bernhard@mime.idt.unit.no (Kjetil Bernhard Thomassen) (10/04/90)
I would like to know the difference between SCSI I and SCSI II. Is there anobody out there who has the details? Is SCSI II 32 bit, or what? What is the teorethical throughput? How many peripherals is it possible to connect to the same controller without using any kind of "undocumented features"? By sending me email, the only thing you obtain is a posted summary. Kjetil Bernhard Thomassen bernhard@idt.unit.no bernhard@solan.unit.no
ben@epmooch.UUCP (Rev. Ben A. Mesander) (10/04/90)
>In article <1990Oct4.003204.20738@idt.unit.no> bernhard@mime.idt.unit.no (Kjetil Bernhard Thomassen) writes: >I would like to know the difference between SCSI I and SCSI II. > >Is there anobody out there who has the details? > No. Nobody knows the details. :-) There is an ANSI (American National Standards Institute) document that covers SCSI-2. I'm not positive of this, but there should be a European spec. as well (There was one for SCSI-1). > >Is SCSI II 32 bit, or what? > Or what. Sometimes. SCSI-2 has 16 and 32 bit options. They require different cabling than SCSI-1, and are not commonly used (yet). SCSI-1 uses an 8 bit bus. > >What is the teorethical throughput? > Well, ethically speaking, that depends on another "or what". There are several options for SCSI-2, including something called "fast SCSI", and the 16 and 32 bit data paths. I can't remember the exact numbers right now, and I'm too sleepy to dig into my SCSI-2 manual. Suffice to say that the maximum throughput is a lot faster than real-world disk drives can sustain. > >How many peripherals is it possible to connect to the same controller >without using any kind of "undocumented features"? > The commonly used method allows 8 devices on a SCSI bus. There may be up to 2048 (I think...) sub-devices for each LUN. I've never seen this feature implemented. Anyone using "undocumented features" on the SCSI bus should be summarily executed. >Kjetil Bernhard Thomassen >bernhard@idt.unit.no >bernhard@solan.unit.no -- | ben@epmooch.UUCP (Ben Mesander) | "Cash is more important than | | ben%servalan.UUCP@uokmax.ecn.uoknor.edu | your mother." - Al Shugart, | | !chinet!uokmax!servalan!epmooch!ben | CEO, Seagate Technologies |
Smyers.S@AppleLink.Apple.COM (Scott Smyers) (10/06/90)
In article <1990Oct4.003204.20738@idt.unit.no> bernhard@mime.idt.unit.no (Kjetil Bernhard Thomassen) writes: > I would like to know the difference between SCSI I and SCSI II. I'm sure you're going to get a lot of replies to this general question, but I thought I'd start with what I know. One of the major differences between SCSI I and II is the addition of command queueing. This allows a mass storage device to receive multiple commands from the same or different initiators and execute them later. For example, during a seek operation, if disconnected, the drive may receive several additional commands and add them to its queue. There are a lot of details to this feature, but the benefits in speed and overall system throughput should be immediately clear. > Is SCSI II 32 bit, or what? SCSI II has a 16 bit and a 32 bit option, but it can still operate in 8 bit mode. There is also a higher speed timing option (called fast SCSI) which permits up to 10 megatransfers/sec. This gives SCSI II the capability to go up to 10 megabytes/sec on an 8 bit cable, or up to 40 megabytes/sec on a 32 bit cable. > How many peripherals is it possible to connect to the same controller > without using any kind of "undocumented features"? Still 8 devices total. ------------------------------ Scott Smyers @ Apple Computer Cupertino CA Apple doesn't necessarily agree with this stuff. ------------------------------ The ideas presented here are my own, not Apple's.
buck@siswat.UUCP (A. Lester Buck) (10/06/90)
In article <ben.2866@epmooch.UUCP>, ben@epmooch.UUCP (Rev. Ben A. Mesander) writes: > >What is the teorethical throughput? > > Well, ethically speaking, that depends on another "or what". There are > several options for SCSI-2, including something called "fast SCSI", and > the 16 and 32 bit data paths. I can't remember the exact numbers right > now, and I'm too sleepy to dig into my SCSI-2 manual. Suffice to say > that the maximum throughput is a lot faster than real-world disk drives > can sustain. If I remember, fast wide SCSI-2 is 40 MB/sec, with 10 M transfers/sec on a 4 byte wide bus. But this is somewhat slow in the current world of fast disks. There have been parallel transfer disks for some time time that gang all the heads together for 20-40 MB/sec. The new disk arrays being marketed by Maximum Strategies, for example, transfer at something really gross like 120 MB/sec and layer IPI-3 on HPPI packets. I think they use two HPPI channels to handle the peak rate. Ciprico had the first SCSI-2 controller on the market and its purpose was to run a RAID array. -- A. Lester Buck buck@siswat.lonestar.org ...!uhnix1!lobster!siswat!buck