1611750 - 14.24 - 30.32% Images (windows7-32, windows7-32-shippable) regression on push b371f612cd60dce156d732f32a50a48eded9d3ab (Wed January 22 2020)

Status: RESOLVED INVALID

(Core :: Graphics: ImageLib, defect, P3)

Description

[Alexandru Ionescu (needinfo me) [:alexandrui]]

We have detected an awsy regression from push:

https://hg.mozilla.org/integration/autoland/pushloghtml?changeset=b371f612cd60dce156d732f32a50a48eded9d3ab

As author of one of the patches included in that push, we need your help to address this regression.

Regressions:

30% Images windows7-32-shippable opt 4,804,730.35 -> 6,261,307.13
28% Images windows7-32-shippable opt 4,879,442.92 -> 6,250,408.42
16% Images windows7-32-shippable opt tp6 6,919,524.97 -> 8,020,466.32
14% Images windows7-32 opt tp6 6,837,913.55 -> 7,811,529.64

Improvements:

7% Heap Unclassified windows7-32 opt 37,805,981.00 -> 35,157,449.98
4% Heap Unclassified windows7-32-shippable opt tp6 53,806,496.75 -> 51,504,654.43
4% Resident Memory windows7-32 opt 618,556,437.43 -> 595,285,905.63
3% Resident Memory windows7-32-shippable opt tp6 684,524,860.82 -> 662,424,210.90
3% Resident Memory windows7-32-shippable opt 618,114,782.20 -> 601,990,560.35
2% JS windows7-32-shippable opt tp6 157,844,937.17 -> 153,936,257.61
0.47% Base Content JS linux1804-64-qr opt 3,729,093.33 -> 3,711,628.00
0.47% Base Content JS macosx1014-64-shippable opt 3,729,860.00 -> 3,712,404.00
0.47% Base Content JS windows10-64 opt 3,789,248.00 -> 3,771,594.67
0.47% Base Content JS windows10-64-qr opt 3,789,393.33 -> 3,771,592.00
0.47% Base Content JS windows10-64-shippable-qr opt 3,789,500.00 -> 3,771,670.67
0.46% Base Content JS linux1804-64 opt 3,728,898.67 -> 3,711,594.67
0.46% Base Content JS linux1804-64-shippable opt 3,728,880.00 -> 3,711,640.00
0.46% Base Content JS linux1804-64-shippable-qr opt 3,728,892.00 -> 3,711,625.33

You can find links to graphs and comparison views for each of the above tests at: https://treeherder.mozilla.org/perf.html#/alerts?id=24803

On the page above you can see an alert for each affected platform as well as a link to a graph showing the history of scores for this test. There is also a link to a treeherder page showing the jobs in a pushlog format.

To learn more about the regressing test(s), please see: https://wiki.mozilla.org/AWSY/Tests

Comment 1

[Andrew Osmond [:aosmond] (he/him)]

This should be fixed by the backout in the regressing bug. However the reports are wrong. There should not be a memory regression; the opposite actually. We don't track memory usage in the surface cache for optimized surfaces; this means things like SourceSurfaceD2D1 and SourceSurfaceCapture claim no heap/nonheap consumption for the raw image data. The "regression" was likely us moving from tracked memory to untracked memory. A patch to fix this should be landed separately so that future regressions can be properly assessed.

Comment 2

[Alexandru Ionescu (needinfo me) [:alexandrui]]

(In reply to Andrew Osmond [:aosmond] from comment #1)

This should be fixed by the backout in the regressing bug. However the reports are wrong. There should not be a memory regression; the opposite actually. We don't track memory usage in the surface cache for optimized surfaces; this means things like SourceSurfaceD2D1 and SourceSurfaceCapture claim no heap/nonheap consumption for the raw image data. The "regression" was likely us moving from tracked memory to untracked memory. A patch to fix this should be landed separately so that future regressions can be properly assessed.

Andrew, I am not sure from your comment if this if fixed or not.

Comment 3

[Andrew Osmond [:aosmond] (he/him)]

Yes it was fixed. Bug 1612589 "regressed" it again, but only because the reporting was flat out wrong, not because I made it worse :).