Windows 10 thread scheduling
-
Hi guys, I've got an MFC program that I've been developing for, well, since Windows 2000. We heavily leverage threads in this app - multiple background worker threads at a low priority, a rendering thread at a higher (but still below-normal) priority, then the main threads at a regular priority for the app to remain responsive to the user interaction with the GUI. It's worked great for W2K, WinXP, Vista, 7, and even 8 along with various flavour of each (though I didn't test much in Win 8). Now, with Win10, the GUI and in fact the entire operating system seems to be lethargic when the low priority threads are working - appearing to defeat the purpose of setting thread priorities. My threading library code hasn't changed, only the OS. And if I take the current build and run it on Win7, it behaves as expected. I've done a cursory review and can't find what's changed in the API to correct this. Did Microsoft change/break basic thread priorities in Windows 10, or am I missing something else? Thanks, --Rob
-
Hi guys, I've got an MFC program that I've been developing for, well, since Windows 2000. We heavily leverage threads in this app - multiple background worker threads at a low priority, a rendering thread at a higher (but still below-normal) priority, then the main threads at a regular priority for the app to remain responsive to the user interaction with the GUI. It's worked great for W2K, WinXP, Vista, 7, and even 8 along with various flavour of each (though I didn't test much in Win 8). Now, with Win10, the GUI and in fact the entire operating system seems to be lethargic when the low priority threads are working - appearing to defeat the purpose of setting thread priorities. My threading library code hasn't changed, only the OS. And if I take the current build and run it on Win7, it behaves as expected. I've done a cursory review and can't find what's changed in the API to correct this. Did Microsoft change/break basic thread priorities in Windows 10, or am I missing something else? Thanks, --Rob
Did you ever resolve this? I'd be interested in the solution. :)
The difficult we do right away... ...the impossible takes slightly longer.
