Add test configs for nvpr with atlas text

Add test configs for nvpr with atlas text

Adds nvpr4 and nvpr16, which use nvpr for paths and atlas for text.

Renames nvprmsaa4 and nvprmsaa16 to nvprdit4, nvprdit16. These use nvpr
for both paths and text.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1834693002

Committed: https://skia.googlesource.com/skia/+/c28afdb4ef9aa1bfc8c424f15abda54247b1b2eb

[Chris Dalton, 2016-03-24 16:44:52]

Description was changed from

==========
Revert back to atlas text on nvpr configs

Reverts the named "nvprmsaa" configs back to using dit=false. The
named configs ought to use the best known implementation by default.
We can still test nvpr text with "gpu(nvpr=true,dit=true..."

BUG=skia:
==========

to

==========
Revert back to atlas text on nvpr configs

Reverts the named "nvprmsaa" configs back to using dit=false. The
named configs ought to use the best known implementation by default.
We can still test nvpr text with "gpu(nvpr=true,dit=true..."

BUG=skia:
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

[Chris Dalton, 2016-03-24 16:47:07]

cdalton@nvidia.com changed reviewers:
+ bsalomon@google.com, mtklein@google.com, scroggo@google.com

[Chris Dalton, 2016-03-24 16:48:34]

Could we still have the gm bots do a run of "gpu(nvpr=true,dit=true..."?
Otherwise this would create a testing hole for nvpr text.

[bsalomon, 2016-03-25 15:42:25]

On 2016/03/24 16:48:34, Chris Dalton wrote:
> Could we still have the gm bots do a run of "gpu(nvpr=true,dit=true..."?
> Otherwise this would create a testing hole for nvpr text.

What will happen in Chromium? Does it use NVPR text? Seems like we at least want
it enabled in DM to avoid rendering regressions.

[Chris Dalton, 2016-03-25 15:48:45]

On 2016/03/25 15:42:25, bsalomon wrote:
> On 2016/03/24 16:48:34, Chris Dalton wrote:
> > Could we still have the gm bots do a run of "gpu(nvpr=true,dit=true..."?
> > Otherwise this would create a testing hole for nvpr text.
> 
> What will happen in Chromium? Does it use NVPR text? Seems like we at least
want
> it enabled in DM to avoid rendering regressions.

Chromium shouldn't be affected, all this does is change the meaning of "--config
nvprmsaa4" in the test apps (dm, nanobench).

I agree we should keep testing alive for nvpr text. I see two ways to do this:

1) nvprmsaa4 means atlas text for nanobench, but nvpr text for dm

2) nvprmsaa4 always means atlas text, and the dm bots start running a new config
of "gpu(nvpr=true,dit=true, samples=4)"

I lean slightly toward #2 because then "nvprmsaa4" always means the same thing.
Which do you guys prefer?

[bsalomon, 2016-03-25 15:52:20]

On 2016/03/25 15:48:45, Chris Dalton wrote:
> On 2016/03/25 15:42:25, bsalomon wrote:
> > On 2016/03/24 16:48:34, Chris Dalton wrote:
> > > Could we still have the gm bots do a run of "gpu(nvpr=true,dit=true..."?
> > > Otherwise this would create a testing hole for nvpr text.
> > 
> > What will happen in Chromium? Does it use NVPR text? Seems like we at least
> want
> > it enabled in DM to avoid rendering regressions.
> 
> Chromium shouldn't be affected, all this does is change the meaning of
"--config
> nvprmsaa4" in the test apps (dm, nanobench).

I know, but I'd like to be sure we test what Chromium is using.

> 
> I agree we should keep testing alive for nvpr text. I see two ways to do this:
> 
> 1) nvprmsaa4 means atlas text for nanobench, but nvpr text for dm
> 
> 2) nvprmsaa4 always means atlas text, and the dm bots start running a new
config
> of "gpu(nvpr=true,dit=true, samples=4)"
> 
> I lean slightly toward #2 because then "nvprmsaa4" always means the same
thing.
> Which do you guys prefer?

I prefer the latter, but giving them distinct alias names e.g. nvprmsaa4 and
nvprmsaa4_bmptext or something along those lines.

[Chris Dalton, 2016-03-25 16:22:18]

On 2016/03/25 15:52:20, bsalomon wrote:
> I know, but I'd like to be sure we test what Chromium is using.

Oh right. Doesn't chromium use atlas text by default unless you enable
device-independent fonts in the flags?

> I prefer the latter, but giving them distinct alias names e.g. nvprmsaa4 and
> nvprmsaa4_bmptext or something along those lines.

What about:

 nvpr4    <- nvprmsaa4
 nvpr16   <- nvprmsaa16
 nvprdit4
 nvprdit16

[bsalomon, 2016-03-25 16:55:17]

On 2016/03/25 16:22:18, Chris Dalton wrote:
> On 2016/03/25 15:52:20, bsalomon wrote:
> > I know, but I'd like to be sure we test what Chromium is using.
> 
> Oh right. Doesn't chromium use atlas text by default unless you enable
> device-independent fonts in the flags?

Yeah, that sounds right.

> 
> > I prefer the latter, but giving them distinct alias names e.g. nvprmsaa4 and
> > nvprmsaa4_bmptext or something along those lines.
> 
> What about:
> 
>  nvpr4    <- nvprmsaa4
>  nvpr16   <- nvprmsaa16
>  nvprdit4
>  nvprdit16

sgtm

[Chris Dalton, 2016-03-25 22:41:02]

Description was changed from

==========
Revert back to atlas text on nvpr configs

Reverts the named "nvprmsaa" configs back to using dit=false. The
named configs ought to use the best known implementation by default.
We can still test nvpr text with "gpu(nvpr=true,dit=true..."

BUG=skia:
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

to

==========
Add test configs for nvpr with atlas text

Adds nvpr4 and nvpr16, which use nvpr for paths and atlas for text.

Renames nvprmsaa4 and nvprmsaa16 to nvprdit4, nvprdit16. These use nvpr
for paths *and* text.

BUG=skia:
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

[Chris Dalton, 2016-03-25 22:41:19]

On 2016/03/25 16:55:17, bsalomon wrote:
> On 2016/03/25 16:22:18, Chris Dalton wrote:
> > On 2016/03/25 15:52:20, bsalomon wrote:
> > > I know, but I'd like to be sure we test what Chromium is using.
> > 
> > Oh right. Doesn't chromium use atlas text by default unless you enable
> > device-independent fonts in the flags?
> 
> Yeah, that sounds right.
> 
> > 
> > > I prefer the latter, but giving them distinct alias names e.g. nvprmsaa4
and
> > > nvprmsaa4_bmptext or something along those lines.
> > 
> > What about:
> > 
> >  nvpr4    <- nvprmsaa4
> >  nvpr16   <- nvprmsaa16
> >  nvprdit4
> >  nvprdit16
> 
> sgtm

Done.

[bsalomon, 2016-03-28 14:41:30]

On 2016/03/25 22:41:19, Chris Dalton wrote:
> On 2016/03/25 16:55:17, bsalomon wrote:
> > On 2016/03/25 16:22:18, Chris Dalton wrote:
> > > On 2016/03/25 15:52:20, bsalomon wrote:
> > > > I know, but I'd like to be sure we test what Chromium is using.
> > > 
> > > Oh right. Doesn't chromium use atlas text by default unless you enable
> > > device-independent fonts in the flags?
> > 
> > Yeah, that sounds right.
> > 
> > > 
> > > > I prefer the latter, but giving them distinct alias names e.g. nvprmsaa4
> and
> > > > nvprmsaa4_bmptext or something along those lines.
> > > 
> > > What about:
> > > 
> > >  nvpr4    <- nvprmsaa4
> > >  nvpr16   <- nvprmsaa16
> > >  nvprdit4
> > >  nvprdit16
> > 
> > sgtm
> 
> Done.

So with this change what configs do the bots run for dm and nanobench?

[Chris Dalton, 2016-03-28 15:55:26]

On 2016/03/28 14:41:30, bsalomon wrote:
> On 2016/03/25 22:41:19, Chris Dalton wrote:
> > On 2016/03/25 16:55:17, bsalomon wrote:
> > > On 2016/03/25 16:22:18, Chris Dalton wrote:
> > > > On 2016/03/25 15:52:20, bsalomon wrote:
> > > > > I know, but I'd like to be sure we test what Chromium is using.
> > > > 
> > > > Oh right. Doesn't chromium use atlas text by default unless you enable
> > > > device-independent fonts in the flags?
> > > 
> > > Yeah, that sounds right.
> > > 
> > > > 
> > > > > I prefer the latter, but giving them distinct alias names e.g.
nvprmsaa4
> > and
> > > > > nvprmsaa4_bmptext or something along those lines.
> > > > 
> > > > What about:
> > > > 
> > > >  nvpr4    <- nvprmsaa4
> > > >  nvpr16   <- nvprmsaa16
> > > >  nvprdit4
> > > >  nvprdit16
> > > 
> > > sgtm
> > 
> > Done.
> 
> So with this change what configs do the bots run for dm and nanobench?

They used to run nvprmsaa4 and nvprmsaa16, but those are being replaced by
nvpr4, nvpr16, nvprdit4, and nvprdit16.

If we aren't pressed for time they might as well both run all 4. Otherwise,
maybe run the 16x configs for dm and the 4x configs for nanobench?

[bsalomon, 2016-03-28 16:02:49]

On 2016/03/28 15:55:26, Chris Dalton wrote:
> On 2016/03/28 14:41:30, bsalomon wrote:
> > On 2016/03/25 22:41:19, Chris Dalton wrote:
> > > On 2016/03/25 16:55:17, bsalomon wrote:
> > > > On 2016/03/25 16:22:18, Chris Dalton wrote:
> > > > > On 2016/03/25 15:52:20, bsalomon wrote:
> > > > > > I know, but I'd like to be sure we test what Chromium is using.
> > > > > 
> > > > > Oh right. Doesn't chromium use atlas text by default unless you enable
> > > > > device-independent fonts in the flags?
> > > > 
> > > > Yeah, that sounds right.
> > > > 
> > > > > 
> > > > > > I prefer the latter, but giving them distinct alias names e.g.
> nvprmsaa4
> > > and
> > > > > > nvprmsaa4_bmptext or something along those lines.
> > > > > 
> > > > > What about:
> > > > > 
> > > > >  nvpr4    <- nvprmsaa4
> > > > >  nvpr16   <- nvprmsaa16
> > > > >  nvprdit4
> > > > >  nvprdit16
> > > > 
> > > > sgtm
> > > 
> > > Done.
> > 
> > So with this change what configs do the bots run for dm and nanobench?
> 
> They used to run nvprmsaa4 and nvprmsaa16, but those are being replaced by
> nvpr4, nvpr16, nvprdit4, and nvprdit16.
> 
> If we aren't pressed for time they might as well both run all 4. Otherwise,
> maybe run the 16x configs for dm and the 4x configs for nanobench?

How about run both 4x configs in dm and nanobench on android, run both 16x
configs on desktop OSs (since they're more likely to be low dpi)?

[Chris Dalton, 2016-03-28 16:17:38]

On 2016/03/28 16:02:49, bsalomon wrote:
> On 2016/03/28 15:55:26, Chris Dalton wrote:
> > On 2016/03/28 14:41:30, bsalomon wrote:
> > > On 2016/03/25 22:41:19, Chris Dalton wrote:
> > > > On 2016/03/25 16:55:17, bsalomon wrote:
> > > > > On 2016/03/25 16:22:18, Chris Dalton wrote:
> > > > > > On 2016/03/25 15:52:20, bsalomon wrote:
> > > > > > > I know, but I'd like to be sure we test what Chromium is using.
> > > > > > 
> > > > > > Oh right. Doesn't chromium use atlas text by default unless you
enable
> > > > > > device-independent fonts in the flags?
> > > > > 
> > > > > Yeah, that sounds right.
> > > > > 
> > > > > > 
> > > > > > > I prefer the latter, but giving them distinct alias names e.g.
> > nvprmsaa4
> > > > and
> > > > > > > nvprmsaa4_bmptext or something along those lines.
> > > > > > 
> > > > > > What about:
> > > > > > 
> > > > > >  nvpr4    <- nvprmsaa4
> > > > > >  nvpr16   <- nvprmsaa16
> > > > > >  nvprdit4
> > > > > >  nvprdit16
> > > > > 
> > > > > sgtm
> > > > 
> > > > Done.
> > > 
> > > So with this change what configs do the bots run for dm and nanobench?
> > 
> > They used to run nvprmsaa4 and nvprmsaa16, but those are being replaced by
> > nvpr4, nvpr16, nvprdit4, and nvprdit16.
> > 
> > If we aren't pressed for time they might as well both run all 4. Otherwise,
> > maybe run the 16x configs for dm and the 4x configs for nanobench?
> 
> How about run both 4x configs in dm and nanobench on android, run both 16x
> configs on desktop OSs (since they're more likely to be low dpi)?

Ok, I like that idea. Do I need to change something on my end to make it happen?

[bsalomon, 2016-03-28 16:52:00]

On 2016/03/28 16:17:38, Chris Dalton wrote:
> On 2016/03/28 16:02:49, bsalomon wrote:
> > On 2016/03/28 15:55:26, Chris Dalton wrote:
> > > On 2016/03/28 14:41:30, bsalomon wrote:
> > > > On 2016/03/25 22:41:19, Chris Dalton wrote:
> > > > > On 2016/03/25 16:55:17, bsalomon wrote:
> > > > > > On 2016/03/25 16:22:18, Chris Dalton wrote:
> > > > > > > On 2016/03/25 15:52:20, bsalomon wrote:
> > > > > > > > I know, but I'd like to be sure we test what Chromium is using.
> > > > > > > 
> > > > > > > Oh right. Doesn't chromium use atlas text by default unless you
> enable
> > > > > > > device-independent fonts in the flags?
> > > > > > 
> > > > > > Yeah, that sounds right.
> > > > > > 
> > > > > > > 
> > > > > > > > I prefer the latter, but giving them distinct alias names e.g.
> > > nvprmsaa4
> > > > > and
> > > > > > > > nvprmsaa4_bmptext or something along those lines.
> > > > > > > 
> > > > > > > What about:
> > > > > > > 
> > > > > > >  nvpr4    <- nvprmsaa4
> > > > > > >  nvpr16   <- nvprmsaa16
> > > > > > >  nvprdit4
> > > > > > >  nvprdit16
> > > > > > 
> > > > > > sgtm
> > > > > 
> > > > > Done.
> > > > 
> > > > So with this change what configs do the bots run for dm and nanobench?
> > > 
> > > They used to run nvprmsaa4 and nvprmsaa16, but those are being replaced by
> > > nvpr4, nvpr16, nvprdit4, and nvprdit16.
> > > 
> > > If we aren't pressed for time they might as well both run all 4.
Otherwise,
> > > maybe run the 16x configs for dm and the 4x configs for nanobench?
> > 
> > How about run both 4x configs in dm and nanobench on android, run both 16x
> > configs on desktop OSs (since they're more likely to be low dpi)?
> 
> Ok, I like that idea. Do I need to change something on my end to make it
happen?

The flags passed to dm and nanobench on the bots are set in tools/dm_flags.py
and tools/nanobench_flags.py. They should probably change as part of this CL.

[Chris Dalton, 2016-03-28 18:11:27]

Description was changed from

==========
Add test configs for nvpr with atlas text

Adds nvpr4 and nvpr16, which use nvpr for paths and atlas for text.

Renames nvprmsaa4 and nvprmsaa16 to nvprdit4, nvprdit16. These use nvpr
for paths *and* text.

BUG=skia:
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

to

==========
Add test configs for nvpr with atlas text

Adds nvpr4 and nvpr16, which use nvpr for paths and atlas for text.

Renames nvprmsaa4 and nvprmsaa16 to nvprdit4, nvprdit16. These use nvpr
for both paths and text.

BUG=skia:
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

[Chris Dalton, 2016-03-28 18:21:11]

On 2016/03/28 16:52:00, bsalomon wrote:
> The flags passed to dm and nanobench on the bots are set in tools/dm_flags.py
> and tools/nanobench_flags.py. They should probably change as part of this CL.

This round does only nvprdit for dm, and nvpr + nvprdit for nanobench.

I don't see a compelling reason to take the extra time to test both nvpr and
nvprdit in dm.

[Chris Dalton, 2016-03-28 18:28:42]

On 2016/03/28 18:21:11, Chris Dalton wrote:
> On 2016/03/28 16:52:00, bsalomon wrote:
> > The flags passed to dm and nanobench on the bots are set in
tools/dm_flags.py
> > and tools/nanobench_flags.py. They should probably change as part of this
CL.
> 
> This round does only nvprdit for dm, and nvpr + nvprdit for nanobench.
> 
> I don't see a compelling reason to take the extra time to test both nvpr and
> nvprdit in dm.

What about dm_flags.json and nanobench_flags.json?

[bsalomon, 2016-03-28 18:41:45]

On 2016/03/28 18:28:42, Chris Dalton wrote:
> On 2016/03/28 18:21:11, Chris Dalton wrote:
> > On 2016/03/28 16:52:00, bsalomon wrote:
> > > The flags passed to dm and nanobench on the bots are set in
> tools/dm_flags.py
> > > and tools/nanobench_flags.py. They should probably change as part of this
> CL.
> > 
> > This round does only nvprdit for dm, and nvpr + nvprdit for nanobench.
> > 
> > I don't see a compelling reason to take the extra time to test both nvpr and
> > nvprdit in dm.
> 
> What about dm_flags.json and nanobench_flags.json?

There is a way to regen them... Does anyone else on this thread remember how to
do that?

[Chris Dalton, 2016-03-29 20:25:04]

ping?

[bsalomon, 2016-03-30 01:43:48]

lgtm

[Chris Dalton, 2016-03-30 02:54:31]

The CQ bit was checked by cdalton@nvidia.com

[commit-bot: I haz the power, 2016-03-30 02:54:41]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1834693002/60001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1834693002/60001

[commit-bot: I haz the power, 2016-03-30 03:05:11]

Description was changed from

==========
Add test configs for nvpr with atlas text

Adds nvpr4 and nvpr16, which use nvpr for paths and atlas for text.

Renames nvprmsaa4 and nvprmsaa16 to nvprdit4, nvprdit16. These use nvpr
for both paths and text.

BUG=skia:
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

to

==========
Add test configs for nvpr with atlas text

Adds nvpr4 and nvpr16, which use nvpr for paths and atlas for text.

Renames nvprmsaa4 and nvprmsaa16 to nvprdit4, nvprdit16. These use nvpr
for both paths and text.

BUG=skia:
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...

Committed:
https://skia.googlesource.com/skia/+/c28afdb4ef9aa1bfc8c424f15abda54247b1b2eb
==========

[commit-bot: I haz the power, 2016-03-30 03:05:12]

Committed patchset #4 (id:60001) as
https://skia.googlesource.com/skia/+/c28afdb4ef9aa1bfc8c424f15abda54247b1b2eb

[scroggo, 2016-04-04 16:59:22]

On 2016/03/28 18:41:45, bsalomon wrote:
> On 2016/03/28 18:28:42, Chris Dalton wrote:
> > On 2016/03/28 18:21:11, Chris Dalton wrote:
> > > On 2016/03/28 16:52:00, bsalomon wrote:
> > > > The flags passed to dm and nanobench on the bots are set in
> > tools/dm_flags.py
> > > > and tools/nanobench_flags.py. They should probably change as part of
this
> > CL.
> > > 
> > > This round does only nvprdit for dm, and nvpr + nvprdit for nanobench.
> > > 
> > > I don't see a compelling reason to take the extra time to test both nvpr
and
> > > nvprdit in dm.
> > 
> > What about dm_flags.json and nanobench_flags.json?
> 
> There is a way to regen them... Does anyone else on this thread remember how
to
> do that?

Sorry, I was OOO last week so I did not get a chance to respond. You can
regenerate by running:

$ python tools/dm_flags.py test

I just ran that locally, and did not see a difference in the json output. This
surprised me, since I thought we required it to change if you changed the python
script. Someone must have already fixed it (possibly incidentally by
regenerating for their own change.)

[Chris Dalton, 2016-04-04 19:28:44]

Yep! Looks like they got updated with this change:

https://codereview.chromium.org/1846603002