1665573 - 2 - 2.31% Resident Memory (linux1804-64-shippable-qr) regression on push 6ca25e02d89355b5d62c4d137d72cd7a0c7e2373 (Fri September 11 2020)

Status: RESOLVED WONTFIX

(Core :: Graphics: WebRender, defect)

Description

[Florin Strugariu [:Bebe]]

Perfherder has detected a awsy performance regression from push 6ca25e02d89355b5d62c4d137d72cd7a0c7e2373. As author of one of the patches included in that push, we need your help to address this regression.

Regressions:

2% Resident Memory linux1804-64-shippable-qr opt 1,133,226,937.38 -> 1,159,396,358.48
2% Resident Memory linux1804-64-shippable-qr opt 1,140,705,112.29 -> 1,163,470,527.40

Details of the alert can be found in the alert summary, including links to graphs and comparisons for each of the affected tests. Please follow our guide to handling regression bugs and let us know your plans within 3 business days, or the offending patch(es) will be backed out in accordance with our regression policy.

For more information on performance sheriffing please see our FAQ.

Comment 1

[BugBot [:suhaib / :marco/ :calixte]]

Set release status flags based on info from the regressing bug 1664084

Comment 2

[Glenn Watson [:gw]]

What's occurring is that we now make much better use of worker threads on machines with available cores in order to rasterize glyphs in the background.

We store a copy of the font template on each worker thread that is active, which is why the memory usage has increased in this test case.

I think it's a reasonable trade off here to use some extra memory when we have CPU cores available to do extra background glyph rasterization.

Comment 3

[Jim Mathies [:jimm]]

Per Glen's comment, a worthy trade off.