diff options
| author | ernstm@chromium.org <ernstm@chromium.org@0039d316-1c4b-4281-b951-d872f2087c98> | 2013-10-11 21:12:33 +0000 |
|---|---|---|
| committer | ernstm@chromium.org <ernstm@chromium.org@0039d316-1c4b-4281-b951-d872f2087c98> | 2013-10-11 21:12:33 +0000 |
| commit | adbe30f035bfc0c63f763a2d4af165ea2d1cf9c0 (patch) | |
| tree | d30cfc87d61a7db96b0793be39364a38040266e9 /cc/cc.gyp | |
| parent | 8d9a3359b14a5bfbbde3fe766231a73935ad065e (diff) | |
| download | chromium_src-adbe30f035bfc0c63f763a2d4af165ea2d1cf9c0.zip chromium_src-adbe30f035bfc0c63f763a2d4af165ea2d1cf9c0.tar.gz chromium_src-adbe30f035bfc0c63f763a2d4af165ea2d1cf9c0.tar.bz2 | |
telemetry: Refactored rasterize_and_record measurement.
- switched rasterize_and_record benchmark to new benchmark architecture.
- detailled traces are not required anymore, fixing the issue with trace
event buffer overflows.
- changed output to show pixels rasterized and recorded for one iteration of the raster/record loop.
- Further reduced variance by excluding lowres tiles and adding another sleep to the measurement. Stdev of pixel counts is now zero (except for amazon.com). Stdev of time metrics with 100 repetitions is well below 1% of the mean for most pages in the top_25.
R=nduca@chromium.org
BUG=280638
Review URL: https://codereview.chromium.org/25353009
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@228253 0039d316-1c4b-4281-b951-d872f2087c98
Diffstat (limited to 'cc/cc.gyp')
| -rw-r--r-- | cc/cc.gyp | 2 |
1 files changed, 2 insertions, 0 deletions
@@ -69,6 +69,8 @@ 'base/tiling_data.cc', 'base/tiling_data.h', 'base/util.h', + 'debug/benchmark_instrumentation.cc', + 'debug/benchmark_instrumentation.h', 'debug/debug_colors.cc', 'debug/debug_colors.h', 'debug/debug_rect_history.cc', |