkit.org/show_bug.cgi?id=152679
3764
3765         Reviewed by Simon Fraser.
3766
3767         Add a new test. The test has a rotating background
3768         gradient, and does a better job physically simulating
3769         particles.
3770
3771         * Animometer/resources/extensions.js: Teach Point to take constants as well as other Points.
3772         (Point.prototype.length): Added.
3773         (Point.prototype.normalize): Added.
3774         * Animometer/resources/runner/tests.js: Add the test to the master suite.
3775         * Animometer/tests/master/particles.html: Added.
3776         * Animometer/tests/master/resources/particles.js: Added. Parent class for different kinds of particles.
3777         (Particle):
3778         (Particle.prototype.reset): If the particle starts slowing down in terms of its animation,
3779         reset it.
3780         (Particle.prototype.animate): Bounce off the walls elastically, and include gravity.
3781         (Particle.prototype.move): Subclasses should override.
3782         (ParticlesStage): Stage includes a rotating gradient background.
3783         * Animometer/tests/master/resources/dom-particles.js: Added. Creates a <div> and adds it to
3784         the stage.
3785         * Animometer/tests/resources/star.svg: Added.
3786
3787 2016-01-03  Jon Lee  <jonlee@apple.com>
3788
3789         Update benchmark test suite
3790         https://bugs.webkit.org/show_bug.cgi?id=152679
3791
3792         Reviewed by Simon Fraser.
3793
3794         Move algorithm.js and sampler.js to tests/ and benchmark-runner.js to runner/.
3795
3796         Needed by both harnesses.
3797         * Animometer/resources/runner/benchmark-runner.js: Renamed from PerformanceTests/Animometer/resources/debug-runner/benchmark-runner.js.
3798         * Animometer/developer.html:
3799         * Animometer/index.html:
3800
3801         Needed only by the tests. Move to tests/. Statistics, in sampler.js, is used by ResultsDashboard, so move that
3802         into extensions.js.
3803         * Animometer/resources/extensions.js:
3804         * Animometer/tests/resources/algorithm.js: Renamed from PerformanceTests/Animometer/resources/algorithm.js.
3805         * Animometer/tests/resources/sampler.js: Renamed from PerformanceTests/Animometer/resources/sampler.js.
3806         * Animometer/tests/bouncing-particles/bouncing-canvas-images.html:
3807         * Animometer/tests/bouncing-particles/bouncing-canvas-shapes.html:
3808         * Animometer/tests/bouncing-particles/bouncing-css-images.html:
3809         * Animometer/tests/bouncing-particles/bouncing-css-shapes.html:
3810         * Animometer/tests/bouncing-particles/bouncing-svg-images.html:
3811         * Animometer/tests/bouncing-particles/bouncing-svg-shapes.html:
3812         * Animometer/tests/master/canvas-stage.html:
3813         * Animometer/tests/misc/canvas-electrons.html:
3814         * Animometer/tests/misc/canvas-stars.html:
3815         * Animometer/tests/misc/compositing-transforms.html:
3816         * Animometer/tests/simple/simple-canvas-paths.html:
3817         * Animometer/tests/template/template-canvas.html:
3818         * Animometer/tests/template/template-css.html:
3819         * Animometer/tests/template/template-svg.html:
3820         * Animometer/tests/text/layering-text.html:
3821         * Animometer/tests/text/text-boxes.html:
3822
3823 2016-01-07  Jon Lee  <jonlee@apple.com>
3824
3825         Update benchmark test suite