Adding a cache + memory pool for GPU TextBlobs

Adding a cache + memory pool for GPU TextBlobs

BUG=skia:

Committed: https://skia.googlesource.com/skia/+/b7133bed55af8dd4ca9427892bb1a5623dbaccf0

[joshualitt, 2015-04-02 17:47:20]

joshualitt@google.com changed reviewers:
+ bsalomon@google.com, joshualitt@google.com, jvanverth@google.com,
robertphillips@google.com

[joshualitt, 2015-04-02 17:47:21]

My naive attempt at a cache for the textblob stuff.

[joshualitt, 2015-04-03 16:44:35]

On 2015/04/02 17:47:21, joshualitt wrote:
> My naive attempt at a cache for the textblob stuff.

updated to include more accurate counting of memory usage.

[jvanverth1, 2015-04-06 20:44:13]

This looks reasonable + 1 comment. I'll defer to someone else on the changes to
GrMemoryPool, though.

https://codereview.chromium.org/1055843002/diff/120001/src/gpu/GrTextBlobCache.h
File src/gpu/GrTextBlobCache.h (right):

https://codereview.chromium.org/1055843002/diff/120001/src/gpu/GrTextBlobCach...
src/gpu/GrTextBlobCache.h:19: GrTextBlobCache() : fPool(2 << 17, 2 << 17),
fBudget(2 << 22) {}
These seem like magic numbers. I'm not sure why they were chosen.

[joshualitt, 2015-04-07 16:18:50]

On 2015/04/06 20:44:13, jvanverth1 wrote:
> This looks reasonable + 1 comment. I'll defer to someone else on the changes
to
> GrMemoryPool, though.
> 
>
https://codereview.chromium.org/1055843002/diff/120001/src/gpu/GrTextBlobCache.h
> File src/gpu/GrTextBlobCache.h (right):
> 
>
https://codereview.chromium.org/1055843002/diff/120001/src/gpu/GrTextBlobCach...
> src/gpu/GrTextBlobCache.h:19: GrTextBlobCache() : fPool(2 << 17, 2 << 17),
> fBudget(2 << 22) {}
> These seem like magic numbers. I'm not sure why they were chosen.

Jim, we chose the budget to arbitrarily be 4 megs for this cache.  I choose the
alloc sizes pretty arbitrarily based off the largest text blob I saw in any of
the skps.  I have added a comment explaining where they come from.  I'm sure
they can still be tuned.

Also, I have updated GrMemoryPool per Brian's and my offline discussion.

[bsalomon, 2015-04-08 02:04:39]

I think the constants Jim commented should be declared static const w/
meaningful names (e.g. static const int kBudgetedByted , static const int
kPreallocBytes, static const int kGrowBytes).

Also, 2 << 22 is 8MB. I'd write 4MB as (1 << 22) or more likely 4 * (1 << 20)
since it is easy to remember that 1 << 10 is a KB, 1 << 20 is a MB, 1 << 30 is a
GB, etc.

[bsalomon, 2015-04-08 02:05:41]

otherwise, lgtm

[joshualitt, 2015-04-08 15:42:42]

The CQ bit was checked by joshualitt@google.com

[joshualitt, 2015-04-08 15:42:43]

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/1055843002/#ps200001
(title: "feedback inc")

[commit-bot: I haz the power, 2015-04-08 15:42:57]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1055843002/200001

[commit-bot: I haz the power, 2015-04-08 16:08:37]

Committed patchset #11 (id:200001) as
https://skia.googlesource.com/skia/+/b7133bed55af8dd4ca9427892bb1a5623dbaccf0