New issue
Advanced search Search tips

Issue 809141 link

Starred by 1 user

Issue metadata

Status: Assigned
Owner:
Cc:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: ----



Sign in to add a comment

cc_perftests failing on chromium.perf/Android Nexus6 Perf

Project Member Reported by sheriff-...@appspot.gserviceaccount.com, Feb 5 2018

Issue description

Filed by sheriff-o-matic@appspot.gserviceaccount.com on behalf of sullivan@google.com

cc_perftests failing on chromium.perf/Android Nexus6 Perf

Builders failed on: 
- Android Nexus6 Perf: 
  https://build.chromium.org/p/chromium.perf/builders/Android%20Nexus6%20Perf

Sample logs:
https://logs.chromium.org/v/?s=chrome%2Fbb%2Fchromium.perf%2FAndroid_Nexus6_Perf%2F1238%2F%2B%2Frecipes%2Fsteps%2Fcc_perftests_on_Android%2F0%2Fstdout

I  490.883s run_tests_on_device(ZX1G22FJB7)  [ RUN      ] CCSerializationPerfTest.DelegatedFrame_ManyRenderPasses_5_100
I  490.883s run_tests_on_device(ZX1G22FJB7)  *RESULT StructTraits serialization min_frame_serialization_time_per_render_pass_shared_quad_state: DelegatedFrame_ManyRenderPasses_5_100= 377.8 us
I  490.883s run_tests_on_device(ZX1G22FJB7)  *RESULT StructTraits serialization: num runs in 2 seconds_per_render_pass_shared_quad_state: DelegatedFrame_ManyRenderPasses_5_100= 4891 runs/s
I  490.884s run_tests_on_device(ZX1G22FJB7)  *RESULT StructTraits deserialization min_frame_deserialization_time_per_render_pass_shared_quad_state: DelegatedFrame_ManyRenderPasses_5_100= 190.1 us
I  490.884s run_tests_on_device(ZX1G22FJB7)  *RESULT StructTraits deserialization: num runs in 2 seconds_per_render_pass_shared_quad_state: DelegatedFrame_ManyRenderPasses_5_100= 10372 runs/s
I  490.884s run_tests_on_device(ZX1G22FJB7)  *RESULT ParamTraits serialization: min_frame_serialization_time_per_render_pass_shared_quad_state: DelegatedFrame_ManyRenderPasses_5_100= 273 us
I  490.884s run_tests_on_device(ZX1G22FJB7)  *RESULT ParamTraits serialization: num runs in 2 seconds_per_render_pass_shared_quad_state: DelegatedFrame_ManyRenderPasses_5_100= 7044 runs/s
I  490.884s run_tests_on_device(ZX1G22FJB7)  [ CRASHED      ]
I  490.886s TimeoutThread-1-for-run_tests_on_device(ZX1G22FJB7)  [host]> /b/swarming/w/ir/third_party/android_tools/sdk/platform-tools/adb -s ZX1G22FJB7 shell '( pm dump org.chromium.native_test | grep dataDir=; echo "PIPESTATUS: ${PIPESTATUS[@]}" );echo %$?'
...
I  511.025s run_tests_on_device(ZX1G22FJB7)  [ RUN      ] CCSerializationPerfTest.DelegatedFrame_ManyQuads_1_4000
I  511.025s run_tests_on_device(ZX1G22FJB7)  *RESULT StructTraits serialization min_frame_serialization_time_per_render_pass_shared_quad_state: DelegatedFrame_ManyQuads_1_4000= 2874.2000000000003 us
I  511.025s run_tests_on_device(ZX1G22FJB7)  *RESULT StructTraits serialization: num runs in 2 seconds_per_render_pass_shared_quad_state: DelegatedFrame_ManyQuads_1_4000= 618 runs/s
I  511.025s run_tests_on_device(ZX1G22FJB7)  *RESULT StructTraits deserialization min_frame_deserialization_time_per_render_pass_shared_quad_state: DelegatedFrame_ManyQuads_1_4000= 2862 us
I  511.025s run_tests_on_device(ZX1G22FJB7)  *RESULT StructTraits deserialization: num runs in 2 seconds_per_render_pass_shared_quad_state: DelegatedFrame_ManyQuads_1_4000= 670 runs/s
I  511.025s run_tests_on_device(ZX1G22FJB7)  *RESULT ParamTraits serialization: min_frame_serialization_time_per_render_pass_shared_quad_state: DelegatedFrame_ManyQuads_1_4000= 2067.5999999999995 us
I  511.025s run_tests_on_device(ZX1G22FJB7)  *RESULT ParamTraits serialization: num runs in 2 seconds_per_render_pass_shared_quad_state: DelegatedFrame_ManyQuads_1_4000= 917 runs/s
I  511.025s run_tests_on_device(ZX1G22FJB7)  [ CRASHED      ]
...

I'm going to kick off a bisect.
 
Failing on Nexus 7 too.
Failing on Nexus 5 too.
😿 Pinpoint job stopped with an error.
https://pinpoint-dot-chromeperf.appspot.com/job/11e73a3e840000
Owner: samans@chromium.org
samans: the pinpoint job in #4 didn't complete, but it has a failure range of r533863 - r533864. Since cc_perftests are failing and your CL r533863 ("viz: Disallow CompositorFrames with empty size") affects cc/, can you take a look?
(hit the rerun button in pinpoint to kick off the bisect in #6)
Project Member

Comment 8 by 42576172...@developer.gserviceaccount.com, Feb 12 2018

Cc: peter@chromium.org brat...@opera.com
Owner: brat...@opera.com
Status: Assigned (was: Available)
📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/14920acd840000

Avoid using the Windows macro PLATFORM_UNKNOWN by bratell@opera.com
https://chromium.googlesource.com/chromium/src/+/e3c10f34e94c0afa1a9cdbb1574085a94939aeca

Understanding performance regressions:
  http://g.co/ChromePerformanceRegressions
Cc: -peter@chromium.org dtu@chromium.org
Owner: samans@chromium.org
Sorry bratell and peter, not sure what happened with the bisect here. The test is failing at your CL, but the previous passing CL it tested was 600 CLs earlier, so it should have kept bisecting. +dtu why did the bisect in #8 blame that CL?

samans: I still think the failure is at your CL, can you take a look?

Comment 10 by dtu@chromium.org, Feb 13 2018

There's a bug with the re-run button where it's running the jobs as try jobs, i.e. it doesn't perform bisection. Sorry about that!

https://github.com/catapult-project/catapult/issues/4210#issuecomment-365110397

Sign in to add a comment