Add lightspeed experiments to disable gpu.

Add lightspeed experiments to disable gpu.

BUG=465771
Committed: https://crrev.com/54199ddc0b3bfb598d41682e6c2f2781050ed067
Cr-Commit-Position: refs/heads/master@{#320125}

[Alexei Svitkine (slow), 2015-03-10 16:59:51]

asvitkine@chromium.org changed reviewers:
+ gab@chromium.org, kalman@chromium.org, vangelis@chromium.org

[Alexei Svitkine (slow), 2015-03-10 16:59:52]

asvitkine@chromium.org changed required reviewers:
+ gab@chromium.org

[Alexei Svitkine (slow), 2015-03-10 16:59:52]

gab: Overall review.
vangelis: FYI for GPU - I've tested that this happens early enough.
kalman: FYI for extensions

[not at google - send to devlin, 2015-03-10 17:10:23]

> kalman: FYI for extensions

Let's discuss this on the bug, and not include it as part of this CL.

[Alexei Svitkine (slow), 2015-03-10 17:14:22]

Okay, re-purposing this CL to just do this for GPU only for now. We can
discuss extensions on the bug and do a follow-up CL if needed.

On Tue, Mar 10, 2015 at 1:10 PM, <kalman@chromium.org> wrote:

>
>  kalman: FYI for extensions
>>
>
> Let's discuss this on the bug, and not include it as part of this CL.
>
> https://codereview.chromium.org/993003002/
>

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[Alexei Svitkine (slow), 2015-03-10 17:15:06]

asvitkine@chromium.org changed reviewers:
- kalman@chromium.org

[Alexei Svitkine (slow), 2015-03-10 17:15:46]

Done, now controlling GPU only.

[not at google - send to devlin, 2015-03-10 17:17:18]

kalman@chromium.org changed reviewers:
+ kalman@chromium.org

[not at google - send to devlin, 2015-03-10 17:17:18]

Thanks!

[Alexei Svitkine (slow), 2015-03-10 18:53:44]

gab and vangelis: PTAL

On Tue, Mar 10, 2015 at 1:17 PM, <kalman@chromium.org> wrote:

> Thanks!
>
> https://codereview.chromium.org/993003002/
>

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[gab, 2015-03-10 19:42:12]

https://codereview.chromium.org/993003002/diff/20001/chrome/browser/chrome_br...
File chrome/browser/chrome_browser_field_trials_desktop.cc (right):

https://codereview.chromium.org/993003002/diff/20001/chrome/browser/chrome_br...
chrome/browser/chrome_browser_field_trials_desktop.cc:69: if
(!variations::GetVariationParams("LightSpeed", &params))
Why not use typical GetVariationParamValue("LightSpeed", "NoGpu")?

https://codereview.chromium.org/993003002/diff/20001/chrome/browser/chrome_br...
chrome/browser/chrome_browser_field_trials_desktop.cc:71: base::CommandLine*
command_line = base::CommandLine::ForCurrentProcess();
Inline this below.

[Alexei Svitkine (slow), 2015-03-10 19:48:36]

https://codereview.chromium.org/993003002/diff/20001/chrome/browser/chrome_br...
File chrome/browser/chrome_browser_field_trials_desktop.cc (right):

https://codereview.chromium.org/993003002/diff/20001/chrome/browser/chrome_br...
chrome/browser/chrome_browser_field_trials_desktop.cc:69: if
(!variations::GetVariationParams("LightSpeed", &params))
On 2015/03/10 19:42:12, gab wrote:
> Why not use typical GetVariationParamValue("LightSpeed", "NoGpu")?

This was setting up up multiple parameters in a previous patchset. If we will
end up adding more things here, it will be a smaller diff. (There are no
efficiency gains since GetVariationParamValue() does a copy of the map
internally anyway.)

[Alexei Svitkine (slow), 2015-03-11 15:57:08]

I updated the code to use APIs that are a bit more concise if only one param is
being checked. PTAL.

https://codereview.chromium.org/993003002/diff/20001/chrome/browser/chrome_br...
File chrome/browser/chrome_browser_field_trials_desktop.cc (right):

https://codereview.chromium.org/993003002/diff/20001/chrome/browser/chrome_br...
chrome/browser/chrome_browser_field_trials_desktop.cc:69: if
(!variations::GetVariationParams("LightSpeed", &params))
On 2015/03/10 19:48:36, Alexei Svitkine (slow) wrote:
> On 2015/03/10 19:42:12, gab wrote:
> > Why not use typical GetVariationParamValue("LightSpeed", "NoGpu")?
> 
> This was setting up up multiple parameters in a previous patchset. If we will
> end up adding more things here, it will be a smaller diff. (There are no
> efficiency gains since GetVariationParamValue() does a copy of the map
> internally anyway.)

Done.

https://codereview.chromium.org/993003002/diff/20001/chrome/browser/chrome_br...
chrome/browser/chrome_browser_field_trials_desktop.cc:71: base::CommandLine*
command_line = base::CommandLine::ForCurrentProcess();
On 2015/03/10 19:42:12, gab wrote:
> Inline this below.

Done.

[gab, 2015-03-11 15:59:41]

lgtm (assuming that adding the --disable-gpu at that point is early enough for
it to disable all it intends to disable)

Thanks,
Gab

[Alexei Svitkine (slow), 2015-03-11 16:21:36]

I tested with chrome://gpu and it did show it disabled.
On Mar 11, 2015 11:59 AM, <gab@chromium.org> wrote:

> lgtm (assuming that adding the --disable-gpu at that point is early enough
> for
> it to disable all it intends to disable)
>
> Thanks,
> Gab
>
> https://codereview.chromium.org/993003002/
>

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[vangelis, 2015-03-11 16:33:08]

On 2015/03/11 16:21:36, Alexei Svitkine (slow) wrote:
> I tested with chrome://gpu and it did show it disabled.
> On Mar 11, 2015 11:59 AM, <mailto:gab@chromium.org> wrote:
> 
> > lgtm (assuming that adding the --disable-gpu at that point is early enough
> > for
> > it to disable all it intends to disable)
> >
> > Thanks,
> > Gab
> >
> > https://codereview.chromium.org/993003002/
> >
> 
> To unsubscribe from this group and stop receiving emails from it, send an
email
> to mailto:chromium-reviews+unsubscribe@chromium.org.

We need to be careful with this as disabling GPU also means no WebGL. Ideally
this experimentation should only be done in the early channels (Canary, Dev).
Would the --disable-gpu flag show up in about:version or about:gpu ? 

LGTM

[Alexei Svitkine (slow), 2015-03-11 17:00:59]

On 2015/03/11 16:33:08, vangelis wrote:
> On 2015/03/11 16:21:36, Alexei Svitkine (slow) wrote:
> > I tested with chrome://gpu and it did show it disabled.
> > On Mar 11, 2015 11:59 AM, <mailto:gab@chromium.org> wrote:
> > 
> > > lgtm (assuming that adding the --disable-gpu at that point is early enough
> > > for
> > > it to disable all it intends to disable)
> > >
> > > Thanks,
> > > Gab
> > >
> > > https://codereview.chromium.org/993003002/
> > >
> > 
> > To unsubscribe from this group and stop receiving emails from it, send an
> email
> > to mailto:chromium-reviews+unsubscribe@chromium.org.
> 
> We need to be careful with this as disabling GPU also means no WebGL. Ideally
> this experimentation should only be done in the early channels (Canary, Dev).
> Would the --disable-gpu flag show up in about:version or about:gpu ? 
> 
> LGTM

Thanks! Let me check about about:gpu and about:version. Note: We do plan to
experiment on only canary in a session-randomized trial (so it will re-roll you
if you restart).

[Alexei Svitkine (slow), 2015-03-11 17:29:30]

On 2015/03/11 17:00:59, Alexei Svitkine (slow) wrote:
> On 2015/03/11 16:33:08, vangelis wrote:
> > On 2015/03/11 16:21:36, Alexei Svitkine (slow) wrote:
> > > I tested with chrome://gpu and it did show it disabled.
> > > On Mar 11, 2015 11:59 AM, <mailto:gab@chromium.org> wrote:
> > > 
> > > > lgtm (assuming that adding the --disable-gpu at that point is early
enough
> > > > for
> > > > it to disable all it intends to disable)
> > > >
> > > > Thanks,
> > > > Gab
> > > >
> > > > https://codereview.chromium.org/993003002/
> > > >
> > > 
> > > To unsubscribe from this group and stop receiving emails from it, send an
> > email
> > > to mailto:chromium-reviews+unsubscribe@chromium.org.
> > 
> > We need to be careful with this as disabling GPU also means no WebGL.
Ideally
> > this experimentation should only be done in the early channels (Canary,
Dev).
> > Would the --disable-gpu flag show up in about:version or about:gpu ? 
> > 
> > LGTM
> 
> Thanks! Let me check about about:gpu and about:version. Note: We do plan to
> experiment on only canary in a session-randomized trial (so it will re-roll
you
> if you restart).

--disable-gpu will show up on chrome://gpu and chrome://version. However,
chrome://version will also show the field trial group, so this can be correlated
with the flag to determine that it's caused by the experiment.

[Alexei Svitkine (slow), 2015-03-11 17:29:35]

The CQ bit was checked by asvitkine@chromium.org

[commit-bot: I haz the power, 2015-03-11 17:30:17]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/993003002/40001

[commit-bot: I haz the power, 2015-03-11 19:48:29]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2015-03-11 19:49:31]

Patchset 3 (id:??) landed as
https://crrev.com/54199ddc0b3bfb598d41682e6c2f2781050ed067
Cr-Commit-Position: refs/heads/master@{#320125}