Messages in this thread | | | Subject | Re: REGRESSION: Performance regressions from switching anon_vma->lock to mutex | From | Tim Chen <> | Date | Thu, 16 Jun 2011 13:26:12 -0700 |
| |
On Wed, 2011-06-15 at 18:50 -0700, Linus Torvalds wrote:
> > Tim, have you tried running your bigger load with that patch? You > could try my patch on top too just to match Peter's tree, but I doubt > that's the big first-order issue. > > Linus
I ran exim with different kernel versions. Using 2.6.39-vanilla kernel as a baseline, the results are as follow:
Throughput 2.6.39(vanilla) 100.0% 2.6.39+ra-patch 166.7% (+66.7%) (note: tmpfs readahead patchset is merged in 3.0-rc2) 3.0-rc2(vanilla) 68.0% (-32%) 3.0-rc2+linus 115.7% (+15.7%) 3.0-rc2+linus+softirq 86.2% (-17.3%)
So Linus' patch certainly helped things over vanilla 3.0-rc2, but throughput is still less than the 2.6.39 with the readahead patch set. The softirq patch I used was from Ingo's combined patch from Shaohua and Paul. It seems odd that it makes things worse. I will recheck this data probably just this patch and without Linus' patch later.
I also notice that the run to run variations have increased quite a bit for 3.0-rc2. I'm using 6 runs per kernel. Perhaps a side effect of converting the anon_vma->lock to mutex?
(Max-Min)/avg 2.6.39(vanilla) 3% 2.6.39+ra-patch 3% 3.0-rc2(vanilla) 20% 3.0-rc2+linus 36% 3.0-rc2+linus+softirq 40%
Thanks.
Tim
| |