Performance difference: mvc, mvc

[pr-benchmarks]

A performance regression has been detected for the following scenarios:

Scenario	Environment	Date	Old RPS	New RPS	Change	Deviation	StDev
mvc [Fixed]	Plaintext Mvc aspnet-citrine-lin	04/10/2021 20:05:42	2,056,959	1,956,553	-4.88 % (-100,406) 👎	-3.57 σ	28,112.93
mvc [Fixed]	Plaintext Mvc aspnet-citrine-amd	04/10/2021 20:20:38	3,949,675	2,025,105	-48.73 % (-1,924,570) 👎	-10.84 σ	177,589.45

Changes:

Runtime	Previous Version	Current Version	Diff
ASP.NET	6.0.0-preview.4.21209.12	6.0.0-preview.4.21210.2	985ef47...81936a7
CLR	6.0.0-preview.4.21209.14	6.0.0-preview.4.21210.4	dotnet/runtime@403ae35...9b647fe
SDK	6.0.100-preview.4.21209.4	6.0.100-preview.4.21210.1	n/a

@sebastienros

[davidfowl]

😮

[halter73]

Interesting that diff range didn't include dotnet/runtime#50463.

[halter73]

@BrennanConroy figured it out. It's because we don't consume DI from the shared runtime and had to wait for the arcade update. I wonder if the bot can be updated to track this.

[sebastienros]

We explained it on Monday, but you preferred to join a different meeting, your loss ;) I implemented custom package references in crank since then, and enabled "latest" extensions (like runtime) in most benchmarks. So now the runtime diff will actually include extensions, even if arcade is slow on aspnet.

[davidfowl]

@sebastienros I should have let the bot closed this huh.