Andrew Morton wrote:

I took the various patches in -mm for a quick ride.  Dual CPU, mem=64m,
`time make -j4 vmlinux':


2.4.25						2:57.34 2:45.62

up to blk_congestion_wait-return-remaining	5:41.52	5:56.37
up to vmscan-remove-priority
up to kswapd-throttling-fixes			7:44.53
up to vm-dont-rotate-active-list		6:29.23
up to vm-dont-rotate-active-list-padding
up to vm-lru-info				9:28.47 6:14.70 5:11.99
up to vm-shrink-zone
up to vm-shrink-zone-div-by-0-fix		9:13.21 8:17.29
up to vm-tune-throttle				7:39.89
up to shrink_slab-for-all-zones			7:06.27
up to zone-balancing-fix			7:46.15
up to zone-balancing-batching
up to zone-balancing-batching-fix		4:44.76 4:27.02 4:05.56 4:31.66 4:06.76


Based on this, and on your totally contradictory benchmarking, all I'm
prepared to say is that kswapd-throttling-fixes slows things down, and we
don't know why.  The rest appears to be worth zilch.  Possibly they slow
some things down as much as they speed other things up.

My machine doesn't touch swap at make -j4 with mem=64m. It is
dual CPU with a SMP kernel but I was using maxcpus=1.

It compiles 2.4.21 with gcc-3.3.3 I think (I can tell you when I
get home).

I can't explain your results. Maybe you have other stuff running.


--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>