Electrolysis/Release Criteria/Jank: Difference between revisions

Electrolysis/Release Criteria/Jank (view source)

Revision as of 00:27, 25 February 2016

576 bytes removed , 25 February 2016

→‎Talos tp5o_responsiveness: updated tp5o_responsiveness status/format

Avih

Confirmed users

95

edits

@@ Line 48: / Line 48: @@
 This is a combined metric, {{bug|1198650}}. We have made the decision that this no longer blocks e10s, because we are looking at the individual components.
-=== Talos tp5o_responsiveness ===
+* e10s comparison validated: jimm
+* Current e10s diff: much better - ~90% on all platforms
-This test is hugely better in e10s. tp5o_responsiveness on WinXP mozilla-central (results not available on Aurora).
+** No results on OS X
+* Note: measures browser responsiveness during page load. In e10s measured only at the chrome process, therefore the improvement seems real. It would still be useful to also collect data for the content process. TBD.
-Contact points for this test are listed in its [https://wiki.mozilla.org/Buildbot/Talos/Tests#Responsiveness|Talos Test Description]. At time of edit, these are :jimm and :jmaher.
+* {{bug|631571}} - add the test to talos
+* {{bug|710296}} - enable the test in e10s (later comments)
-On consulting the relevant bug ({{bug|710296}}), :avih and :jmaher, and the test's bug ({{bug|631571}}) I learned that the test results should be expected. The test does the following: while loading the tp5 top websites, inject custom tracer events into the parent process' event loop and measure how long it takes for them to percolate through.
-This test is a valid test of chrome responsiveness during content load. Chrome responsiveness under heavy content load is expected to improve (that being one of the reasons for E10s in the first place).
-'''next steps''': PASS (pending :avih's validation and documentation of tp5o_responsiveness' usefulness as an e10s metric)
 === GC pauses ===