dotnet / dotnet/yarp

Investigate 6.0 vs 7.0 perf

Open
#1,495 0 comments 5 reactions 1 assignee Claimed by @MihaZupan View on GitHub
tenet-performance
Dominant language
C#
Stars
9.6k
Forks
933
Avg merge
12d 18h
Merged PRs (30d)
2

Description

Since the perf benchmarks were updated to track .NET 6.0 and .NET 7.0 back in November, 6.0 has been showing consistently better results on most benchmarks. Why?
![image](https://user-images.githubusercontent.com/1821173/148257783-9cb90bd0-cd42-429a-854b-0a74be7ef562.png)

Interestingly, Miha's 0-byte-read change that caused a big improvement in most charts did not affect the 7.0 10240 benchmark. Why not?
![image](https://user-images.githubusercontent.com/1821173/148257970-d5cdd271-1888-438e-b248-b59eeee70a9b.png)

The one benchmark showing better 7.0 results is 102400. The 0-byte-read change had a very out-sized impact on 7.0.
![image](https://user-images.githubusercontent.com/1821173/148258216-837e898f-8c2c-4bbe-9cbd-44793ed5c202.png)

Others:
h2-h2 is similar or better on 7.0.
h2-http 1024 is better on 7.0, but 10240 and 102400 are much worse, and likely regressed by the 0-byte read change.
https-http 10240 6.0 looks improved by the 0-byte-read change, but 7.0 regressed.

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.