* 2.4.17 raid5 checksumming function selection wierdness
@ 2002-01-11 12:12 Anders Vedmar
2002-01-11 12:17 ` David S. Miller
2002-01-11 12:26 ` Dave Jones
0 siblings, 2 replies; 3+ messages in thread
From: Anders Vedmar @ 2002-01-11 12:12 UTC (permalink / raw)
To: linux-kernel
The raid5 code in 2.4.17 seems to select the slowest available
checksumming function.
Part of dmesg output follows:
[...]
raid5: measuring checksumming speed
8regs : 2148.000 MB/sec
32regs : 1900.400 MB/sec
pIII_sse : 1272.000 MB/sec
pII_mmx : 3291.200 MB/sec
p5_mmx : 4212.400 MB/sec
raid5: using function: pIII_sse (1272.000 MB/sec)
[...]
The machine is an Athlon XP 1400MHz on an MSI K7T266Pro2 board with
512 Mb of PC2100 DDR.
--
/A
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: 2.4.17 raid5 checksumming function selection wierdness
2002-01-11 12:12 2.4.17 raid5 checksumming function selection wierdness Anders Vedmar
@ 2002-01-11 12:17 ` David S. Miller
2002-01-11 12:26 ` Dave Jones
1 sibling, 0 replies; 3+ messages in thread
From: David S. Miller @ 2002-01-11 12:17 UTC (permalink / raw)
To: anders.vedmar; +Cc: linux-kernel
From: Anders Vedmar <anders.vedmar@interactiveinstitute.se>
Date: Fri, 11 Jan 2002 13:12:53 +0100
The raid5 code in 2.4.17 seems to select the slowest available
checksumming function.
It does so because: 1) It performs better for the cache miss
case and 2) it is more cache friendly
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: 2.4.17 raid5 checksumming function selection wierdness
2002-01-11 12:12 2.4.17 raid5 checksumming function selection wierdness Anders Vedmar
2002-01-11 12:17 ` David S. Miller
@ 2002-01-11 12:26 ` Dave Jones
1 sibling, 0 replies; 3+ messages in thread
From: Dave Jones @ 2002-01-11 12:26 UTC (permalink / raw)
To: Anders Vedmar; +Cc: linux-kernel
On Fri, 11 Jan 2002, Anders Vedmar wrote:
> The raid5 code in 2.4.17 seems to select the slowest available
> checksumming function.
> 8regs : 2148.000 MB/sec
> pIII_sse : 1272.000 MB/sec
> raid5: using function: pIII_sse (1272.000 MB/sec)
SSE is favoured due to it avoiding trashing the cache contents.
--
| Dave Jones. http://www.codemonkey.org.uk
| SuSE Labs
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2002-01-11 12:27 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2002-01-11 12:12 2.4.17 raid5 checksumming function selection wierdness Anders Vedmar
2002-01-11 12:17 ` David S. Miller
2002-01-11 12:26 ` Dave Jones
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.