Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!csd4.csd.uwm.edu!uakari.primate.wisc.edu!unmvax!ncar!ames!sun-barr!newstop!sun!sally!plocher
From: plocher%sally@Sun.COM (John Plocher)
Newsgroups: comp.unix.i386
Subject: Re: How to choose a new 386 UNIX PC...
Message-ID: <124379@sun.Eng.Sun.COM>
Date: 8 Sep 89 23:11:29 GMT
References: <21969@cup.portal.com>
Sender: news@sun.Eng.Sun.COM
Reply-To: plocher@sun.UUCP (John Plocher)
Organization: Sun Microsystems, Mountain View
Lines: 91

In article <21969@cup.portal.com> cliffhanger@cup.portal.com (Cliff C Heyer) writes:
>2. 33MHz hardware not yet reviewed in key areas of: 
>bus speed, paged/interleaved memory, shadow 
>(BIOS/video) RAM, disk cache(memory or controller), 
>extended memory speed, wait states.

The issues here can be grouped like this:
paged/interleaved memory and/or cache
	Cache is best (32k is absolute minimum, >64k doesn't show much improvement)
	Static column (page mode) is almost as good,
	Interleaved is better than "normal" but not worth it if you are spending
	the $ on a 25/33 MHZ system... (Go with cache or page mode)
	
memory speed and wait states
	Without cache you need faster than 60ns for 33MHz no wait state,
			                   80ns for 25MHz no wait state
	With a cache you can get by with 80-100ns for the 33MHZ box and 100ns
	for the 25MHZ box.  Same for Page Mode.

bus speed
	If it isn't 8MHZ (i.e., if it is 10 or 12) don't touch it!  Many cards
	won't run faster than 8MHZ on the bus.

shadow RAM
	Not meaningful for Unix since Unix doesn't use the BIOS


>4. 33MHz hardware disk I/O only 0-5% better than 25MHz. In
>other words, it might as well be the same.

As stated in this group, the DPT caching controller with 2.5MB+ of ram is FAST.
So is a good ESDI or smart SCSI

>1. UNIX (or any multitasking OS) and the effects of 
>the on-board cache:
>
> 	While multitasking, does flushing the cash waste a 

Flushing the cache means that the next 32,000 unique instructions/data references
will run at the slower memory speed (3-8 wait states each)

See the notes in this group about interactions between dual ported memory on 
I/O cards (serial/ethernet) and the cache.  If you can't disable the cache
for these cards, you can't use the cards.

>
>2. Is memory technology (cost/speed ) lagging behind 
>microprocessor technology? All the newest 33MHz 
>80386 PCs are using 70+ ns DRAMs when the 386 is 
>running at 30 ns and the on-board caches are rated 
>at 25 ns. You can't get 0 wait states 100% of the 
>time with this approach.

If you could get affordable 50ns dynamic RAM you could ignore all the cache
questions for a 33MHz box.  Since you can't, you use fast (25ns) cache RAM
and slower dynamic RAM.  (The cache has to be accessed every time memory is,
so it must be quite a bit faster than main memory)

>3. Is it impractical (cost and/or size) to put 40 256KB
>25ns SRAMs
>alot of trouble to go to...but how much do SRAMs cost?

LOTS.  At that speed they would be at least an order of magnitude more expensive
than the 70ns DRAMS per megabyte.
Also SRAM uses *POWER* which means big power supplies and lots of heat.

Back in '84 I bought a complete PCAT clone (kbd, mono, 1MB RAM, 30MB hard disk...)
for LESS than a 1MB 8MHz Static Ram board for my S100 system cost.


>4. Are any board makers making (or have made) 
>motherboards with ESDI and/or SCSI interfaces ON 
>BOARD to bypass the 8MHz AT bus? Also hopefully 

Mylex has a 32bit Disk controller (I think it is SCSI) and it is very fast!

>this mfg. would include shadow RAM (BIOS & video) 

Why?  Unix can't use it.

>and extended/expanded memory that is as fast as 
>main memory. (eg. add on memory boards have same 
>cycle time as the first 2MB.)

If the Motherboard has a 32 bit connector card then the expansion board 
should run at full speed as main memory.


I will leave the rest of the comments/questions for others ...

   -John