Memory benchmark for showing large images in succession #64762
Conversation
flutter-dashboard
bot
added
the
c: contributor-productivity
|
It looks like this pull request may not have tests. Please make sure to add tests before merging. If you need an exemption to this rule, contact Hixie on the #hackers channel in Chat. Reviewers: Read the Tree Hygiene page and make sure this patch meets those guidelines before LGTMing. |
| imageIndex = 0; | ||
| } else { | ||
| imageIndex += 1; | ||
| } |
There was a problem hiding this comment.
Nit: line 33-37 can be simplified as imageIndex = (imageIndex + 1) % 6.
| const String kPackageName = 'com.example.macrobenchmarks'; | ||
| const String kActivityName = 'com.example.macrobenchmarks.MainActivity'; | ||
|
|
||
| class LargeImageChangerPerfTest extends MemoryTest { |
There was a problem hiding this comment.
I wonder if more memory info (e.g., peak memory, average, percentile, etc.) from DevToolsMemoryTest (see, e.g., this example) or Timeline (see, e.g., profiling_summarizer and flutter/engine#18516) would be helpful here? The DevTools works for Android and the Timeline works for iOS.
|
Updated to use a DevTools memory test. |
|
Is it feasible to measure CPU usage as well? |
|
@liyuqian might have more context on that. What CPU benchmarks do you think this would capture that we're not already capturing though? |
|
specifically the cost of increased GCs to reduce image memory usage |
…o image_page_benchmark
|
Added an iOS variant that does CPU as well. |
| await driver.close(); | ||
| }); | ||
|
|
||
| // test('Measure CPU/GPU/Memory', () async { |
| }, onDone: () { observatoryUri.complete(null); }); | ||
| }, onDone: () { | ||
| if (!observatoryUri.isCompleted) { | ||
| observatoryUri.complete(null); |
There was a problem hiding this comment.
unless you need the null, this can be observatoryUri.complete()
liyuqian
added
perf: energy
4 participants
Benchmark for #56482
This will be improved when the following finish landing:
Current behavior observed is memory growth until a GC is finally performed after multiple iterations of changing the image. Expected behavior will be that a GC is performed at least once after a large image is disposed of.