Article View: comp.sys.next.hardware
Article #599Re: Terrible SCSI Transfer
From: M Carling
Date: Tue, 01 Sep 1992 01:05
Date: Tue, 01 Sep 1992 01:05
47 lines
2556 bytes
2556 bytes
In article <1992Aug31.112327.2138@osf.org> david@postman.gr.osf.org (David
George) writes:
> NeXT uses old NCR 5390 chips for doing SCSI.
Actually, NeXT used those only in the earliest 030 cubes. All other NeXTs
use the NCR 53C90A which is basically the same part as the Emulex ESP100A
SCSI protocol chip found in many SparcStations.
> These generate a lot of interrupts for each SCSI transaction (>2).
True.
> The NeXT would benchmark worse against machines using the NCR 537x0
chips or even machines with intelligent host adapaters built around the
5390 (for example).
Not necessarily. There is more to a SCSI subsystem than its protocol chip.
The reason why benchmarks are used is so that men will have a more
reliable basis for comparison than speculation. Most benchmarks are less
than perfect, so benchmarking doesn't eliminate speculation; it only
reduces it because one is still speculating about the acuracy of the
benchmark. Speculation about benchmarks is doubly speculative.
As I've posted before, the way to do this is to get identical disks (or
use the same one twice), devise a good benchmark and actually try it on
the systems of interest. Benchmark systems, not datasheets!
> The 5390 doesn't support SCSI2 and is limited in some other areas.
The 53C90A corrected several deficiencies of the 53C90 and also added
support for some (optional) SCSI-2 features. The 53C90A *does* support
SCSI-2 -- depending on how one uses it. Since these chips are
"low-level" protocol controllers, the implementation the majority of
the SCSI-2 protocol is the responsibility of the software driving the
part, not the part itself.
> The software does a pretty good job under these circumstances.
Agreed.
M Carling
Director, Bay Area NeXT Group
--
"High tech is potent, precise, and in the end, unbeatable. The truth is,
it reminds a lot of people of the way I pitch horseshoes. Would you
believe some of the people? Would you believe our dog? Look, I want to
give the high-five signal to high tech."
George Bush, 4/25/89
Technical Details
Message-ID:
<1992Sep1.010514.432@leland.Stanford.EDU>Path:
arc1-ca.pugleaf.net!sparky!uunet!stanford.edu!leland.Stanford.EDU!newsReferences:
<1992Aug31.112327.2138@osf.org>Show Full Headers
Complete Article Headers
Newsgroups: comp.sys.next.hardware Path: arc1-ca.pugleaf.net!sparky!uunet!stanford.edu!leland.Stanford.EDU!news From: m@crito.stanford.edu (M Carling) Subject: Re: Terrible SCSI Transfer Message-ID: <1992Sep1.010514.432@leland.Stanford.EDU> Sender: news@leland.Stanford.EDU (Mr News) Organization: DSO, Stanford University References: <1992Aug31.112327.2138@osf.org> Distribution: usa Date: Tue, 1 Sep 92 01:05:14 GMT Lines: 47