Newsgroups: comp.arch
Path: utzoo!henry
From: henry@utzoo.uucp (Henry Spencer)
Subject: Re: Self-modifying code
Message-ID: <1988Jul26.024039.28579@utzoo.uucp>
Organization: U of Toronto Zoology
References: <1988Jul22.164129.5495@utzoo.uucp> <4912@killer.DALLAS.TX.US>
Date: Tue, 26 Jul 88 02:40:39 GMT

In article <4912@killer.DALLAS.TX.US> elg@killer.DALLAS.TX.US (Eric Green) writes:
>> ... A good implementation of BitBlt,
>> whether in software *or* hardware, will run the memory flat out, leaving
>> nothing for "servicing other processes".  
>
>But the 68000 (uncached) version would use 3/4ths of that bandwidth 
>to fetch INSTRUCTIONS...

Sigh.  Please think before answering.  An artfully-build MOVEM loop does
*not* spend 3/4 of its time fetching instructions.
 
>... the bandwidth of modern 100ns 
>memories is much greather than a 8mhz 68000 can take advantage of...

Suggesting that one should either buy a faster processor or spend less
money on the memory.  There are sillier things than putting 100ns RAMs
on an 8MHz 68000, yes, but I'd have to think for a moment to come up
with specific examples.

>Given the cost constraints, and the state of technology in 1984, it
>made a lot of sense to have a hardware blitter. The fact that the Blit
>terminal did fast blits without a blitter probably has more to do with
>the fact that a) the machine was designed before modern high-speed
>DRAMs, when 250ns RAMs were fast...

The 630, designed rather more recently, does things the same way.  So
do the monochrome Suns.

> and b) the machine had a much higher
>resolution screen than the Amiga, meaning that much more bus bandwidth
>was taken up with video refresh (thereby improving the desirability of
>caching the loop, since you couldn't do data accesses any faster
>anyhow).

I don't follow this.  There is no cache in the Blit.

>> Note that some of the earlier Suns had quite a bit of hardware BitBlt
>> assist, and the newer ones don't.  Sun learned...
>
>Earlier Suns were based upon 68000/68010 techynology, where hardware
>bitblt's make a lot of sense. Later Sun's, based upon 68020s, don't
>need the hardware assist (see the beginning of this message). 

No, they dumped the hardware assist *before* they switched to the 020,
not after, unless I'm much mistaken.  Most of the Sun-2s (68010) had
no hardware assist.

More generally, I will repeat -- more explicitly -- something I pointed
out before:  the fair comparison is not to the same system without BitBlt
hardware, but to a system where the same effort and funding (custom chips
are *not* cheap to design) have been invested in making the main CPU faster
instead.
-- 
MSDOS is not dead, it just     |     Henry Spencer at U of Toronto Zoology
smells that way.               | uunet!mnetor!utzoo!henry henry@zoo.toronto.edu