If you’re doing I/O you’re entering the kernel anyway. Besides, syscalls themselves are extremely fast (page 16). See benchmarks: https://blog.linuxplumbersconf.org/2013/ocw/system/presentations/1653/original/LPC%20-%20User%20Threading.pdf …
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
I'm trying to remember if 2005 was before, after, or during the time when Solaris and NetBSD tried to do scheduler activations and found it to be a nice idea that's seeply cursed to actually implement & went to 1:1.
-
“deeply cursed”, even. And to give an idea of how cursed, NetBSD was actually willing to break backward compatibility to get rid of it. (IIRC it was unusably crashy unless restricted to a single CPU right up to the end, so it didn't get much use, but still.)
- 1 more reply
New conversation -
-
-
Kernel entry occurs much more often for syscall reasons (I/O, etc.) than it does for rescheduling reasons most of the time, unless you have near-max CPU usage. If Meltdown/Spectre are the reasons for your worry, buy a Ryzen.
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.