[bisect-kit] optimization: cros flash with --send-payload-in-parallel |
||
Issue descriptionOur bisect server is in tpe and lab DUTs are (mostly?) in mtv, cros flash is slow (about 25-50 minutes, IIRC) cros flash has option --send-payload-in-parallel which may improve cros flash speed.
,
Sep 6
The following revision refers to this bug: https://chromium.googlesource.com/chromiumos/platform/bisect-kit/+/73e6017f5bdcae864a4a2524a6a60d04f555d415 commit 73e6017f5bdcae864a4a2524a6a60d04f555d415 Author: Kuang-che Wu <kcwu@chromium.org> Date: Thu Sep 06 15:46:33 2018 bisect-kit: use --send-payload-in-parallel speed up cros flash By my not scientific experiments on 4 devices, this flag reduced 2-3 minutes for 3 devices, and no difference for 1 device. See detail numbers in the bug comments. BUG= chromium:878705 TEST=switch_cros_prebuilt.py still works Change-Id: I4c41f9f080e3f7d0dd23aa61f96eb32f1999854c Reviewed-on: https://chromium-review.googlesource.com/1208126 Commit-Ready: Kuang-che Wu <kcwu@chromium.org> Tested-by: Kuang-che Wu <kcwu@chromium.org> Reviewed-by: Chung-yih Wang <cywang@chromium.org> [modify] https://crrev.com/73e6017f5bdcae864a4a2524a6a60d04f555d415/bisect_kit/cros_util.py
,
Sep 6
|
||
►
Sign in to add a comment |
||
Comment 1 by kcwu@chromium.org
, Sep 6I randomly choosed 4 devices (different boards) to test the flag --send-payload-in-parallel. All 4 devices are in MTV lab (but I don't know their network condition/routing etc.) And I run following command on my workstation in TPE. $ time cros flash $host xbuddy://remote/R71-11030.0.0/test and $ time cros flash $host xbuddy://remote/R71-11030.0.0/test --send-payload-in-parallel ================================================================== host 1 x without + with --send-payload-in-parallel +------------------------------------------------------------+ | + + | | + ++ + xx x xx| | |___AM__|| ||_____MA_______| | +------------------------------------------------------------+ N Min Max Median Avg Stddev x 5 759.68404 788.59561 779.76808 775.46912 12.854345 + 6 606.43274 671.35565 620.91557 624.73901 23.81555 Difference at 95.0% confidence -150.73 +/- 26.9988 -19.4373% +/- 3.48161% (Student's t, pooled s = 19.7114) ================================================================== host2 x without + with --send-payload-in-parallel +------------------------------------------------------------+ | + | | x + | | xx+ | | +xx* | | +*x*+ | | ++****x x x | | +****** x + x + x | | + *+****** x ++ xx ++x xx + + x +| | |_____M____A_________| | ||_______M___A__________| | +------------------------------------------------------------+ N Min Max Median Avg Stddev x 39 475.63217 596.94456 486.69279 496.59971 23.862541 + 40 469.13674 602.59352 485.31915 494.06053 26.315035 No difference proven at 95.0% confidence ================================================================== host3 x without + with --send-payload-in-parallel +------------------------------------------------------------+ |+ ++ + + + xxxx x| | |_A__| | ||______MA______| | +------------------------------------------------------------+ N Min Max Median Avg Stddev x 6 809.20534 833.29637 815.64993 817.09348 8.5752985 + 6 660.99566 719.19032 682.41796 683.27478 22.025438 Difference at 95.0% confidence -133.819 +/- 21.4987 -16.3774% +/- 2.63112% (Student's t, pooled s = 16.7131) ================================================================== host4 x without + with --send-payload-in-parallel +------------------------------------------------------------+ | + + x x | |+ + ++ + + xx xx x x| | |___M_A_____| | ||____M_A_______| | +------------------------------------------------------------+ N Min Max Median Avg Stddev x 8 960.89797 1040.0684 976.02419 986.21065 25.99646 + 8 778.7987 882.26731 800.51862 809.80691 32.715244 Difference at 95.0% confidence -176.404 +/- 31.6896 -17.887% +/- 3.21327% (Student's t, pooled s = 29.5474) ================================================================== Conclusion: no difference on one device, reduced 2-3 minutes on other 3 devices.