1008762 - Rethink our GC scheduling heuristics in light of GGC

Status: RESOLVED WORKSFORME

(Core :: JavaScript: GC, defect)

Description

[Terrence Cole [:terrence]]

Now that we have GGC, some workloads that would have triggered high-frequency GC will now no longer do so. This got me thinking about a couple other potential pathological cases. Our first step here is to surface enough data and add enough telemetry that we can figure out if there are new pathological behaviors at all.

Comment 1

[Terrence Cole [:terrence]]

I don't think there are; at least nothing too bad has popped up in the last year and a half. We lower the amount of garbage, but the distance from low to high frequency gc is not all that big.