https://bugs.winehq.org/show_bug.cgi?id=38272
--- Comment #9 from Nikolay Sivov bunglehead@gmail.com --- (In reply to Alex Brainman from comment #8)
(In reply to Nikolay Sivov from comment #6)
Do you happen to know why QueryPerformanceCounter/QueryPerformanceFrequency/NtQueryPerformanceCounter wasn't used for that?
As I recall it, QPC didn't work properly on my PC. From https://codereview.appspot.com/108700045#msg11
... But it does not work on my XP https://code.google.com/p/go/issues/detail?id=6007#c36 ...
If you follow that link, you can see small program and its output. As I interpret my results, QPC is wrong.
Yes, looks strange. However it will be interesting to see native program result on this system. Also msdn claims that in some cases XP will fail to provide accurate values, depending on hardware. How can I run this Go test program of yours in Wine? Could you attach a binary here?
And second reason was performance. Some CL reviewers raised concerns about QPC performance. And Dmitriy suggested something simple. I was sceptical first, but we haven't had anyone complained yet. Except people who use WINE.
Yes, for Wine QPC would be preferred currently, as long as we don't have an acceptable solution that updates this process memory data you're using (and having another entity like a separate thread to update it probably won't help performance, but it needs testing).
Alex