Re: 5.x filesystem performance regression

[Edgar Fuß <ef%math.uni-bonn.de@localhost>]

Hello again.

> Could anyone else please me[a]sure on a RAID 5?
ws@ kindly measured this (on -current) on his four-SATA-disc RAID5.
With writeback cacheing disabled, his system is as slow as mine.
With writeback cacheing enabled, it's tenfold faster (i.e. like my 4.0.1).

On my machine (using SCSI discs), the write cache hardly influences performance 
at all.

Could someone please be so kind as to benchmark 5.1 or -current with SCSI discs?

> I sincerly hope someone can make sense out of these hours and hours of 
> testing.
Anyone? This really drives me nuts.
No matter what may be broken with my setup, with writeback cacheing disabled, 
Wolfgang's -current is tenfold slower compared to my 4.0.1.