Add perf tests based on json representation of layer tree

Add perf tests based on json representation of layer tree

This extends cc_perftests to run perf tests based on a json dump of a layer tree
with minimal code changes and adds a tree representing a tab switch UI with
two stacks of 10 layers each.

BUG=


Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=170411

[jamesr, 2012-11-29 08:36:02]

This isn't fully clean yet, but it's darn handy.

I generated the JSON by dumping layerTreeAsText() and then running some very
ugly processing code on it.  It's not the prettiest JSON - in particular it
could be a lot more fewer newlines and could deal with omitting useless fields -
but it's a start.

Since there's file I/O and a good bit of processing in the test setup now we
need a way to separate that from the normal timing.  Will think about that next.

[jamesr, 2012-11-30 00:12:04]

I think this is good enough for a first cut.  Can you do a review, Dana?

Once this is in TonyG has offered to help us get this up+running on the perf
infrastructure so we can monitor the results of all the optimization work across
all platforms, etc

[danakj, 2012-11-30 01:21:00]

LGTM with some minor comments

https://codereview.chromium.org/11418235/diff/3001/cc/layer_tree_host_perftes...
File cc/layer_tree_host_perftest.cc (right):

https://codereview.chromium.org/11418235/diff/3001/cc/layer_tree_host_perftes...
cc/layer_tree_host_perftest.cc:38: impl->setNeedsRedraw();
Probly won't matter, but this should be the else clause of endTest() so we get a
reliable # of frames.

https://codereview.chromium.org/11418235/diff/3001/cc/layer_tree_host_perftes...
cc/layer_tree_host_perftest.cc:270: } else if (layer_type == "ContentLayer") {
aside: we should dump the opaque info for these and parse it. i was just
assuming fully opaque for this scenario since it's tab contents. for solid color
too, as color affects opacity.

https://codereview.chromium.org/11418235/diff/3001/cc/layer_tree_host_perftes...
cc/layer_tree_host_perftest.cc:284: } else { // Type "Layer" or unknown"
mismatched "

https://codereview.chromium.org/11418235/diff/3001/cc/layer_tree_host_perftes...
cc/layer_tree_host_perftest.cc:290: for (int i=0; i<16; ++i) {
no {}

https://codereview.chromium.org/11418235/diff/3001/cc/layer_tree_host_perftes...
cc/layer_tree_host_perftest.cc:290: for (int i=0; i<16; ++i) {
spaces around operators

https://codereview.chromium.org/11418235/diff/3001/cc/layer_tree_host_perftes...
cc/layer_tree_host_perftest.cc:305: ADD_FAILURE();
<< "Unable to parse layer."

https://codereview.chromium.org/11418235/diff/3001/cc/layer_tree_host_perftes...
cc/layer_tree_host_perftest.cc:311: gfx::Size viewport = gfx::Size(720, 1038);
aside: maybe this should be in the json too, or be set in TEST_F where we choose
the data file.

[danakj, 2012-11-30 01:53:03]

LGTM still

[commit-bot: I haz the power, 2012-11-30 02:07:59]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jamesr@chromium.org/11418235/11001

[commit-bot: I haz the power, 2012-11-30 03:34:01]

Step "update" is always a major failure.
Look at the try server FAQ for more details.

[commit-bot: I haz the power, 2012-11-30 04:43:48]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jamesr@chromium.org/11418235/11001

[commit-bot: I haz the power, 2012-11-30 05:51:47]

Change committed as 170411