Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 20 Feb 1999 16:45:19 -0700 (MST)
From:      "Kenneth D. Merry" <ken@plutotech.com>
To:        allenc@verinet.com (Allen Campbell)
Cc:        stable@FreeBSD.ORG
Subject:   Re: Tagged queueing
Message-ID:  <199902202345.QAA15054@panzer.plutotech.com>
In-Reply-To: <36CF45B2.14AE3598@verinet.com> from Allen Campbell at "Feb 20, 1999  4:30:58 pm"

next in thread | previous in thread | raw e-mail | index | archive | help
Allen Campbell wrote...
> 1.5MB/sec with tagged queuing (not queueing) turned on?  Hmmm...
> The following is from an Asus P97S with a P2/233:
> 
> uname -rm:
> 
> 2.2.7-STABLE i386
> (no CAM, btw)
> 
> Bit of dmesg (with AHC_TAGENABLE):
> 
> ahc0: aic7880 Wide Channel, SCSI Id=7, 16 SCBs
> ahc0: Resetting Channel A
> ahc0: Downloading Sequencer Program...ahc0: 374 instructions downloaded
> Done
> ahc0: Probing channel A
> ahc0 waiting for scsi devices to settle
> ahc0: target 0 using 16Bit transfers
> ahc0: target 0 synchronous at 20.0MHz, offset = 0x8
> ahc0: target 0 Tagged Queuing Device
> (ahc0:0:0): "WDIGTL ENTERPRISE 1.91" type 0 fixed SCSI 2
> sd0(ahc0:0:0): Direct-Access 4157MB (8515173 512 byte sectors)
> sd0(ahc0:0:0): with 5720 cyls, 8 heads, and an average 186 sectors/track
> ahc0: target 1 synchronous at 20.0MHz, offset = 0xf
> 
> bonnie results with and without AHC_TAGENABLE:
> 
>               -------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
> TAG-Q-ON  100  4936 37.6  4548  8.4  2274  6.2  8288 66.1  7973 13.2 140.8  2.9
> TAG-Q-OFF 100  3986 30.5  3986  7.1  2410  5.3  8091 64.3  8001  9.2 125.7  2.1
> 
> About the only loser with AHC_TAGENABLE is Block Sequential Input.
> To be honest, that makes me wonder more about the SCSI driver than
> the drive firmware.
> 
> How was this 1.5MB/sec measured?  I'd like to duplicate the test.
> 
> Ruslan's drive reported 'WDIGTL ENTERPRISE 1.80', as opposed to
> the 1.91 from my dmesg output.  Perhaps the newer firmware has
> something to do with this.
> 
> Sorry for the long lines.  I didn't wrap parts of this to keep
> the bonnie output straight.

It's possible they have fixed the problem in newer firmware revisions.  I
don't have any Western Digital drives, so I can't say for sure.

It's also possible that the drive functions okay with only 4 tags, but not
with as many as it can handle.  The old SCSI layer would only do 4
transactions at a time, but CAM will queue as many transactions to the
drive as it can handle.

I would suggest mailing Andrew Gallatin <gallatin@cs.duke.edu>.  He
originally submitted the bug report.  Y'all may be able to figure out which
versions of Western Digital firmware are broken and which ones work
properly.


Ken
-- 
Kenneth Merry
ken@plutotech.com


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




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199902202345.QAA15054>