Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 05 Oct 1998 13:26:46 -0700
From:      Greg Hutchison <gregh@nw.imagestat.com>
To:        Michael Kunze <mkunze@ivw.de>
Cc:        "aic7xxx@FreeBSD.ORG" <aic7xxx@FreeBSD.ORG>
Subject:   Re: Ultra2 transfer mode?
Message-ID:  <36192B86.ED972E99@nw.imagestat.com>
References:  <36190ABA.C835F0B8@ivw.de>

next in thread | previous in thread | raw e-mail | index | archive | help
Michael Kunze wrote:
> 
> Hi all,
> 
> pre 15 works for me without problems on an Asus P2B-LS. There is still a
> question left i asked some mails ago. If i take a look into
> /proc/scsi/aic7xxx/0, i see
> 
> Statistics:
> (scsi0:0:0:0)
> Device using Wide/Async transfers.
>                          ^ ^ ^ ^
> Device Negotiation Settings
> Period Offset Bus Width
> User       000    000        1
> Goal       000    000        1
> Current    000    000        1
> Total transfers 3595 (3370 read;225 written)
> blks(512) rd=16267; blks(512) wr=654
> 
> This should be definitely synchronous, because the harddisk is an IBM
> DGHS09V Ultra2-SCSI, which of course is capable of ultra2 synchronous
> transfers.
> 
> So, what is the REAL transfer mode? 80 Mbyte/s synchronous or 20 MByte/s
> or
> less asynchronous?
> 

less asynch :(

I just sorted this out here this morning on three separate systems with
pre-15 and IBM Ultrastar drives. 
Drives are variously FastWide (Ultrastar XP and 2XP) and Ultra Fast Wide
(Ultrastar 2XP and Ultrastar 9ZX).

Two of the systems are ASUS P2B-DS's with onboard AIC7890 controllers,
the other uses an onboard AIC7880.

In all case with pre-15, the IBM drives were reporting as Wide/Async
where other drives (Quantum Atlas) were reporting Wide/Sync.

Jumpering all the Ultrastars to 'Disable Target Initiated Synchronous
Negotiation', (forcing the initiator to negotiate for Synchronous and/or
wide transfers), corrected the behavior of the pre-15 driver - all
drives are now reported as Wide/Synchronous at 20Mb/s or 40Mb/s as
appropriate.

Earlier revisions of the driver behaved differently. Some negotiated
sync/wide UNLESS tagged queuing was turned on. 

Tests with Bonnie show that thruput more than doubled with the,
(shouldn't be necessary), jumper change.

Greg H.

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



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?36192B86.ED972E99>