All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=827144

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=96ae633288f89c278d1c037f1044e4db2f6333f66812e547f57086bdd653bbde


Bot(s) for this bug's original alert(s):

android-nexus5

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/14b09697440000

📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/14b09697440000

viz: Ensure DirectLayerTreeFrameSink frame acks are async. by khushalsagar@google.com
https://chromium.googlesource.com/chromium/src/+/4361dcb9a14ccbf76b21abc3c3592a905430c3b1

Understanding performance regressions:
  http://g.co/ChromePerformanceRegressions

I looked at the load_games_lazors trace on Nexus5x and the complete increase is from a single extra texture of ~730K in the gpu process. The change above affects the timings of when the frame ack is processed in the browser and returned resources are recycled maybe which could affect what we see when a memory dump is taken. I need to look more to be certain.

 Issue 827139  has been merged into this issue.

I could repro it locally, and the increase is definitely from an additional texture in the GPU process. Since the patch above makes a callback async, it changes the timing for when resources are returned to cc and deleted on the GPU.

Issue 827133 has been merged into this issue.