Re: Speed freak needs help

[moon (David A. Moon)]

> Date: 23 Dec 92 14:46 GMT
> From: UK0392@AppleLink.Apple.COM (EHN & DIJ Oakley,GB,IDV)
>  
> I hate to say this, but we are now running into a speed problem with MCL and a
> Quadra 950.
>  
> John Koza extols TI Explorers, but after finally tracking down someone who
> claims to have a small number of new Explorer systems, I am not sure whether we
> would see anything like the necessary acceleration, and the systems are not
> cheap.

TI stopped making Explorers several years ago.  Without having measured either of 
them, I think Explorers were much slower than Quadras.  Symbolics still makes 
Lisp machines, but again without having measured them I think you would find 
performance on the latest Symbolics machine similar to or less than the Quadra.  
(The Symbolics clock rate is half the Quadra clock rate, but the architecture is 
not much more than twice as efficient.)

There are Lisp implementations available for a few supercomputers such as the 
Cray-1.  I don't recall that the speedup was in proportion to the increased cost, 
though.  The architecture is not suited to Lisp and probably the implementations 
were not very clever.

You might get a small speedup (factor of 3 or 4?  just guessing) by using Common 
Lisp on the latest and greatest RISC workstation, whichever that is this month.  
Use about 3 times as much memory as you use in MCL, times 2 if it uses a copying 
garbage collector, and you should be alright on memory configuration.  Do a test 
run before buying anything.

Are you sure the algorithm can't be optimized?  Sometimes that's the fastest and 
cheapest way to get more speed.  But maybe you already tried that.