Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sdd.hp.com!wuarchive!zaphod.mps.ohio-state.edu!uwm.edu!ogicse!hsdndev!rutgers!cbmvax!daveh
From: daveh@cbmvax.commodore.com (Dave Haynie)
Newsgroups: comp.sys.amiga.advocacy
Subject: Re: A3000UX - Born to run UNIX SVR4
Message-ID: <18892@cbmvax.commodore.com>
Date: 13 Feb 91 00:10:40 GMT
References: <1991Feb7.151106.4795@cc.helsinki.fi> <32530@auc.UUCP> <1512@pdxgate.UUCP> <11964@helios.TAMU.EDU>
Reply-To: daveh@cbmvax.commodore.com (Dave Haynie)
Organization: Commodore, West Chester, PA
Lines: 43

In article <11964@helios.TAMU.EDU> n177ac@tamuts.tamu.edu (Daryl Biberdorf) writes:
>In article <1512@pdxgate.UUCP> hal@eecs.UUCP (Aaron Harsh) writes:
>>In article <32530@auc.UUCP> rar@auc.UUCP (Rodney Ricks) writes:

>Since when does the processor alone determine the total capability
>of the machine?  

Never did, never will.  You have to not only consider the whole system, but
the software that's running on it.  Many things you do under AmigaOS on an
'030 system will seem quite a bit faster than under NeXTSTeP on a slab.  CPU
and FPU bound crunching operations will always be faster on an '040 NeXT
than an '030 Amiga, but a good deal of what you run is bound by hard disk and 
user interaction.  You also have to consider where the machine is headed.  As
a UNIX-based network workstation, a slab type configuration makes some sense.
Sun and many others figured this out before NeXT did.  As a personal computer,
it doesn't make lots of sense.

>We had an 030 NeXT in the integration lab at my last co-op assignment.  As 
>long as you wanted to run ONE job at a time the machine's performance was 
>excellent.  But try feeding a score file to the DSP (yes, teh sound *is* 
>wonderful) while doing one or two other tasks and watch the whole thing grind
>to a halt (INCLUDING the music).  The NeXT's architecture must be such that 
>the *CPU* is responsible for feeding data to the DSP (and, by analogy, to the
>other specialized chips).  

Reasonable conclusion, but wrong.  The DSP itself is a slave-only device; the
CPU must feed data into it, while it operates out of a small amount of very
fast internal RAM and 24K of pretty fast external SRAM (expandable on the 
slabs, not on the cubes).  Most of the remainder of the NeXT architecture,
including disk I/O and video fetch, seems to be reasonably DMA and interrupt
oriented.  I have yet to find any real detailed descriptions of how they do
it, and doubt it's anything quite as sophisticated (architecturally speaking)
as the way the Amiga chips handle DMA.  More along the lines of the shared,
arbitrated bus and FIFO approach used on the A3000 for hard disk, rather than
dedicated DMA slots.  

>--Daryl Biberdorf,  n177ac@tamuts.tamu.edu  OR dlb5404@rigel.tamu.edu


-- 
Dave Haynie Commodore-Amiga (Amiga 3000) "The Crew That Never Rests"
   {uunet|pyramid|rutgers}!cbmvax!daveh      PLINK: hazy     BIX: hazy
	"What works for me might work for you"	-Jimmy Buffett