Disk Speed

[William Roberts]

In article <AUXPOST%90040422380113 at PUCC.BITNET> MATLEVAN at EKU.BITNET (Jerry LeVan) writes:
>timex dd if=/dev/rdsk/<whole disk> of=/dev/null bs=12k count=1024
>
>The above times a 12 meg transfer from the raw disk to the null device
>Some of the times I obtained are:
>
>Apple 160 MB disk       44.08 secs              .2727 MB/Sec
>Apple  80 MB disk       58.23 secs              .2060 MB/Sec
>
>At&T 3b2/1000-70 330 MB disk    20.24 secs      .5928 MB/Sec
>DecStation 3100  330 MB disk    29.8  secs      026 MB/Sec

Sun 3/50 with SCSI 141M disk     35 secs        SunOS 3.5
Sun 3/60 with SCSI 141M disk     30 secs        SunOS 4.0
SparcStation1 SCSI 327M disk     33 secs
Sequent B21000  377M Eagle       25 secs
Sequent B21000  795M Swallow5    23 secs

>
>Do you think the above test (cribbed from a trade journal) gives
>a reasonable estimate of relative disk performance?

Ought to be OK - you'd get better actual performance out of
using larger block sizes (e.g. 60k or so). It looks to be a
little bit OS dependent, judging from the Sun 3/50 to Sun 3/60
figures, and it also probably varies according to machine load.
But I'd go for it as a broad comparison.

>Some SCSI accelerator's are starting to appear on the market.
>(Daystar and Surfboard?) Has anybody had any experience with
>these type of devices?

I tried the Maccelerate card from NatSemi - ran OK under 1.1.1
but we had compatibility problems with some drives and NatSemi
have apparently decided that it is not for retail sale any more
(i.e. it works well if you know what you are doing, so OEMs and
developers can build it into systems if they want).

The main problem is that such products are mostly trying to
second guess what MacOS disk drivers are going to do, and the
MacOS SCSI manager (prior to System 7) does so little that
there is complete anarchy! This isn't true under A/UX, but the
manufacturers haven't worried that much about the A.UX market,
and I'd guess that the IIfx product will kill off any interest
that they had...

><Sigh> I really broke down and had a
>good cry when if found out that Apple was going to have a
>MacII - Mac fx board swap for less than I paid for the DayStar
>accelerator :-(

I'll bet DayStar weren't over the moon either.
-- 

William Roberts                 ARPA: liam at cs.qmw.ac.uk
Queen Mary & Westfield College  UUCP: liam at qmw-cs.UUCP
Mile End Road                   AppleLink: UK0087
LONDON, E1 4NS, UK              Tel:  01-975 5250 (Fax: 01-980 6533)