Re: FreeBSD 5.3b7and poor ata performance

From: fandino <fandino_at_ng.fadesa.es>
Date: Tue, 19 Oct 2004 13:39:48 +0200
Søren Schmidt wrote:
>> # dd if=/dev/ad4 of=/dev/null bs=1024k count=1024
>> 1024+0 records in
>> 1024+0 records out
>> 1073741824 bytes transferred in 31.090536 secs (34535970 bytes/sec)
>
> etc, and I get this:
> 1073741824 bytes transferred in 18.488903 secs (58074934 bytes/sec)
> 1073741824 bytes transferred in 14.956484 secs (71791059 bytes/sec)

>> I'd like to know wich is you opinion about this.
>
> If you run a stock generic kernel with the debug (WITNESS etc) taken out 
> you should see the same raw performance as I do.

I redid all tests with __RC1__ and performance is as bad as before.

> Now, raw performance is one thing, filesystem I/O something entirely

I think the problem is with raw performace, just see this example
with a test raid0. Each disk by separate has a throughput of 26000 K/sec,
with /dev/stripe/test they have a throughput of 15000 K/sec each
(raid0 throughput 30000 K/sec) and with four disks 7500 K/sec each
(30000 K/sec again).

# dd if=/dev/stripe/test of=/dev/null bs=1024k count=1024
1024+0 records in
1024+0 records out
1073741824 bytes transferred in 30.072215 secs (35705445 bytes/sec)

There is a constant limit of 30000 K/sec independent of the number
of disks used.

simply it doesn't makes any sense. With two disk I'd must have
approx. 26K+26K=52 M/sec and with four 26K+26K+26K+26K=104 M/sec

Other systems like OpenBSD had a good throughput 55000K/sec by disk

I send you in another email with all info that you need.

Regards,
Received on Tue Oct 19 2004 - 09:39:58 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:18 UTC