[MLton] Multicore CPU's and MLton

Daniel C. Wang danwang@CS.Princeton.EDU
Tue, 05 Jul 2005 10:05:22 -0700


Stephen Weeks wrote:

>I think that multi-core is a logical (even necessary) direction for
>chips to go.  But I think that process-level parallelism is usually
>the way to take advantage of it. 
>
Oh, I guess this got lost in the discussion, but I don't really care how 
parallelism is implemented. Process, level ala parallel make would be 
fine for me.  I just want to make sure that MLton can use that extra 
core for something!

I'll have to say I was a little skeptical about this whole multi-core 
thing, but then I started noticing all these Pentium Ds floating around. 
A dual core Pentuim D running at 2.6hz per core is around $268. Shipping 
systems are  below $1500. AMD seems to be aiming their dual cores at the 
higher end server market. Intel is playing catchup and I think is 
attacking the low end. If you look at the roadmaps for both Intel and 
AMD they are in a pissing match to figure out who can deliver the quad 
core chip! Remember all the real profits are in the server market. Us 
desktop users litteraly get the server chips that can't run fast and 
cool enough! BTW you would be quite surprizsed about how all the thermal 
and performance concrens of a laptop chip and a server chip are more 
alike than different. I expect to see an Apple branded x86 with a dual 
core pentium M in a little more than a year. :)

In less than five years, everything, execpt for the very low end, I 
suspect will be multicore.  Languages that have some sort of seperate 
compilation mechanisms can take advantage of the extra cycles. I'm 
concerned that the whole program methodology while quite elgeant and 
practical today, will look less and less so in the future.