New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Fixed resetting performance counters related to idle-rate, etc. #1887
Conversation
- also fly-by changes to preprocessor constant usage
Okay I'll check it out |
got some negative counts so I don't think this is working. I will investigate further but in parallel I am going to make another branch for my work around |
In debug this should trigger assertions. In any case, what counters show those negative numbers? |
[20:47] patg: hkaiser: to answer your question from earlier the counter that gave me negative values was /threads/time/average-overhead |
@hkaiser (had to add .txt to upload it here) the output is:
for comparison
example_fib.sh.txt
and no evaluation:
|
I fixed the above calculation but while running some tests saw other problems. I need to investigate more tomorrow .....actually that would be later today |
Found and fixed the problems which were for both get_thread_duration and get_thread_phase_duration calculations for individual worker threads. |
@pagrubel Are the problems you had fix now? |
Those problems seem fixed. However, if someone queries total counts and worker-thread counts at the same time there will be problems. Should we just document this or prevent it. Also, I would like to check this out some more. |
I think we should document that resetting counters may have unwanted side-effects. The problems you're listing are not unique. Other scenarios could create problems as well. Is it ok to merge this now? |
Yes, all is working now. |
Fixed resetting performance counters related to idle-rate, etc.
@pagrubel Please verify whether this fixes your issue reported by #1859