Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!milano!bigtex!james
From: james@bigtex.cactus.org (James Van Artsdalen)
Newsgroups: comp.unix.sysv386
Subject: Re: Need buying advice for 386 and Unix
Message-ID: <49991@bigtex.cactus.org>
Date: 19 Nov 90 19:38:47 GMT
References: <5682@crash.cts.com>
Reply-To: james@bigtex.cactus.org (James Van Artsdalen)
Organization: Institute of Applied Cosmology, Austin TX
Lines: 46

In <5682@crash.cts.com>, jca@pnet01.cts.com (John C. Archambeau) wrote:

> A 486/33 motherboard will yield about 14 to 15 MIPS.  But keep in
> mind the bottleneck going across the ISA bus.  A 64-bit processor
> running on a 16-bit bus.

There's nothing 64-bit about a 486.  It does have 128-bit cache lines.

> Sort of reminds me of a traffic jam when a highway suddenly goes
> from 4 lanes to 1 (which is the correct ratio).  Do NOT get a 486
> unless you're going to go EISA or MCA.  It's a waste of CPU bus
> bandwidth if you don't.

I disagree here.  The data coming off of the hard disk is much less
than the bandwidth of the AT bus.  Therefore you can't win with just
an EISA hard disk controller.  What you CAN win with is a caching
controller, or a controller that can do DMA & has a unix driver that
can use it, or some other optimization not related to the bus
bandwidth.  Caching helps ISA too.

It's no accident that the EISA hard disk controllers that are
appearing have better hardware support (DMA, cache) than their ISA
counterparts: they wouldn't be much faster if they didn't (note: DMA
is slower on ISA anyway).

Also note that the fastest EISA Ethernet cards are maybe 5% faster
than the 8-bit WD-8003.  20Mbytes/sec bus bandwidth isn't needed to
communicate over a 10Mbit wire.

I will be buying EISA stuff for bigtex, but for the caching and DMA
features.

| (3) get Toshiba SIMMs.

> Why Toshiba?  Memory is memory.  NEC, TI, et. al.

No.  Completely wrong.  Look at a DRAM specification sheet some day.
There are at least twenty different parameters that must be met.  They
are are a little different for each SIMM that plugs into the same
socket.  Part of the design effort in a system is to allow as many
different SIMMs to be used as possible, and then make sure you *don't*
ship the ones that won't work.  RAS precharge time & friends: all that
fun stuff.
-- 
James R. Van Artsdalen          james@bigtex.cactus.org   "Live Free or Die"
Dell Computer Co    9505 Arboretum Blvd Austin TX 78759         512-338-8789