| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
page_test (patchset #4 id:60001 of https://codereview.chromium.org/1151933005/)
Reason for revert:
I suspect this is causing failures on telemetry_perf_unittests - see https://build.chromium.org/p/chromium.win/builders/XP%20Tests%20%281%29 build #37836 and later
Original issue's description:
> Make sure that tracing is always cleaned up for v8_gc_times page_test
>
> BUG=491104
> CQ_EXTRA_TRYBOTS=tryserver.chromium.perf:linux_perf_bisect;tryserver.chromium.perf:mac_perf_bisect;tryserver.chromium.perf:win_perf_bisect;tryserver.chromium.perf:android_nexus5_perf_bisect
>
> Committed: https://crrev.com/1d1d10c687de9424db5a66a89d179f11784d7ee8
> Cr-Commit-Position: refs/heads/master@{#331590}
TBR=sullivan@chromium.org,rmcilroy@chromium.org,prasadv@chromium.org,nednguyen@google.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=491104
Review URL: https://codereview.chromium.org/1147093008
Cr-Commit-Position: refs/heads/master@{#331669}
|
|
|
|
|
|
|
|
|
| |
BUG=491104
CQ_EXTRA_TRYBOTS=tryserver.chromium.perf:linux_perf_bisect;tryserver.chromium.perf:mac_perf_bisect;tryserver.chromium.perf:win_perf_bisect;tryserver.chromium.perf:android_nexus5_perf_bisect
Review URL: https://codereview.chromium.org/1151933005
Cr-Commit-Position: refs/heads/master@{#331590}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Patch 1 is from issue 1140353003 at patchset 60001 (http://crrev.com/1140353003#ps60001)
Patch 2 add a workaround to make sure that service_worker
benchmark create interation record.
BUG=444705
CQ_EXTRA_TRYBOTS=tryserver.chromium.perf:linux_perf_bisect;tryserver.chromium.perf:mac_perf_bisect;tryserver.chromium.perf:android_nexus5_perf_bisect
Review URL: https://codereview.chromium.org/1146263004
Cr-Commit-Position: refs/heads/master@{#331008}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
#3 id:60001 of https://codereview.chromium.org/1140353003/)
Reason for revert:
Seems to have broken service_worker tests (http://crbug.com/490130)
Original issue's description:
> [Telemetry] Disable auto issuing record for thread_times
>
> BUG=444705
> CQ_EXTRA_TRYBOTS=tryserver.chromium.perf:linux_perf_bisect;tryserver.chromium.perf:mac_perf_bisect;tryserver.chromium.perf:android_nexus5_perf_bisect
>
> Committed: https://crrev.com/2905d303db21f120aab1914f98fe0e44284f82ed
> Cr-Commit-Position: refs/heads/master@{#330613}
TBR=eakuefner@chromium.org,nednguyen@google.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=444705
Review URL: https://codereview.chromium.org/1151543002
Cr-Commit-Position: refs/heads/master@{#330728}
|
|
|
|
|
|
|
|
|
| |
BUG=444705
CQ_EXTRA_TRYBOTS=tryserver.chromium.perf:linux_perf_bisect;tryserver.chromium.perf:mac_perf_bisect;tryserver.chromium.perf:android_nexus5_perf_bisect
Review URL: https://codereview.chromium.org/1140353003
Cr-Commit-Position: refs/heads/master@{#330613}
|
|
|
|
|
|
|
|
|
|
|
| |
This can be merged to ValidateAndMeasurePage method.
BUG=455391, 470147
CQ_EXTRA_TRYBOTS=tryserver.chromium.perf:linux_perf_bisect;tryserver.chromium.perf:mac_perf_bisect;tryserver.chromium.perf:android_nexus5_perf_bisect
Review URL: https://codereview.chromium.org/1124033004
Cr-Commit-Position: refs/heads/master@{#329468}
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The current threadtime timeline measurement only supports per-frame
metrics. Generalize this to support total and per-second metrics for
threadtimes and tasks.
BUG=466213
CQ_EXTRA_TRYBOTS=tryserver.chromium.perf:mac_perf_bisect;tryserver.chromium.perf:android_nexus5_perf_bisect
Review URL: https://codereview.chromium.org/1084533005
Cr-Commit-Position: refs/heads/master@{#329322}
|
|
|
|
|
|
|
|
|
|
|
| |
Merge this with DidNavigateToPage since they are called and essentially called
in the same time (after page.Navigate & before page.RunPageInteractions)
BUG=455391, 470147
Review URL: https://codereview.chromium.org/1139823002
Cr-Commit-Position: refs/heads/master@{#329292}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When impl-side painting is on, the main thread will use property trees
to calculate visible rects and update lists. Because impl-side painting
is not shipped everywhere, this cannot be fully turned on everywhere.
This leaves testing in a bit of a weird state. Previously, using
the "verify property trees" setting would let us catch differences
between CDP and property trees. However, running CDP has side effects,
so to make sure that property trees on their own are valid, it
can't be run.
This means that any changes to CDP need to run in both states.
As CDP is in maintenance mode, this seems like the right balance.
Review URL: https://codereview.chromium.org/867273004
Cr-Commit-Position: refs/heads/master@{#329196}
|
|
|
|
|
|
|
|
|
|
|
|
| |
This CL reenables the superfluous-parens and unpacking-non-sequence Pylint
warnings for tools/telemetry and tools/perf
BUG=475714
R=nednguyen,dtu
Review URL: https://codereview.chromium.org/1137563003
Cr-Commit-Position: refs/heads/master@{#329134}
|
|
|
|
|
|
|
|
|
|
|
| |
almost the same to smoothness, except for customized browser option.
Remove smoothness_controller since it's no longer used.
BUG=444703
Review URL: https://codereview.chromium.org/1127413002
Cr-Commit-Position: refs/heads/master@{#329018}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The refactoring that modifies smoothness to use TimelineBasedMeasurement's
control code (https://codereview.chromium.org/1117953004)
has change the tracing categories used for smoothness slightly, which
may cause a small regression to queuing_durations metric.
This patch explictly specifies the tracing categories to be used for smoothness
instead of using the default no-overhead-level
(https://code.google.com/p/chromium/codesearch#chromium/src/tools/telemetry/telemetry/core/platform/tracing_category_filter.py&l=18).
BUG=484802, 444701
Review URL: https://codereview.chromium.org/1131883002
Cr-Commit-Position: refs/heads/master@{#328994}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
navigation.
This is the first step in merging repaint with smoothness measurement, allowing
both to use TBM control code.
BUG=444701
CQ_EXTRA_TRYBOTS=tryserver.chromium.perf:linux_perf_bisect;tryserver.chromium.perf:mac_perf_bisect;tryserver.chromium.perf:android_nexus5_perf_bisect
Review URL: https://codereview.chromium.org/1126183002
Cr-Commit-Position: refs/heads/master@{#328740}
|
|
|
|
|
|
|
|
| |
BUG=475714
Review URL: https://codereview.chromium.org/1126523002
Cr-Commit-Position: refs/heads/master@{#328519}
|
|
|
|
|
|
|
|
| |
BUG=483212
Review URL: https://codereview.chromium.org/1123843002
Cr-Commit-Position: refs/heads/master@{#328398}
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Tryjob run:
./tools/perf/run_benchmark --browser=trybot-all-android smoothness.top_25_smooth
https://codereview.chromium.org/1122503002
BUG=444701
CQ_EXTRA_TRYBOTS=tryserver.chromium.perf:linux_perf_bisect;tryserver.chromium.perf:mac_perf_bisect;tryserver.chromium.perf:win_perf_bisect;tryserver.chromium.perf:android_nexus5_perf_bisect
Review URL: https://codereview.chromium.org/1117953004
Cr-Commit-Position: refs/heads/master@{#328293}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The smoothness.tough_animation_cases tests occasionally fail to execute
the first frame within the 60s timeout period on Android devices causing
the test to fail.
This change resets the browser state before each test to ensure
things like garbage collection of the previous test don't cause
tests to time out before they're able to initialise their state.
BUG=469684
CQ_EXTRA_TRYBOTS=tryserver.chromium.perf:android_nexus6_perf_bisect
Review URL: https://codereview.chromium.org/1112213006
Cr-Commit-Position: refs/heads/master@{#328112}
|
|
|
|
|
|
|
|
|
| |
BUG=465594
CQ_EXTRA_TRYBOTS=tryserver.chromium.perf:linux_perf_bisect;tryserver.chromium.perf:mac_perf_bisect;tryserver.chromium.perf:win_perf_bisect;tryserver.chromium.perf:android_nexus5_perf_bisect
Review URL: https://codereview.chromium.org/1103163008
Cr-Commit-Position: refs/heads/master@{#327990}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Some sites, notably techcrunch, in top25 were failing to reach readyState
complete or quiesence due to continual requests (possibly an artifact of
the wpr). As this benchmark normalizes/categories results, it shouldn't
be necessary to reach the same state on every run.
Also switches to a tighter tracing category which allows the reference build
to be re-enabled, since the metrics will not appear there until they are
valid to be consumed by the blink_style measurement.
BUG=479048
Review URL: https://codereview.chromium.org/1100853003
Cr-Commit-Position: refs/heads/master@{#327235}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
controller.
Add logic to make sure that page sets paired with
smoothness/repaint only have interaction records with same label & they have repeatable flags enabled.
To make sure that this doesn't crash any smoothness
benchmarks, I modified benchmark_smoke_unittest to run all
smoothness benchmarks (including ones that only support
android) in patch #3. To validate the test runs, see
telemetry_perf_unittest's result in patch #5
smoothness.key_silk_cases: https://codereview.chromium.org/1111543002/
smoothness.polymer:
https://codereview.chromium.org/1060583012
BUG=444701,444703
Review URL: https://codereview.chromium.org/1104573002
Cr-Commit-Position: refs/heads/master@{#327186}
|
|
|
|
|
|
|
|
| |
BUG=475714
Review URL: https://codereview.chromium.org/1103273002
Cr-Commit-Position: refs/heads/master@{#327078}
|
|
|
|
|
|
|
|
|
|
|
| |
Add benchmark and measurement for measuring quiescent power, and enable
it on Mac.
Moved the workaround for http://crbug.com/419786 so that we don't lose idle
wakeups numbers for the renderer process.
Review URL: https://codereview.chromium.org/1092393002
Cr-Commit-Position: refs/heads/master@{#326698}
|
|
|
|
|
|
|
|
| |
BUG=476043
Review URL: https://codereview.chromium.org/1080463003
Cr-Commit-Position: refs/heads/master@{#326657}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Plumb the "painting disabled" mode through to disable the GraphicsContext
in Slimming Paint perf testing. This mode causes Graphics Context to do
no work, producing empty SkPictures for all drawing display items. This
seems the best interpretation of "painting disabled".
The perf results lines for slimming_paint record_time_null_canvas will
go to zero after this patch, and the record_time_construction_disabled
result will start appearing and be zero for non slimming paint.
R=sullivan,skyostil,enne,ajuma
BUG=471348
Review URL: https://codereview.chromium.org/1104433005
Cr-Commit-Position: refs/heads/master@{#326586}
|
|
|
|
|
|
|
|
| |
BUG=475714
Review URL: https://codereview.chromium.org/1103613002
Cr-Commit-Position: refs/heads/master@{#326520}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
smoothness_controller.
To make sure that this doesn't cause any problem to existing
benchmarks, I modified benchmark_smoke_unittest to make sure it run
all smoothness benchmarks on all page sets in temporary patches.
(See the telemetry_perf_unittest's log in patch 6)
BUG=444701
Review URL: https://codereview.chromium.org/1092723002
Cr-Commit-Position: refs/heads/master@{#326310}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
by default
This is a reland of https://codereview.chromium.org/1094723002
This fixes crbug/478268 by removing the test class in
linux_platform_backend_unittest that subclass LinuxPlatformBackend.
Previously, the change to discover may shuffle the ordering of discover claassees.
Since platform/__init__.py use discover to find all platform classes that support
the given device
(see https://code.google.com/p/chromium/codesearch#chromium/src/tools/telemetry/telemetry/core/platform/__init__.py&l=41),
the shuffling might made the test class used instead of the actual
LinuxPlatformBackend class.
BUG=466804, 478268
Review URL: https://codereview.chromium.org/1064523004
Cr-Commit-Position: refs/heads/master@{#326159}
|
|
|
|
|
|
|
|
| |
BUG=479048
Review URL: https://codereview.chromium.org/1080583005
Cr-Commit-Position: refs/heads/master@{#326001}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
smoothness.maps (https://codereview.chromium.org/1094973002)
smoothness.key_silk_cases (https://codereview.chromium.org/1082583003)
smoothness.key_mobile_sites_smooth (https://codereview.chromium.org/1076313004)
smoothness.gpu_rasterization.key_mobile_sites_smooth (same page set as smoothness.key_mobile_sites_smooth)
smoothness.sync_scroll.key_mobile_sites_smooth (same page set as smoothness.key_mobile_sites_smooth)
smothness.simple_mobile_sites (https://codereview.chromium.org/1084193003)
In order to move smoothness to use the control code of TBM,
we first need to remove the code that automatically issuing
marker in smoothness_controller & make sure that all the
pages that are paired with smoothness must issue the
markers themselves in the page.
BUG=444701
Review URL: https://codereview.chromium.org/1088813003
Cr-Commit-Position: refs/heads/master@{#325912}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Covers CSS Parsing and full document style update.
For CSS Parsing we export separate metrics for tokenization and parsing that
occur during page load, ignoring UA and trivial stylesheets. The results are
classified based on token size into regular, large and enormous.
For style update a full style recalc is triggered. The result is normalized
by the number of styles that were updated. There's still variance across
pages for this metric -- my plan is to export additional statistics for the
style engine (eg. StyleResolverStats) to classify into different style
update kinds.
BUG=475370
Review URL: https://codereview.chromium.org/1076053006
Cr-Commit-Position: refs/heads/master@{#325852}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
to True by default (patchset #2 id:20001 of https://codereview.chromium.org/1094723002/)
Reason for revert:
Speculative revert for BUG=478268
Original issue's description:
> [Telemetry] Switch discover.DiscoverClasses's index_by_class_name to True by default
>
> BUG=466804
>
> Committed: https://crrev.com/2d35135080ac73b7d602bb8c6d699293de5dd4d9
> Cr-Commit-Position: refs/heads/master@{#325669}
TBR=dtu@chromium.org,aiolos@chromium.org
BUG=466804
Review URL: https://codereview.chromium.org/1083133004
Cr-Commit-Position: refs/heads/master@{#325754}
|
|
|
|
|
|
|
|
|
|
| |
default
BUG=466804
Review URL: https://codereview.chromium.org/1094723002
Cr-Commit-Position: refs/heads/master@{#325669}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is due to 2 reasons:
+ Page specific logic should belong to page, not page test.
+ Simplify the repaint measurement control code, which make
it easier to port repaint to TimelineBasedMeasurement.
Trybot link for repaint.gpu_rasterization.key_mobile_sites_repaint:
https://codereview.chromium.org/1098543002
BUG=444703, 455391
Review URL: https://codereview.chromium.org/1089223003
Cr-Commit-Position: refs/heads/master@{#325558}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
smoothness_controller.
In order to move repaint to use the control code of TBM, we
first need to remove the code that automatically issuing
marker in smoothness_controller & make sure that all the pages that are paired with repaint must issue the markers
themselves in the page.
BUG=444703
Review URL: https://codereview.chromium.org/1083413002
Cr-Commit-Position: refs/heads/master@{#325355}
|
|
|
|
|
|
|
|
|
|
|
| |
The dashboard is currently confused about which direction indicates an
improvement. This CL makes that explicit.
BUG=470272
Review URL: https://codereview.chromium.org/1057823005
Cr-Commit-Position: refs/heads/master@{#325317}
|
|
|
|
|
|
|
|
|
|
| |
places in perf/.
BUG=475090
Review URL: https://codereview.chromium.org/1081193002
Cr-Commit-Position: refs/heads/master@{#324924}
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The v8_gc_times benchmark was accidentally counting all the accumulated gc
times for previous pages in a page set instead of just the current page.
This CL fixes this.
Perf Sheriffs: This will cause movement in the V8 GC benchmark results but is
not a regression, just a correction in measurement.
Review URL: https://codereview.chromium.org/1073203002
Cr-Commit-Position: refs/heads/master@{#324667}
|
|
|
|
|
|
| |
Review URL: https://codereview.chromium.org/1073983002
Cr-Commit-Position: refs/heads/master@{#324561}
|
|
|
|
|
|
|
|
| |
BUG=475090
Review URL: https://codereview.chromium.org/1070173002
Cr-Commit-Position: refs/heads/master@{#324524}
|
|
|
|
|
|
|
|
| |
BUG=444701
Review URL: https://codereview.chromium.org/1074833002
Cr-Commit-Position: refs/heads/master@{#324480}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In tracing the target pages, we can get 'BlinkGCTimeMeasurement'
events which show time ranges to measure forced GC.
This CL makes oilpan_gc_times test to check each forced GC event is in the ranges.
Background:
OilpanGCTimes test runs forced GCs in each test run and also between them.
We want to track only former ones to measure GC marking time durations precisely.
BUG=438074,465997
TEST=./tools/perf/run_tests oilpan --browser=content-shell-release
TEST=./tools/perf/run_benchmark run oilpan_gc_times.blink_perf_stress --browser=content-shell-release
Review URL: https://codereview.chromium.org/1065573002
Cr-Commit-Position: refs/heads/master@{#324443}
|
|
|
|
|
|
|
|
|
|
|
|
| |
record to TBM.
Move smooth_gesture_util from tools/perf/ to tools/telemetry/telemetry/web_perf/
BUG=468061
Review URL: https://codereview.chromium.org/1067883002
Cr-Commit-Position: refs/heads/master@{#324323}
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Report durations of performLayout directly without doing any analysis.
Ignore other events.
Use interactions.
A "load" event may be implemented in another cl, possibly using https://code.google.com/p/chromium/issues/detail?id=345845
BUG=460208
Review URL: https://codereview.chromium.org/979243003
Cr-Commit-Position: refs/heads/master@{#323984}
|
|
|
|
|
|
|
|
|
|
|
| |
Move getGCReason() and two string definitions into _AddTracingResults()
BUG=438074
TEST=./tools/perf/run_tests oilpan --browser=content-shell-release
Review URL: https://codereview.chromium.org/1067503002
Cr-Commit-Position: refs/heads/master@{#323899}
|
|
|
|
|
|
|
|
| |
BUG=473414
Review URL: https://codereview.chromium.org/1058153002
Cr-Commit-Position: refs/heads/master@{#323791}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
and idle GC.
Before this CL, OilpanGCTimes metric measured time for
'precise' and 'conservative' GC, but now Oilpan GC can run for
other reasons, and some tests need to measure such 'other' GC runs.
In short, we had ['precise', 'conervative'] metrics, and
we will have ['precise', 'conervative', 'forced', 'idle']
metrics.
BUG=438074
TEST=./tools/perf/run_tests oilpan --browser=content-shell-release
Review URL: https://codereview.chromium.org/968773002
Cr-Commit-Position: refs/heads/master@{#323651}
|
|
|
|
|
|
|
|
|
|
|
|
| |
This CL is a preparation to update the arguments of Heap::collectGarbage event.
https://codereview.chromium.org/1046033003/
BUG=438074
TEST=./tools/perf/run_tests oilpan --browser=content-shell-release
Review URL: https://codereview.chromium.org/1054463002
Cr-Commit-Position: refs/heads/master@{#323192}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Before this CL, no GC was run in between test, so setting --pageset-repeat > 1
lead to increased memory usage per test run, reporting inaccurate results.
This CL forces V8 GC (and attached Oilpan GC) after each test run for stable result.
More background:
Blink used to trigger V8 major GC after page unload, but this was changed to
less aggressive GC since r181187, suspected as source of this "leak".
BUG=465997
Review URL: https://codereview.chromium.org/1027133002
Cr-Commit-Position: refs/heads/master@{#322318}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
As part of the Telemetry public API refactor, I wrote a tool that sorts imports.
The tool will be checked in separately, but here's all the import changes in Telemetry and its dependencies.
It'll be easier to test the refactor tool if we do the import sorting in advance, so it doesn't clutter the diffs for the module moves.
Style guide rule: "Within each grouping, imports should be sorted lexicographically, ignoring case, according to each module's full package path."
BUG=449308
TEST=trybots (tools/telemetry/run_tests; tools/perf/run_tests)
Review URL: https://codereview.chromium.org/1029263003
Cr-Commit-Position: refs/heads/master@{#322211}
|
|
|
|
|
|
|
|
|
|
|
|
| |
This CL removes the is_smooth flag from Telemetry, and refactors existing tests
and usages on a case-by-case basis to take this into account.
BUG=444697
R=nednguyen,sullivan
Review URL: https://codereview.chromium.org/1013803003
Cr-Commit-Position: refs/heads/master@{#321015}
|