Stephen Illingworth
Stephen Illingworth
The remote testing I've managed to do so far, has established that the error is in the GUI part of the program. If I run the emulator in terminal mode,...
> I need some more specific instructions to reproduce. So far: > > ``` > git clone https://github.com/JetSetIlly/Gopher2600 > cd Gopher2600 > git checkout d1482c139ab9c78b18902426e87367a58cf8b101 > ~/go1.22.0/bin/go build > ```...
@randall77 If you clone this project now and run 'go build .' it should compile without requiring any external C libraries (ie. no SDL or OpenGL). There's also no need...
> https://github.com/JetSetIlly/gopher2600_performance_profile 404s, is it private ? No. I missed the s off the end when I was copying and pasting. Doh! https://github.com/JetSetIlly/gopher2600_performance_profiles
@randall77 Thanks for looking at this. If we take an average of the runs for 1.21.6 and 1.22.0 then there's a drop of 1.86% in performance. I would say that's...
> > If we take an average of the runs for 1.21.6 and 1.22.0 then there's a drop of 1.86% in performance. > > Sure, but the run-to-run variance can...
I've updated the test program to output a rating every second, running for a total of one minute. This gives a clearer picture of what's happening I think. An example...
@qiulaidongfeng Nice! I didn't know about benchstat. These are my results ``` │ benchstat_1.27.1_maxgoproc_n.txt │ benchstat_1.22.0_maxgoproc_n.txt │ │ fps/op │ fps/op vs base │ Test 145.9 ± 1% 139.0 ±...
@ouvaa You've made me very sad. In case it wasn't obvious, I'm not complaining about the development of Go. I hope I'm helping to improve it by identifying an area...
> @JetSetIlly Can you have your test program output the [format](https://go.googlesource.com/proposal/+/master/design/14313-benchmark-format.md) that [benchstat](https://pkg.go.dev/golang.org/x/[email protected]/cmd/benchstat) accepts, and then use benchstat and git bisect (good=go1.21.0 bad=go1.22.0) ? Run the test at least 10...