7.1-PRERELEASE: arcmsr write performance problem
Jan Mikkelsen
janm at transactionware.com
Tue Dec 2 20:05:55 PST 2008
David Kelly wrote:
>
> On Dec 1, 2008, at 11:45 PM, Jan Mikkelsen wrote:
>
>> Replying to my own post ...
>>
>> I have done a test on the same machine comparing 6.3-p1 to 7.1-PRE.
>> The performance is the expected ~6MB/s (because of the lack of cache)
>> on 6.3-p1, so the BIOS change doesn't seem to be at fault.
>>
>> This seems to be a regression somewhere between 6.3 to 7.1. The Areca
>> driver is the same in 6.3 and 7.1, so the problem seems to be elsewhere.
>>
>> I think this is more than just a "performance" problem. The
>> observations with gstat showing extremely high ms/w values (I have
>> seen them as high as 22000) makes it look like IO completion
>> interrupts are being lost.
>>
>> Any suggestions on where to look next? Are there obvious candidates?
>
>
> ATA maximum block transfer has dropped from 128k to 64k in 7.x. Am not
> sure where the handle is to tweak it back up but has slowed peak thruput
> on my Dell PE400SC. Can watch with "systat -v"
Interesting, thanks.
> Worse, I have a stripped array of 2 drives that won't transfer more than
> 43k at a chunk because apparently the stripe metadata didn't align
> nicely on 64k multiples.
I know the partitions start at 64kB multiples in my case. I did,
however, reduce the RAID-6 stripe size to 4kB (from 64kB) and that
improved things slightly, but the throughput on 7.1 is still slower by
about a factor of a about 6 (measured by untaring ~3GB across ~167k
files into a fresh filesystem). And on 7.1 the machine is unusable
during much of the time.
> --
> David Kelly N4HHE, dkelly at HiWAAY.net
> ========================================================================
> Whom computers would destroy, they must first drive mad.
More information about the freebsd-stable
mailing list