Path: utzoo!attcan!uunet!decwrl!sgi!shinobu!odin!anchor!olson
From: olson@anchor.esd.sgi.com (Dave Olson)
Newsgroups: comp.periphs.scsi
Subject: Re: More advice sought for r/w optical system
Message-ID: <1991Feb28.055757.11038@odin.corp.sgi.com>
Date: 28 Feb 91 05:57:57 GMT
References: <1991Feb13.134708.3440@news.larc.nasa.gov> <3721@island.COM> 	<1991Feb25.045950.17710@odin.corp.sgi.com> <HARAKI.91Feb27105332@fermi.rc.m-kasei.co.jp>
Sender: news@odin.corp.sgi.com (Net News)
Organization: Silicon Graphics, Inc.  Mountain View, CA
Lines: 54

In <HARAKI.91Feb27105332@fermi.rc.m-kasei.co.jp> haraki@fermi.rc.m-kasei.co.jp (Susumu Haraki 3844) writes:
|   In article <1991Feb25.045950.17710@odin.corp.sgi.com> olson@anchor.esd.sgi.com (Dave Olson) writes:
|  <I don't know where you got your info, but even the slowest M-O
|  <drive I saw (an old sony) wrote at 100Kbytes/sec WITHOUT pre-erasing.
|     On DOS environment,Drives must updates FAT and directry.
|     And most of case,they are located most inner area.
|     Do you know the seeking speed of MO drives?
|     In addition to, an old sony drive are as fast as New one.

I didn't see any mention of DOS, I was talking about real computers
and OS's :-).  The very first sony c501 I had for eval was definitely
slower than newer ones; most likely a firmware fix.

|  <Current Sony, Ricoh, and Maxtor M-O's all write at 250-300Kbytes/s
|  <(at least) without pre-erasure.
|     Definitely NO!!
|     Because most of drives use three passes method in writing mode.(Except
|     old Sony drives[NWP-539S 2.05] )
|    I don't check New Sony drive yet ;-<
|      (Erase,Write & Verify)

I can only tell you that on standard out of the box SGI 4D25's, running
IRIX 3.3, that I have measured these values, both through the
filesystem and direct to the disk, both with software timers, and SCSI
analyzers.  You may not like it, but that is life.  Zcav on the Tahiti
helps, but not all that much (~ 10% improvement, by vague
recollection), since the current Tahiti's spin at the same speed
(2200?) for both media types (older ones spun at 1800 for non-ZCAV)

My tests were typically writing and reading 16-32 Mbytes, with
i/o sizes from 8K to 256Kbytes.  The 600Kb read/250Kb write
is at large block sizes (128Kb), and it does drop off at lower
block sizes.  As I recall, it was down to 400 Kb read, 150 Kb write
at 16Kbytes per i/o, but this was some time ago, and I don't have
the numbers handy at the moment.

|  <The next generation of drives
|  <will not require erase before write, so write speeds should be 
|  <500-600 Kbytes/sec or more, which matches current read speed.
|   Yes. But including seeking time, the speed will be less than 500Kbytes/sec.
|   The heads of drives are too heavy to move quickly.

True, seeking is somewhat slow, but even the current Tahiti is 35ms
avg 1/3 stroke seek.  Also, remember that reasonable designed
OS's have file systems that can write large contiguous files,
and therefore (for these types of files) seek time is not much
of an issue.  Using them as the main disk in a multi-user
system would not be appropriate...

--

	Dave Olson

Life would be so much easier if we could just look at the source code.