Re: Multithreaded scientific programming

["jimmaureenrogers@worldnet.att.net" <jimmaureenrogers@worldnet.att.net>]

Björn Persson wrote:
> Martin Krischik wrote:
> > David B. Chorlian wrote:
> >> Talk about cache, TLB, and other memory management topics which
> >> are the serious problems in large-scale scientific computing.
> >
> > Well, for that I would have to know what a TLB is...
>
> Translation lookaside buffer - a small cache memory for the page table,
> which the memory management unit uses to speed up the translation from
> virtual addresses to hardware addresses. I guess David wants to write
> programs such that they only relatively rarely need to access memory
> pages that aren't in the TLB.
>

This sounds highly system dependent. Do all processors have the same
size TLB?

I know some current Intel dual core processors have a 16 Megabyte
cache, while the roughly equivalent AMD processors have a 2
Megabyte cache. It would seem that cache size and contents
could be very important for optimizations at the level one worrys
about TLB contents. On the other hand, the AMD dual core
processors have up to a 3000 Mts speed while the roughly
equivalent Intel processors have a 1600 Mts speed.

Both AMD and Intel employ the same instruction sets.
Compilers are not likely to recognize differences between
processors from the two vendors.

Is it common to write scientific programming in such an
unportable manner?

Jim Rogers