Path: utzoo!mnetor!tmsoft!torsqnt!hybrid!scifi!bywater!uunet!brunix!rca
From: rca@cs.brown.edu (Ronald C.F. Antony)
Newsgroups: comp.sys.next
Subject: Re: Build an external disk as boot disk with 2.0
Message-ID: <64594@brunix.UUCP>
Date: 12 Feb 91 06:48:14 GMT
References: <1991Feb7.225634.28188@news.cs.indiana.edu> <16112@milton.u.washington.edu> <1991Feb9.152316.22061@news.cs.indiana.edu>
Sender: news@brunix.UUCP
Reply-To: rca@cs.brown.edu (Ronald C.F. Antony)
Organization: Brown University Department of Computer Science
Lines: 46

In article <1991Feb9.152316.22061@news.cs.indiana.edu> stung@greatwhite.cs.indiana.edu (Sho-Huan Simon Tung) writes:
>In article <16112@milton.u.washington.edu> wjs@milton.u.washington.edu (William Jon Shipley) writes:
>>~$1410 and has a five year warranty.  According to the drive manual,
>>jumpers 1-2 allow you to set the drive in SCSI-2 mode, enabling
>>synchronous transfer at up to 4.8 MB/s.  (This is the same transfer rate
>>Kind of beats the pant off of the 700 KB/s it runs normally, and the
>>900 KB/s Wrens get.
>I have tried to put back the jumpers I removed from CNH2 and it works.
>However, I only get a small improvement on the transfer rate compared
>to the 771374 bytes/s posted by Izumi with an 030. May be it is because that
>my drive was build with asynchronous transfer mode on or may be the cable
>is too long. Anyone knows the reason?
>disk> read
>starting block? 0
># sectors per transfer? 16
>number of transfers? 1000
>sector increment? 16
>16384000 bytes in 19477 ms = 841499 bytes/s

The reason lies as far as I know somewhere else:

There is a huge difference between the transfer rate that the device
can handle to and from the SCSI bus, and the transfer rate to and from
the disk.

The former comes only to play when you dump the drives cache or
something like that. For everything else the drive remains the
bottleneck.

At 54 sectors per track at 512 bytes, i.e. 27 kbytes and 3600 rpm you
can get 27*3600/60 = 1620/1024 =1.58MB per second theoretical transfer
rate as a maximum. This implies that there are no reallocated blocks,
and continous reads, no seeks etc. Every seek however has a heavy
impact on this dream figure.

As long as the drive controlers do no parallel reads from multiple
heads or the rpm figure is increased insanely, there is not too much
room for improvement.

Ronald


------------------------------------------------------------------------------
"The reasonable man adapts himself to the world; the unreasonable one persists
in trying to adapt the world to himself. Therefore all progress depends on the
unreasonable man."   G.B. Shaw   |  rca@cs.brown.edu or antony@browncog.bitnet