Re: Accelerator //e

[nathan@visi.com (Nathan Mates)]

In article <350FF8F4.1BD1@library.ucla.edu>,  <pubpc@library.ucla.edu> wrote:
>True, but try using your graphical word processor or page
>layout software and see which one is faster.  Why are you
>so offended when I say _my_ accelerated GS is faster than
>_my_ 486SLC for tasks like word processing?  I am just
>reporting what I see.  That's no offense to your Pentium.

   You're comparing one big bloated GUI program on the PC side (and
laptops have historically have always had worse video thruput than the
average desktop of the same era, another fact you conveniently ignore)
to something written mostly in asm on the GS side. For a more
realistic benchmark, try something that's very processor intensive on
both sides, and *not* written in asm: a compiler. 

   Gcc on the 386/40 I own to be a personal linux box on my desk at
work (want something for telnet out that doesn't suck as bad as a
Microsoft OS-- and as I'm doing game development for PSX and Direct3D,
NT is right out) seriously smokes Orca/C. gcc is one big, bloated,
unwieldy program in its own right, but even with optimizing code well
vs Orca/C w/o optimizations, my linux box is at least 5-10x
faster. Plus, running any processor intensive programs is way faster
on the linux box.

   Put simply: your benchmarks are misrepresentative of system power.
Plus, with accelerators for the GS not available new anymore, you
really going to think anyone's going to want to but a GS, outbid many
others to get an accelerator card, upgrade the card to at least 12Mhz
just to do "real" work better than on a 4-year old PC? Sorry, a
refurb'd P90/16MB Ram/750MB HDs can be found for $299, about the price
of a GS + enough upgrades to make it a power system. 

   Direct your system advocacy into telling existing GS users to vote
with their pocketbooks and buying GS hardware and software to keep
supporters in business. Alltech sold (1) new Zip accelerator in 1997,
and then when the last one went, users who'd not bought before
suddenly panicked and realized that they'd waited too long. I hear the
waiting list for refurbed GS accelerators from Alltech is now upper
2-digits in depth. 

   Why should it take a near-crisis in hardware upgrades to force
people out of their complacency and not wanting to purchase anything
for their GSs? I'd get close to calling this a community of
cheapskates, after this and their general not wanting to buy software
and hardware in general. Your time would be better spent getting
people to support the Apple II via their pocketbooks than irrational
and illogical arguments about system speed that only drag down and
expose your ignorance of the system capabilities of both sides.

>Btw, I researched the 486SLC some more and found that this
>chip, designed by Cyrix, although it is based on a 386 can
>perform _much_ faster than a regular i386.  According to
>one source, a Cyrix 486SLC-25 can keep up with a 486DX-66
>in some tasks.  :)  

   Yeah, they optimized one instruction better than a genuine intel
486DX-66, and if the program just happens to consist of that
instruction over and over again, it's faster. Try some real benchmarks
for 32-bit code (which will bite on a SLC as it's got a 16 bit bus
versus the 32 bit bus of a DX), video benchmarks, and the like. I bet
www.zdnet.com has some benchmark programs you could take a look at, or
you can go for dhrystone (processor-intensive, another benchmark that
shows a 386/40 to have more potential power)

Nathan Mates
--
<*> Nathan Mates http://www.visi.com/~nathan/      <*>
# What are the facts? Again and again and again-- what are the _facts_?
# Shun wishful thinking, avoid opinion, care not what the neighbors
# think-- what are the facts, and to how many decimal places? -R.A. Heinlein