Conference hydra::amiga_v1

    
    There is also going to be a 68040 board for both the 2000 and 3000
    machines...
    
    The 68040 for the 2000 was seen in Amiga World issue JULY or August
    
    And the board for 3000 was by Supra approx $1999 as seen in Popular
    Computing Weekly of New Computer Express (By Steve Gold in the States..
    [no relation..])
    
    	Steve G

re 68040:

What clock speed?

This should hit 20+ mips!

john

    
    re .-1
    
    	The speed is 25MHz.
    
    	The company is RCS of Canada...
    
    	They give it a speed of 20-25MIPS and 3.5+MFlops...
    
    	It is in Amiga World July Issue...
    	Steve G

>   	The speed is 25MHz.
>    
>    	The company is RCS of Canada...
>    
>    	They give it a speed of 20-25MIPS and 3.5+MFlops...
>    
>    	It is in Amiga World July Issue...
>    	Steve G

Hmm, 20-25 MIPS at 25mhz... What's it at 50mhz??  :-))))

john

    I wonder if it supports TU58s.....
    
    
    ..... No no It's Ok I was just having a nightmare......
    
    
    You forgot the Co-processor as well in those figures....
    
    
    			- Pete -

>>  Hmm, 20-25 MIPS at 25mhz... What's it at 50mhz??  :-))))

Unavailable.

	<mike

That's amazing.

A 68030 at 25mhz is around 7 mips. A 68004 a the same sclock is THREE times
that!

john

One of the 'spring-time' Bytes (it's at home) had a decent article on the
68040.  Apparently, Motorola really squeezed a lot onto that chip.  
It contains something like (brain don't fail me now) 900k transistors, and
that's more than twice as many as the 80486.

Combining smaller transistor size (shorter paths?) with advanced pipe-lining,
onboard cache, built in FPU, and other optimizations and you get one hell of 
a CISC (COMPLETE not complex Instruction Set Computer) CPU.

If I dig up anymore interesting things I'll post them.

Steve

    Re: .7, et al
    
    <Dave Haynie emulation ON ;-) >
    
    Most of the speed increase of the 040 in comparison to the 030 is due
    to a much enhanced design and features like independent
    data/instruction caches of 4KB size, pipe-lining, etc.
    However, detail contributing most to the tripling of the the performance is 
    the fact, that an 25MHz 040 will run internally at 50MHz (at least most
    of the places).
    
    <Hazy mode OFF>
    
    And since I'm a software weenie, I'll disclaim even my mere
    existance... ;-)
    
    <CB>

    The 68040 does run internally at twice the supplied clock speed which
    is one of the reasons it goes so fast. Another thing that will boost
    speeds on an Amiga is that with it's 8kB total cache size, it generally
    needs the bus an average of like 11% of the time. That leaves the 89%
    of the bus cycles to support chips and DMA. Since the support chips are
    designed to cycle steal (every other cycle), I don't know if this will
    really help, but it may. Even if it doesn't speed things up that way,
    you still get the speed increase from the fact that almost 90% of the
    time, no data/instruction fetches are needed.
    
    bob pomeroy
    refine::pomeroy
    

        From the July 1990 issue of Systems Integration - page 17,
        without permission.
        
        There's now a way to increase system performance at the processor
        level.  A new ICEC (integrated circuit, environmentally
        controlled) from Velox Computer Technology that sits on top of an
        Intel 80486 can increase clock speeds from 33 MHz to 50 MHz
        simply by cooling the chip.
        
        Velox's ICECap 486 uses supercomputer design principles to reduce
        processor operating temperatures to 0 degrees Celsius, while
        precisely controlling voltage levels.  ICECap consists of a
        thermoelectric cold plate, temperature sensor, environmental
        controller circuit and dual-clock speed generator.  It draws
        power from a systems's disk drive using a connector and wire
        assembly.
        
        The Santa Clara, CA manufacturer is readying other versions of
        its product.  Velox announced that it will ship other versions of
        ICECap 1 that will support Intel's 860 and Motorola's 68040 by
        the end of the year.  The next processors that Velox plans to
        support with its ICECap 2 are the DEC MicroVAX, Hewlett-Packard
        Precision, Intel 80386, Intergraph Clipper, MIPS R3000 and
        Motorola 68030.  Next year Velox will ship ICECap 3, a version of
        the product that's designed for OEM's that want to design the
        product into integrated circuit packages in bare die form.
        
        ICECap 486 is available in OEM quantities for $150 each.
        

Will one of those ICEcaps (for the 68030) be usable on the Commodore '30 board,
or perhaps with the A3000.  I assume that both designs are done asynchronously.

How about one for the 6888x?  That might be a more valuable increase for a
lot of people.

Incorporating one of those on a board would not only increase the speed, but
would save you a substansial amount of money on a new "faster" CPU.

Steve

RE:.12

>   Will  one  of  those  ICEcaps  (for  the  68030) be usable on the
>   Commodore  '30  board,  or  perhaps with the A3000. I assume that
>   both designs are done asynchronously.

    It  may  be  useful on the GVP 50mhz upgrade from 33 mhz, but for
other boards I would be skeptical. There  is  a  whole  lot  more  to
timing   than  the  CPU  and  clock.  Things  act  a  lot  more  like
radio/analog than Digital above 25mhz...


>   How  about  one  for  the  6888x?  That  might be a more valuable
>   increase for a lot of people.

    This  is  one  place  it  would likely be more useful all the way
around since most  boards  have  independently  (or  capable  of  it)
clocked FPU's.

	-SES

    
    	Anybody heard about the A3500.
    
     It's supposed to be a tower system with more ZorroIII slots..!!
    
    Also there is rumors of a A3000 portable...??!!
    
    	Steve G