Teaser: Another purchase made, more NOS equipment on the way!

[redrumloa]

redrumloa wrote:
I just made another nice purchase which will start making it's way into my auctions within ~3-4 weeks. No full systems this time, bot alot of neat NEW loot :-D  :-D

Stay tuned:-)

Damn I have no patience, this lot is on it's way... But I want it now :-(

Oh and I am in negotiations for 2 other lots ATM. This is getting really interesting :-D

[DamienMcKenna]

A warehouse full of Eclipsis machines? )

Damien

[downix]

@-D-

A DSP is not the same as a CPU, how it's math results are measured do not come up equal.  Let's dig up the AT&T DSP3210 performance specs here for a moment:

The AT&T DSP3210 is capable of 12.5 million instructions per section, 25 million floating point operations per second according to Dave Haynie.  Yes folk, that's SIMD for ya.  Each instruction does 2 math operations.  All of these, 32-bit floating-point.  No intermix of 8 or 16-bit here, simplified the design.  But real world tests reveal the same DSP getting 16.7 MIPS / 33MFLOPS.  The DSP3210 was introduced in 1991, back when 386's were still speed-kings.  And a 386 can't even do a single MFLOP.

Now, the 68030 does 1.3 MFLOPS @ 50Mhz.  The 68040-25 did about twice that, at 2.4 MFLOPS  The Pentium 2 gets 9 MFLOPS @ 450Mhz.  So looks like I am right here.

As for audio, the DSP3210 was not limited to audio, nor was it ever intended to be limited to audio.  They had an audio chip that was to run alongside it (from Crystal Semiconductor or Analog Devices) for audio.

As well, the DSP3210 continued to be evolved, with 55Mhz, 66Mhz,  and even 80Mhz models arriving.  But it ended it's life in 1999, along with all of Lucent's 32-bit DSP solutions.  Sad really.  It could have given the Amiga a real shot in the arm.

[alx]

Think much much smaller :-)

No-one suspected it, but C= was working on an Amiga laptop - and it's the only prototype :-o

...or maybe not ;-)

[sdesros]

A full size arcade game and loose Jamma boards.

I don't suppose you also frequent(ed) the Neo-Geo.com forums?  There's a "redrum" on those forums that apparently hail from Florida.  :-?

[Bobsonsirjonny]

Matt_H wrote:
At the rate things are going, I'd estimate you're only a few months away from finding the legendary warehouse of brand new CD32s.  :-D

There was a £ shop in Wrexham selling CD32's off. Brand new ones at £1 each. I bought 5 and gave em all away......


And now when I'm really broke - totally skink and actually thinking about selling my computer.... damn.

[GadgetMaster]

@ Bobson

You gave all of them away?

Next time put such things on e-bay. You will find that people are prepared to pay you to have them.

Anyway, spilt milk and crying and all that....:-P

Hope your fortunes change for the better soon.  :-)

[that_punk_guy]

Bobsonsirjonny wrote:


There was a £ shop in Wrexham selling CD32's off. Brand new ones at £1 each. I bought 5 and gave em all away......

You... did... WHAT?!?!  :-o

 :-)

[Damion]

@Downix

I'm afraid you're mistaken; let's look at some
figures from the Sandra benchmarking utility:

Intel PII @450mhz   1220 MIPS   590 MFLOPS
AMD K6-2 @400mhz    1165 MIPS   475 MFLOPS
Intel PI @266mhz     630 MIPS   300 MFLOPS

And a few Pentium 3 benchmarks:

Intel PIII @500mhz  1350 MIPS   670 MFLOPS
Intel PIII @750mhz  2012 MIPS   998 MFLOPS

Oh and what the hell,

68060/60 @66mhz       86 MIPS    34 MFLOPS

Yes I'm well aware that the DSP in the A3000+
was also to be used for some maths processing,
but let's be realistic, the '030 and the DSP
combined wouldn't have equalled the FPU
performance of a P266, and once we factor
integer performance (MIPS) it's really in for
a beating.

Then we have to consider other things such as
bus, memory speed and graphics performance,
and I'm sorry but in every respect your average
PIII had the 3000+ beaten by a considerable
margin.

So looks like I am right here.

Please check the number(s), and try your call
again later.

[downix]

@-D-

I pulled my figured right from overclockers websites for 32-bit MFLOP.  What bit-depth are those numbers, because they look like 8 or 16-bit to me.

[redrumloa]

I don't suppose you also frequent(ed) the Neo-Geo.com forums?  There's a "redrum" on those forums that apparently hail from Florida.  :-?

Nope, not me! Interesting coincidence :-)

[Damion]

All of the processors listed have full 32bit
architectures (branch, integer and floating
point) and I assume have been measured as such.

[downix]

@-D-

But did the benchmarking programs state as such?  I made very well sure the spec I grabbed said so.  (Mind you, it also said "L2 cache disabled" on the P2)  It's an old trick to run smaller/faster code to inflate a benchmark.  (I remember old coders that ran the 68000 in 16-bit mode, which ment it executed 150% faster from the benchmark's angle)

[Damion]

Well to be sure I'll doublecheck the sources and
post back later today.

[downix]

@-D-

I'd also appreciate some P2-450 scores with the L2 engaged.  I know that the FSB and memory access would kill the perf without it.

[redrumloa]

OT! Yeesh :-D