On Wed 12-08-15 11:45:26, Mel Gorman wrote: [...] > 4-node machine stutter > 4-node machine stutter > 4.2.0-rc1 4.2.0-rc1 > vanilla nozlc-v1r20 > Min mmap 53.9902 ( 0.00%) 49.3629 ( 8.57%) > 1st-qrtle mmap 54.6776 ( 0.00%) 54.1201 ( 1.02%) > 2nd-qrtle mmap 54.9242 ( 0.00%) 54.5961 ( 0.60%) > 3rd-qrtle mmap 55.1817 ( 0.00%) 54.9338 ( 0.45%) > Max-90% mmap 55.3952 ( 0.00%) 55.3929 ( 0.00%) > Max-93% mmap 55.4766 ( 0.00%) 57.5712 ( -3.78%) > Max-95% mmap 55.5522 ( 0.00%) 57.8376 ( -4.11%) > Max-99% mmap 55.7938 ( 0.00%) 63.6180 (-14.02%) > Max mmap 6344.0292 ( 0.00%) 67.2477 ( 98.94%) > Mean mmap 57.3732 ( 0.00%) 54.5680 ( 4.89%) Do you have data for other leads? Because the reclaim counters look quite discouraging to be honest. > 4.1.0 4.1.0 > vanilla nozlc-v1r4 > Swap Ins 838 502 > Swap Outs 1149395 2622895 Twice as much swapouts is a lot. > DMA32 allocs 17839113 15863747 > Normal allocs 129045707 137847920 > Direct pages scanned 4070089 29046893 7x more scanns by direct reclaim also sounds bad. > Kswapd pages scanned 17147837 17140694 while kswapd is doing the same amount of work so we are moving considerable amount of reclaim activity into the direct reclaim > Kswapd pages reclaimed 17146691 17139601 > Direct pages reclaimed 1888879 4886630 > Kswapd efficiency 99% 99% > Kswapd velocity 17523.721 17518.928 > Direct efficiency 46% 16% which is just a wasted effort because the efficiency is really poor. Is this the effect of hammering a single zone which would be skipped otherwise while the allocation would succed from another zone? The latencies were not very much higher to match these numbers though. Is it possible that other parts of the benchmark suffered? The benchmark has measured only mmap part AFAIU. > Direct velocity 4159.306 29687.854 > Percentage direct scans 19% 62% > Page writes by reclaim 1149395.000 2622895.000 > Page writes file 0 0 > Page writes anon 1149395 2622895 > > The direct page scan and reclaim rates are noticeable. It is possible > this will not be a universal win on all workloads but cycling through > zonelists waiting for zlc->last_full_zap to expire is not the right > decision. As much as I would like to see zlc go it seems that it won't be that easy without regressing some loads. Or the numbers -- Michal Hocko SUSE Labs -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>