VRQ 0.96b is released with the following changes
1. smt sensitive scheduling v0.2, which fix issue with NR_CPUS > real cpu cores
Thank all for testing and finally bring SMT sensitive scheduling feature to VRQ. Feel free to test this release comparing to previous and enjoy the pleasure of unlocking the SMT capability, :)
I'm planing further improvement for SMT sensitive scheduling. Currently, I am not happy with implement code, there are duplicated code with migration code, and others. And it's late in this kernel release and time to look at sync-up works in 4.12 mainline.
Enjoy VRQ 0.96b for
v4.11 kernel, and unlock your SMT cpu ability with VRQ, :)
code are available at
https://bitbucket.org/alfredchen/linux-gc/commits/branch/linux-4.11.y-vrq
and also
https://github.com/cchalpha/linux-gc/commits/linux-4.1.y-vrq
All-in-one patch is available too.
BR Alfred
Thanks Alfred for this release.
ReplyDeleteI've done my usual throughput tests. The results are here :
https://docs.google.com/spreadsheets/d/163U3H-gnVeGopMrHiJLeEY1b7XlvND2yoceKbOvQRm4/edit?usp=sharing
VRQ 0.96b has the best overall results.
This release also fix the bad results under half cpu load (ex. lame x4).
Well done !
Pedro
@Pedro
DeleteThanks for your test result.
There is regression comparing to 095b using single core. I'll see if any further improvement can be make for this.
@Alfred,
ReplyDeleteBack from vacation, it was pretty good :)
Compiled fine (unlike 4.12) + 1 hour usage + one kernel build, no problems so far.
Will use this as my main kernel on my Ryzen build. Let's see how it fares in day to day usage + games.
Great job Alfred ;)
Br, Eduardo