AHA2790UW has speed-limit problems ?

[Doug Ledford]

Matti Aarnio wrote:
> 
> Hello,
> 
>   Some days ago I sent a plea for fix to <linux-scsi at vger.rutgers.edu>
> list.  As it turns out, my problem seem to relate on the chassis internal
> ultra-wide cable, and to the devices attached to it.
> 
> Even the lattest beta (5.1.0-*) sequencer and associated code apparently
> can not fix marginally working hardware.  (  :-|  )
> 
> The symptoms:
> 
>   Four Ultra-Wide (SingleEnded) disks on high quality cable inside
> DEC PWS 433a workstation chassis attached to AHA2940UW host adapter.
> The host adapter has termination configured (Active LOW, active HIGH),
> and the other end of the cable has an active terminator for it.
> 
> Generating a load-set at which all disks (at least two) are actively
> written at the same time seems to cause SCSI-bus timeouts IF the channel
> is being run at the fastest mode.
> 
> The Kludge-Cure:
> 
>   Lower the speed of the device in the adapter bios configuration
> utility.  In fact lower them all, and no SCSI-timeout problems appear.
> 
> Question:
> 
>   Is there ANYTHING that is possible to do to in this type of environment
> so that I could (reliably) run faster disk-transfer speeds ?

The most common causes of this problem are either bad termination or cabling
problems.  You've already covered termination issues (although one should
never under estimate the problem either, for instance the external
terminator you have at the end of the cable could be bad, or the terminator
on the card could be bad, there could be a lack of term power on the bus for
that external terminator to use, etc).  The cable itself you said was high
quality.  This leaves one last thing I always tell people to check.  The
cable run from the card to the last device needs to be as free of twists and
turns as possible.  The more you can straighten that cable out, the better. 
In general, I find that loops don't cause any problems  but that twists are
horrible for reliability.  IOW, if you fold the cable with a sharp crease
such that pin one on the cable ends up laying against pin one on the cable,
that's OK, when there is a twist such that pin one on the cable ends up
running past or parallel to pin 68 of the cable, that's bad.

> Now my RAID0 write throughput has dropped from 12.odd MB/sec to 7.odd
> MB/sec speed.

I take it these drives aren't speed demons on their own?  FWIW, I see 25.odd
MB/sec sustained writes on a 4 drive RAID0 array here.

-- 

 Doug Ledford  <dledford at dialnet.net>
  Opinions expressed are my own, but
     they should be everybody's.

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