To be honest, that may be enough to get me to switch... not having that in GL/Vulkan makes them almost unusable for serious MT.
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
No app-side oversubscription.
-
(my preference would be post-to-IO-completion-port, really, but since Win32 seems to have given up on them, oh well :( )
End of conversation
New conversation -
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Metal also has you to wait to standard OS sync primitives & have a callback from Metal allow you to unlock the mutex/barrier/sema
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Having GL-specific wait calls means you can't wait on anything else, hence that thread is immediately oversubscribed.
-
You want all your threads to be able to wait on any available work, so you don't have oversubscribed threads sitting around.
- Show replies
New conversation -
-
-
_on what_? That's the whole problem. I need to know when I can start reusing the memory from the sync'd task.
-
So you want to do "wait on any IO completion and any GL completion" and service whichever happens first.
- Show replies
New conversation -
-
-
Because you need to oversubscribe to that! That's the whole point.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.