Blazor time to first UI regression: blazorwasmbenchmark

[pr-benchmarks]

Scenario	Environment	Date	Old (ms)	New (ms)	Change	Dependencies
blazorwasmbenchmark [Fixed]		08/16/2023 04:36:58	258	285	10.49 % (27) 👎	Changes Name Version Diff

PowerBI Dashboard

[MackinnonBuck]

The aspnetcore commit range for this regression is 19e6bdb .. 324a376.

I found that #50080 introduced the regression, which means it probably came from a runtime change.

The commit range for that runtime dependency update is dotnet/runtime@8986374 .. dotnet/runtime@1eb702c.

Within that range, it looks like the primary (or only?) change to the WebAssembly runtime is dotnet/runtime#88279.

@kg, do you think dotnet/runtime#88279 might have increased WebAssembly startup times?

[kg]

@kg, do you think dotnet/runtime#88279 might have increased WebAssembly startup times?

It did, I noticed it a couple days ago in our measurements. One partial fix just landed on main, and another one is in the works.

Some of it is a measurement artifact - the thread safety changes moved an allocation from runtime (during JIT) to startup, causing it to become measured when it wasn't before.

[MackinnonBuck]

Thanks @kg! Glad to hear this is known and fixes are in progress. I'll leave this issue open until the fixes make their way to the Blazor WebAssembly benchmarks.

[kg]

According to our latest measurements, startup time on dotnet/runtime main should now be below where it was in July. Let us know if you're still seeing problems after integrating the latest changes. We still have work in progress to improve it further.