freebsd stable - May 2003 - bad performance on ATA promise controllers

Hello friends,

 I'm having a problem with my home server (ASUS A7V133 motherboard) which
has
a horrible performance with ATA disks connected to an integrated promise
controller.

Below you can see iozone results of the same disk connected to the primary/
secondary controller versus the promise one.

Promise ATA100 controller:
========================Version 1.02a       ------Sequential Output------
--Sequential Input- --Random-
                    -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine        Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP  /sec %CP
damocles         1G  7285  18  7235   3  5198   3 18830  58 19332   7 180.5   0


VIA 82C686 ATA100 controller:
===========================Version 1.02a       ------Sequential Output------
--Sequential Input- --Random-
                    -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine        Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP  /sec %CP
damocles         1G 24301  63 25047  11 13182   8 29843  85 30678  12 185.1   1

There is a big diference, 7MB/s connected in the promise and 24MB/s in the 686
and using UltraDMA-100 in both cases.
Transfers between ATA devices are even worse, certain combinations give me
transfer rates as low as 1MB/s :-?
I tried with other disks/ata cables and the results where identical.

If I boot GNU/Linux in the same machine all disks work as expected in
the promise controller, so this make me think that it's a specific problem
with FreeBSD.

Until now, I haven't had any luck searching in the mailing list archives
about this problem so I'm trying to asking here. Does anyone know where the
problem could be?

Thanks

** dmesg: 
http://www.fadesa.es/misc/promise.txt
-- 
-----BEGIN GEEK CODE BLOCK-----
Version: 3.1
GCS/IT d- s+:+() a- C+++ UBL+++$ P+ L+++ E--- W++ N+ o++ K- w---
O+ M+ V- PS+ PE+ Y++ PGP+>+++ t+ 5 X+$ R- tv-- b+++ DI D++>+++
G++ e- h+(++) !r !z
------END GEEK CODE BLOCK------

"Jim" wrote:
> I've had uglyness between the earlier via chipsets and the promise
ata-100 set (data corruption of large files during dma mode transfers).  Your
problem almost seems like the promise adapter is falling back to one of the PIO
modes under freebsd.
Hello Jim,

CPU was 90% idle during tests, so I suppose
that disks weren't in PIO mode.

regards,
> -----Original Message-----
> From: Jos? M. Fandi?o [mailto:freebsd@fadesa.es]
> Sent: Monday, May 26, 2003 10:14 AM
> To: stable@freebsd.org
> Subject: bad performance on ATA promise controllers
> 
> Hello friends,
> 
>  I'm having a problem with my home server (ASUS A7V133 motherboard)
which has
> a horrible performance with ATA disks connected to an integrated promise
> controller.
> 
> Below you can see iozone results of the same disk connected to the primary/
> secondary controller versus the promise one.
> 
> Promise ATA100 controller:
> ========================> Version 1.02a       ------Sequential
Output------ --Sequential Input- --Random-
>                     -Per Chr- --Block-- -Rewrite- -Per Chr- --Block--
--Seeks--
> Machine        Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP  /sec
%CP
> damocles         1G  7285  18  7235   3  5198   3 18830  58 19332   7 180.5
0
> 
> VIA 82C686 ATA100 controller:
> ===========================> Version 1.02a       ------Sequential
Output------ --Sequential Input- --Random-
>                     -Per Chr- --Block-- -Rewrite- -Per Chr- --Block--
--Seeks--
> Machine        Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP  /sec
%CP
> damocles         1G 24301  63 25047  11 13182   8 29843  85 30678  12 185.1
1
> 
> There is a big diference, 7MB/s connected in the promise and 24MB/s in the
686
> and using UltraDMA-100 in both cases.
> Transfers between ATA devices are even worse, certain combinations give me
> transfer rates as low as 1MB/s :-?
> I tried with other disks/ata cables and the results where identical.
> 
> If I boot GNU/Linux in the same machine all disks work as expected in
> the promise controller, so this make me think that it's a specific
problem
> with FreeBSD.
> 
> Until now, I haven't had any luck searching in the mailing list
archives
> about this problem so I'm trying to asking here. Does anyone know where
the
> problem could be?
> 
> Thanks
> 
> ** dmesg:
> http://www.fadesa.es/misc/promise.txt
-- 
-----BEGIN GEEK CODE BLOCK-----
Version: 3.1
GCS/IT d- s+:+() a- C+++ UBL+++$ P+ L+++ E--- W++ N+ o++ K- w---
O+ M+ V- PS+ PE+ Y++ PGP+>+++ t+ 5 X+$ R- tv-- b+++ DI D++>+++
G++ e- h+(++) !r !z
------END GEEK CODE BLOCK------