home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.os.linux
- Path: sparky!uunet!boulder!ophelia!drew
- From: drew@ophelia.cs.colorado.edu (Drew Eckhardt)
- Subject: Re: help selecting SCSI hardware.
- Message-ID: <1992Aug23.202902.14772@colorado.edu>
- Keywords: SCSI hardware linux
- Sender: news@colorado.edu (The Daily Planet)
- Nntp-Posting-Host: ophelia.cs.colorado.edu
- Organization: University of Colorado at Boulder
- References: <1992Aug23.172424.10606@m.cs.uiuc.edu>
- Date: Sun, 23 Aug 1992 20:29:02 GMT
- Lines: 66
-
- In article <1992Aug23.172424.10606@m.cs.uiuc.edu> najem@milton.cs.uiuc.edu (Zeyad Najem) writes:
- >
- >I'm about to buy the following three items. Linux will be the main OS
- >installed on my system (I still need DOS from time to time). I would
- >appreciate any comments regarding the selected hardware and/or prices.
- >The major use of the system will be LISP/Prolog/TeX/Postscript.
- >
- >
- > o 520MB Fujitsu 3.5" 12ms HD $1099.00
- > o Always SCSI controller $ 179.00
-
- BAD choice. Currently, the Always controllers are NOT supported,
- and according to the always owners who have looked at writing
- an appropriate low level driver, Always is not real good about
- providing information, so it is unlikely that one will ever
- be written.
-
- > o Archive Viber 250MB Tape drive $ 499.00
- >
- >
- >The current specs of the system is:
- >
- > o 486/33 ISA
- > o 16MB RAM
- > o 170MB IDE Hard drives.
- >
- >
- >My specific questions are:
- >
- > 1) Is the Always controller a good choice ?
-
- No, it doesn't work now, and it will probably never work under Linux.
-
- >
-
- > 2) Is there any inheritied limitations in the controller/tape drive
- > that prohibits linux from supporting it in the future ? If enough
- > material is avaialable, I can try writing a driver myself.
-
- No, SCSI tapes are essentially all the same, just like SCSI disks.
-
- > 3) Do SCSI and IDE coexist ?
-
- Yes.
-
- You should also know that right now there is a performance problem in
- the stock SCSI drivers, where we're reading/writing one block per revolution
- on unbuffered drives, and aren't explicitly turning on the buffer in the
- code page on buffered drives. The net effect is a transfer rate of
- one block per revolution, which means about 60K/sec on most drives.
-
- People have gotten up to ~300K/sec on buffered drives where the buffer
- is working.
-
- This problem will be solved when the SCSI rewrite is finished.
-
- The code is about a week of work away from testing at most, the same state it's
- been in since spring - I do less "real" work during the school year, but
- also have classes getting in the way - so ya pays your money and ya takes your
- chances.
-
- --
- Microsoft is responsible for propogating the evils it calls DOS and Windows,
- IBM for AIX (appropriately called Aches by those having to administer it), but
- marketing's sins don't come close to those of legal departments.
- Boycott AT&T for their absurd anti-BSDI lawsuit.
-