Began logging more gpu stats from nanobench

Began logging more gpu stats from nanobench

BUG=skia:

Committed: https://skia.googlesource.com/skia/+/e45c81c8f3bb62e45f2ff3b8772b4b23f1ddc6ca

[joshualitt, 2015-12-01 21:09:47]

joshualitt@google.com changed reviewers:
+ bsalomon@google.com, joshualitt@google.com, mtklein@google.com

[joshualitt, 2015-12-01 21:09:47]

So, not sure if we want to use a dictionary, or just a TArray of key/value
struct, so I thought I'd just put this up for review and we can discuss.  Also,
I'm not sure if we want to make one dump function and have a printer wrap it in
an SkString.  I'm flexible.

[bsalomon, 2015-12-01 21:25:07]

Should we make the SkpBench do an extra untiled draw iff on a gpu config and
return the generated stats along with the benchmark time?

[joshualitt, 2015-12-01 22:19:27]

On 2015/12/01 21:25:07, bsalomon wrote:
> Should we make the SkpBench do an extra untiled draw iff on a gpu config and
> return the generated stats along with the benchmark time?

feedback incorporated(I think)

[bsalomon, 2015-12-02 15:06:53]

lgtm

[joshualitt, 2015-12-02 15:37:44]

On 2015/12/02 15:06:53, bsalomon wrote:
> lgtm

take another quick look at the delta between 4 & 3 please, its fairly minor but
incorporates what we just discussed.

[bsalomon, 2015-12-02 15:44:26]

https://codereview.chromium.org/1489033004/diff/100001/bench/SKPBench.cpp
File bench/SKPBench.cpp (right):

https://codereview.chromium.org/1489033004/diff/100001/bench/SKPBench.cpp#new...
bench/SKPBench.cpp:159: static void draw_pic(SkCanvas* canvas, GrContext*
context, const SkPicture* picture,
draw_pic_for_stats?

Does offset need to be a param or could you just do
int offset = keys.count();
before running the picture?

https://codereview.chromium.org/1489033004/diff/100001/bench/SKPBench.cpp#new...
bench/SKPBench.cpp:186: context->resetGpuStats();
maybe make draw_pic reset stats before calling drawPicture rather than after?

https://codereview.chromium.org/1489033004/diff/100001/bench/nanobench.cpp
File bench/nanobench.cpp (right):

https://codereview.chromium.org/1489033004/diff/100001/bench/nanobench.cpp#ne...
bench/nanobench.cpp:1230: if (runs++ % FLAGS_flushEvery == 0) {
why move this?

[joshualitt, 2015-12-02 15:53:06]

feedback inc

https://codereview.chromium.org/1489033004/diff/100001/bench/SKPBench.cpp
File bench/SKPBench.cpp (right):

https://codereview.chromium.org/1489033004/diff/100001/bench/SKPBench.cpp#new...
bench/SKPBench.cpp:159: static void draw_pic(SkCanvas* canvas, GrContext*
context, const SkPicture* picture,
On 2015/12/02 15:44:26, bsalomon wrote:
> draw_pic_for_stats?
> 
> Does offset need to be a param or could you just do
> int offset = keys.count();
> before running the picture?

Acknowledged.

https://codereview.chromium.org/1489033004/diff/100001/bench/SKPBench.cpp#new...
bench/SKPBench.cpp:186: context->resetGpuStats();
On 2015/12/02 15:44:26, bsalomon wrote:
> maybe make draw_pic reset stats before calling drawPicture rather than after?

Acknowledged.

https://codereview.chromium.org/1489033004/diff/100001/bench/nanobench.cpp
File bench/nanobench.cpp (right):

https://codereview.chromium.org/1489033004/diff/100001/bench/nanobench.cpp#ne...
bench/nanobench.cpp:1230: if (runs++ % FLAGS_flushEvery == 0) {
On 2015/12/02 15:44:26, bsalomon wrote:
> why move this?

I accidentally moved it in my initial CL, its back to where it is at TOT

[bsalomon, 2015-12-02 15:58:11]

lgtm

https://codereview.chromium.org/1489033004/diff/120001/bench/SKPBench.cpp
File bench/SKPBench.cpp (right):

https://codereview.chromium.org/1489033004/diff/120001/bench/SKPBench.cpp#new...
bench/SKPBench.cpp:159: static void draw_pic_for_stats(SkCanvas* canvas,
GrContext* context, const SkPicture* picture,
might need to protect this by a SK_SUPPORT_GPU

https://codereview.chromium.org/1489033004/diff/120001/src/gpu/GrTest.cpp
File src/gpu/GrTest.cpp (right):

https://codereview.chromium.org/1489033004/diff/120001/src/gpu/GrTest.cpp#new...
src/gpu/GrTest.cpp:174: keys->push_back(SkString("shader_compilations"));
values->push_back(fShaderCompilations);
do these get collected? I can't remember if freeGpuResource frees the program
cache.

[joshualitt, 2015-12-02 16:01:45]

The CQ bit was checked by joshualitt@google.com

[commit-bot: I haz the power, 2015-12-02 16:01:59]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1489033004/120001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1489033004/120001

[commit-bot: I haz the power, 2015-12-02 16:07:27]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2015-12-02 16:07:28]

Try jobs failed on following builders:
  Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Debug-Trybot on client.skia (JOB_FAILED,
http://build.chromium.org/p/client.skia/builders/Test-Ubuntu-GCC-GCE-CPU-AVX2...)

[joshualitt, 2015-12-02 16:52:31]

The CQ bit was checked by joshualitt@google.com

[joshualitt, 2015-12-02 16:52:31]

The patchset sent to the CQ was uploaded after l-g-t-m from bsalomon@google.com
Link to the patchset: https://codereview.chromium.org/1489033004/#ps140001
(title: "no gpu support")

[commit-bot: I haz the power, 2015-12-02 16:52:39]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1489033004/140001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1489033004/140001

[commit-bot: I haz the power, 2015-12-02 17:05:40]

Description was changed from

==========
Began logging more gpu stats from nanobench

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

to

==========
Began logging more gpu stats from nanobench

BUG=skia:

Committed:
https://skia.googlesource.com/skia/+/e45c81c8f3bb62e45f2ff3b8772b4b23f1ddc6ca
==========

[commit-bot: I haz the power, 2015-12-02 17:05:40]

Committed patchset #8 (id:140001) as
https://skia.googlesource.com/skia/+/e45c81c8f3bb62e45f2ff3b8772b4b23f1ddc6ca