aic7895, IBM 9ZX, Quantum Viking, slow transfer rates

[Bjorn T Johansen]

I was just wondering what is the most optimal depth on tagged queueing?
And where can I find the benchmark program you use?

Regards,

BTJ

>On Wed, 17 Mar 1999, Karl-Heinz Herrmann wrote:
>
>I just recompiled my kernel to automatically enable tagged queueing with a
>depth of 8. This seems to improve speeds on the 9ZX to something more like
>what I was expecting, but slowing down the Viking slightly (which I gather
>means that I should disable tagged queueing on that device). I would like
>to thank everyone for all of their help. (Although I'd still be very
>interested if anyone had a 9ZX that they could try the same benchmark on,
>just for comparison purposes). 
>
>Here's what I got after enabling tagged queueing:
>IBM 9ZX
>             -------Sequential Output-------- ---Sequential Input----Random--
>             -Per Char- --Block--- -Rewrite-- -Per Char- --Block-----Seeks---
>Machine   MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU  /sec
%CPU
>         512  4944 76.4  9916 16.2  4007 13.3  6170 85.9  9644 13.5  99.9 3.3
>
>Quantum Viking
>             -------Sequential Output-------- ---Sequential Input----Random--
>             -Per Char- --Block--- -Rewrite-- -Per Char----Block-----Seeks---
>Machine   MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU  /sec
%CPU
>         512  4117 63.7  7044 11.8  2877  9.5  5938 82.8  7459 10.9  81.4 2.8
>
>
>> 
>> On 17-Mar-99 Albert Max Lai wrote:
>> >  Tag Queue Enable Flags: 0x0000
>> > Ordered Queue Tag Flags: 0x0000
>> > Default Tag Queue Depth: 8
>> >     Tagged Queue By Device array for aic7xxx host instance 0:
>> >      
>> > {255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255}
>> >     Actual queue depth per device for aic7xxx host instance 0:
>> >       {1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}
>> 
>> Hallo Albert,
>> 
>> you have tagged queuing disabled (Actual depth 1 on all targets) and
>> Tag Queue Enable Flags: 0x0000.
>> So your drive is working in dumb mode one command at a time without any
>> optimisation.
>> 
>> put some bootflag (see somewhere in the aicxxx doc) in lilo.conf.
>> Sorry I'm not at my Linux-Box right now so I can't look it up.
>> It's something like tag_queue{{0,0,0,0,,,,}} where 0 means "use
>> default".
>> 
>> If you enable it (-> Tag Queue Enable Flags: 0x[nonzero]) and the
>> actual depth is still 1 you can have a look with scsiinfo if the drive
>> has tagged queuing switched on.
>> 
>> regards,
>> 
>> K.-H.
>> 
>> 
>> 
>> 
>> 
>> ----------------------------------------------------------
>> E-Mail: Karl-Heinz Herrmann <k.-h.herrmann at fz-juelich.de>
>> Date: 17-Mar-99      Time: 15:20:59
>> ----------------------------------------------------------
>> 
>> 
>> To Unsubscribe: send mail to majordomo at FreeBSD.org
>> with "unsubscribe aic7xxx" in the body of the message
>> 
>
>--
>Albert Lai <amlai at columbia.edu>           1018D East Campus
>Residential Computer Consultant           411 W. 116th St.
>Columbia University                       New York, NY 10027
>http://www.columbia.edu/~aml61            (212)853-4854
>
>
>
>
>
>
>
>
>To Unsubscribe: send mail to majordomo at FreeBSD.org
>with "unsubscribe aic7xxx" in the body of the message
>
>
_______________________________________________________________________
Bjørn T Johansen (BSc,MNIF,MCP)	Dept. of Customized Solutions
Senior Systems Engineer		Software Innovation asa
bjorntj at softinn.no			Drammensveien 175
Phone	: +47 22 51 85 09		P.O.Box 390 Skøyen
Fax	: +47 22 73 07 14		N-0212 Oslo, Norway
Cellular: +47 926 93 298		http://www.softinn.no
________________________________________________________________________
"The stickers on the side of the box said "Supported Platforms: Windows 95,
Windows NT 4.0, or better", so clearly Linux was a supported platform."


To Unsubscribe: send mail to majordomo at FreeBSD.org
with "unsubscribe aic7xxx" in the body of the message