Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 17 Apr 2005 22:23:14 -0700
From:      Mike Smith <msmith@mu.org>
To:        Mikhail Teterin <mi@corbulon.video-collage.com>
Cc:        trhodes@freebsd.org
Subject:   Re: speed of a ciss-based pseudo-disk
Message-ID:  <1CA47A90-7B13-4EFC-9E3A-4A6441353DBD@mu.org>
In-Reply-To: <200504172305.j3HN53SD098019@corbulon.video-collage.com>
References:  <200504172305.j3HN53SD098019@corbulon.video-collage.com>

next in thread | previous in thread | raw e-mail | index | archive | help

On Apr 17, 2005, at 4:05 PM, Mikhail Teterin wrote:

>> Are you asking that someone go implement a 'nominal' speed tester in
>> the kernel that will accurately determine the speed of each attached
>> storage device?
>
> Scott, I'm asking about a nominal number. If my controller can not  
> talk
> to the attached drives at U320 and negotiates down to lower speeds,  
> I'd
> like to know about it.

There's no way to obtain a "nominal" number.  The 133MHz number
comes from the maximum theoretical sustained bandwidth of a
32-bit 33MHz PCI bus.

Since there's no good way in the driver to detect the bus width, or any
way to determine the negotiated speed for a given drive, I'm afraid
you're SOL.

> If, as Paul Saab responded, even the nominal number is unobtainable  
> for
> HP/Compaq arrays -- fine. The driver's output confused me, so I asked
> about its meaning.

CAM insists on having a number.  I gave it a number:

   2587:        cpi->base_transfer_speed = 132 * 1024;  /* XXX what  
to set this to? */


  = Mike



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1CA47A90-7B13-4EFC-9E3A-4A6441353DBD>