Chromium Code Reviews Chromium Code Reviews
Issue
1371113004:
Debugging perf regression after refactoring (Closed)
DescriptionDebugging perf regression after refactoring
Output of the analyze_results.py script (mac, media_bitrate):
Patch #14: (mod == master, but CreateCapturer on a different position)
ScreenshareSlidesVP8_2TL_Before 294835.2 +- 7469.0 (n=10)
ScreenshareSlidesVP8_2TL_Master 326086.8 +- 11721.0 (n=10)
ScreenshareSlidesVP8_2TL_Mod 299158.8 +- 10296.9 (n=10)
Increase after the new patch:
ScreenshareSlidesVP8_2TL master: 10.6% mod: 1.5%
(broken) (NOT broken)
Patch #15: (mod == master)
ScreenshareSlidesVP8_2TL_Before 295947.6 +- 11251.2 (n=10)
ScreenshareSlidesVP8_2TL_Master 326542.4 +- 11255.0 (n=10)
ScreenshareSlidesVP8_2TL_Mod 327613.7 +- 8351.3 (n=10)
Increase after the new patch:
ScreenshareSlidesVP8_2TL master: 10.3% mod: 10.7%
(broken) (broken)
BUG=534220
Patch Set 1 #Patch Set 2 : rebase master #Patch Set 3 : Repeat the tests 10 times #Patch Set 4 : rebase master #Patch Set 5 : Yeah, each test two seconds, great #Patch Set 6 : Running 3x8 tests #Patch Set 7 : Run screenshare_slides #Patch Set 8 : Enable LayerFilteringTransport #Patch Set 9 : Enable almost everything #Patch Set 10 : Alternative version #Patch Set 11 : rebase master #Patch Set 12 : Reverting encode_time_ms... #Patch Set 13 : Alternative again #Patch Set 14 : baremetal not rel #Patch Set 15 : Only the CreateCapturer on a different position #Patch Set 16 : Just in case, everything the same as master #Patch Set 17 : Adding analyze_results.py #
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||
