Chromium Code Reviews
|
|
Chromium Code Reviews
Issue
1712007:
Remove power test from bvt and nightly suite since now we had a dedicated power test suite. (Closed)
|
Created:
10 years, 8 months ago by ericli Modified:
9 years, 7 months ago CC:
chromium-os-reviews_chromium.org, sosa+cc_chromium.org, seano, petkov+cc_chromium.org Visibility:
Public. |
DescriptionRemove power test from bvt and nightly suite since now we had a dedicated power test suite.
Patch Set 1 #
Total comments: 1
Patch Set 2 : patch 1 #
Messages
Total messages: 9 (0 generated)
http://codereview.chromium.org/1712007/diff/1/2 File server/site_tests/suite_BuildVerify/control (left): http://codereview.chromium.org/1712007/diff/1/2#oldcode81 server/site_tests/suite_BuildVerify/control:81: 'power_Resume', Lets keep the power_Resume test in the BVT since it measures the time it takes to go into and out of suspend-to-RAM state and not battery power draw. It would be good to measure these times when the system is running on AC power (which is this case) as well as on Battery power, which will be covered in the power nightly. Other (perhaps better) option is to have the power nightly cover both the AC & battery powered cases for this test.
OK. The only concern is the same keyval is coming from both battery powered system (NightlyPower) and AC powered system (BVT). And they will be plot on the same chart. It will confuse people unless you want the numbers to be plotted this way. Another suggestion would be to emit different key names based on the power input. I guess the test could easily distinguish that. On 2010/04/22 15:53:11, sameer wrote: > http://codereview.chromium.org/1712007/diff/1/2 > File server/site_tests/suite_BuildVerify/control (left): > > http://codereview.chromium.org/1712007/diff/1/2#oldcode81 > server/site_tests/suite_BuildVerify/control:81: 'power_Resume', > Lets keep the power_Resume test in the BVT since it measures the time it takes > to go into and out of suspend-to-RAM state and not battery power draw. It would > be good to measure these times when the system is running on AC power (which is > this case) as well as on Battery power, which will be covered in the power > nightly. > > Other (perhaps better) option is to have the power nightly cover both the AC & > battery powered cases for this test.
On 2010/04/22 16:41:16, ericli wrote: > OK. The only concern is the same keyval is coming from both battery powered > system (NightlyPower) and AC powered system (BVT). And they will be plot on the > same chart. It will confuse people unless you want the numbers to be plotted > this way. > > Another suggestion would be to emit different key names based on the power > input. I guess the test could easily distinguish that. > Can the dashboard differentiate the runs and plot them on different charts on the basis of autotest TAG? I think we will end up wrapping up a bunch of existing tests (wifi, disk, video etc. for example) and measure battery power draw across them. It would be too invasive to go into every test and modify the key names to differentiate AC vs Battery runs. > > > > > On 2010/04/22 15:53:11, sameer wrote: > > http://codereview.chromium.org/1712007/diff/1/2 > > File server/site_tests/suite_BuildVerify/control (left): > > > > http://codereview.chromium.org/1712007/diff/1/2#oldcode81 > > server/site_tests/suite_BuildVerify/control:81: 'power_Resume', > > Lets keep the power_Resume test in the BVT since it measures the time it takes > > to go into and out of suspend-to-RAM state and not battery power draw. It > would > > be good to measure these times when the system is running on AC power (which > is > > this case) as well as on Battery power, which will be covered in the power > > nightly. > > > > Other (perhaps better) option is to have the power nightly cover both the AC & > > battery powered cases for this test.
This is not supported in our chart data model yet, but we can. It will take some time. On Thu, Apr 22, 2010 at 10:16 AM, <snanda@chromium.org> wrote: > On 2010/04/22 16:41:16, ericli wrote: > >> OK. The only concern is the same keyval is coming from both battery >> powered >> system (NightlyPower) and AC powered system (BVT). And they will be plot >> on >> > the > >> same chart. It will confuse people unless you want the numbers to be >> plotted >> this way. >> > > Another suggestion would be to emit different key names based on the power >> input. I guess the test could easily distinguish that. >> > > > Can the dashboard differentiate the runs and plot them on different charts > on > the basis of autotest TAG? I think we will end up wrapping up a bunch of > existing tests (wifi, disk, video etc. for example) and measure battery > power > draw across them. It would be too invasive to go into every test and > modify the > key names to differentiate AC vs Battery runs. > > > > > > > On 2010/04/22 15:53:11, sameer wrote: >> > http://codereview.chromium.org/1712007/diff/1/2 >> > File server/site_tests/suite_BuildVerify/control (left): >> > >> > http://codereview.chromium.org/1712007/diff/1/2#oldcode81 >> > server/site_tests/suite_BuildVerify/control:81: 'power_Resume', >> > Lets keep the power_Resume test in the BVT since it measures the time it >> > takes > >> > to go into and out of suspend-to-RAM state and not battery power draw. >> It >> would >> > be good to measure these times when the system is running on AC power >> (which >> is >> > this case) as well as on Battery power, which will be covered in the >> power >> > nightly. >> > >> > Other (perhaps better) option is to have the power nightly cover both >> the AC >> > & > >> > battery powered cases for this test. >> > > > > http://codereview.chromium.org/1712007/show > -- Eric Li 李咏竹 Google Kirkland
On 2010/04/22 17:51:13, ericli wrote: > This is not supported in our chart data model yet, but we can. It will take > some time. > Lets leave power_Resume in BVT for now. Should I file a feature request to add the tag support in dashboard? This will very useful going forward as we start tracking power draw of subsystems. > On Thu, Apr 22, 2010 at 10:16 AM, <mailto:snanda@chromium.org> wrote: > > > On 2010/04/22 16:41:16, ericli wrote: > > > >> OK. The only concern is the same keyval is coming from both battery > >> powered > >> system (NightlyPower) and AC powered system (BVT). And they will be plot > >> on > >> > > the > > > >> same chart. It will confuse people unless you want the numbers to be > >> plotted > >> this way. > >> > > > > Another suggestion would be to emit different key names based on the power > >> input. I guess the test could easily distinguish that. > >> > > > > > > Can the dashboard differentiate the runs and plot them on different charts > > on > > the basis of autotest TAG? I think we will end up wrapping up a bunch of > > existing tests (wifi, disk, video etc. for example) and measure battery > > power > > draw across them. It would be too invasive to go into every test and > > modify the > > key names to differentiate AC vs Battery runs. > > > > > > > > > > > > > > On 2010/04/22 15:53:11, sameer wrote: > >> > http://codereview.chromium.org/1712007/diff/1/2 > >> > File server/site_tests/suite_BuildVerify/control (left): > >> > > >> > http://codereview.chromium.org/1712007/diff/1/2#oldcode81 > >> > server/site_tests/suite_BuildVerify/control:81: 'power_Resume', > >> > Lets keep the power_Resume test in the BVT since it measures the time it > >> > > takes > > > >> > to go into and out of suspend-to-RAM state and not battery power draw. > >> It > >> would > >> > be good to measure these times when the system is running on AC power > >> (which > >> is > >> > this case) as well as on Battery power, which will be covered in the > >> power > >> > nightly. > >> > > >> > Other (perhaps better) option is to have the power nightly cover both > >> the AC > >> > > & > > > >> > battery powered cases for this test. > >> > > > > > > > > http://codereview.chromium.org/1712007/show > > > > > > -- > Eric Li > 李咏竹 > Google Kirkland >
PTAL. Add it back to BVT. Dont worry about the feature request, I will take care about it. On 2010/04/22 18:09:19, sameer wrote: > On 2010/04/22 17:51:13, ericli wrote: > > This is not supported in our chart data model yet, but we can. It will take > > some time. > > > Lets leave power_Resume in BVT for now. > > Should I file a feature request to add the tag support in dashboard? This will > very useful going forward as we start tracking power draw of subsystems. > > > > On Thu, Apr 22, 2010 at 10:16 AM, <mailto:snanda@chromium.org> wrote: > > > > > On 2010/04/22 16:41:16, ericli wrote: > > > > > >> OK. The only concern is the same keyval is coming from both battery > > >> powered > > >> system (NightlyPower) and AC powered system (BVT). And they will be plot > > >> on > > >> > > > the > > > > > >> same chart. It will confuse people unless you want the numbers to be > > >> plotted > > >> this way. > > >> > > > > > > Another suggestion would be to emit different key names based on the power > > >> input. I guess the test could easily distinguish that. > > >> > > > > > > > > > Can the dashboard differentiate the runs and plot them on different charts > > > on > > > the basis of autotest TAG? I think we will end up wrapping up a bunch of > > > existing tests (wifi, disk, video etc. for example) and measure battery > > > power > > > draw across them. It would be too invasive to go into every test and > > > modify the > > > key names to differentiate AC vs Battery runs. > > > > > > > > > > > > > > > > > > > > > On 2010/04/22 15:53:11, sameer wrote: > > >> > http://codereview.chromium.org/1712007/diff/1/2 > > >> > File server/site_tests/suite_BuildVerify/control (left): > > >> > > > >> > http://codereview.chromium.org/1712007/diff/1/2#oldcode81 > > >> > server/site_tests/suite_BuildVerify/control:81: 'power_Resume', > > >> > Lets keep the power_Resume test in the BVT since it measures the time it > > >> > > > takes > > > > > >> > to go into and out of suspend-to-RAM state and not battery power draw. > > >> It > > >> would > > >> > be good to measure these times when the system is running on AC power > > >> (which > > >> is > > >> > this case) as well as on Battery power, which will be covered in the > > >> power > > >> > nightly. > > >> > > > >> > Other (perhaps better) option is to have the power nightly cover both > > >> the AC > > >> > > > & > > > > > >> > battery powered cases for this test. > > >> > > > > > > > > > > > > http://codereview.chromium.org/1712007/show > > > > > > > > > > > -- > > Eric Li > > 李咏竹 > > Google Kirkland > >
On 2010/04/22 19:23:25, ericli wrote: > PTAL. Add it back to BVT. LGTM, but interested if it looks good to Sameer too. Cheers, David
LGTM |
||||||||||||||||||||||||||||||||||||||||||||||||||||||
