Re: [CFT] delayed allocation and multipage I/O patches for 2.5.6.

2.5.6 with Andrew's everything patch on ext2 
filesystem mounted with delalloc came up with
these MB/second on k6-2/475 with IDE disk:

dbench 128	
2.5.6		2.5.6-akpme	akpm % faster
8.4 		12.5 		48

tiobench seq reads (8 - 128 threads avg 3 runs)
2.5.6		2.5.6-akpme	%
9.36 		12.97		38

tiobench seq writes (8 - 128 threads avg 3 runs)
2.5.6		2.5.6-akpme	%
15.3		19.29		26

Both kernels needed reiserfs patches for 2.5.6, but
above tests are on ext2.  

More on these tests and a few other akpm patches at:
http://home.earthlink.net/~rwhron/kernel/akpm.html

-- 
Randy Hron