Messages in this thread | | | From | riel@redhat ... | Subject | [PATCH 0/3] sched,numa: further numa balancing fixes | Date | Sat, 14 Jun 2014 15:18:56 -0400 |
| |
A few more bug fixes that seem to improve convergence of "perf bench numa mem -m -0 -P 1000 -p X -t Y" for various values of X and Y, on both 4 and 8 node systems.
This does not address the issue I highlighted Friday:
https://lkml.org/lkml/2014/6/13/529
I have an idea on how to fix that issue, but implementing that as part of this series would be silly, since I would have to rip the code back out and completely rewrite it once I started working on placement for systems with complex NUMA topologies. Better to leave that code for then...
Patches are against the latest -next and -tip trees.
| |