Erlang & performance / watt

Peter-Henry Mander <>
Tue Dec 13 10:00:46 CET 2005


I'm intrigued by this:

"If there are enough threads (hardware and software), one should never
have to speculate. The eight CPU cores were a throwback to early RISC
designs: single-issue, in-order, nonspeculative."

So, it looks as if Intel was going down a cul-de-sac with speculative,
out-of-order, hyperthreaded, overheating billion transistor etc... CPUs,
and that the RISC paradigm is destined for a resurgence. I have a
fondness for simple, orthogonal CPUs.

... And that a multicore Erlang CPU might find a market yet? (ducking to
avoid the flames, not just those from Buncefield nearby:
http://news.bbc.co.uk/1/hi/uk/4523430.stm).

Pete.

On Tue, 2005-12-13 at 09:21 +0100, Michael Fogeborg wrote:
> How does Erlang compete here ?
> Any advantages when it comes to performance / watt ?
> 
> http://www.acmqueue.com/modules.php?name=Content&pa=showpage&pid=330
> 
> A quote from the article:
> "....
> The complexity of parallel software can slow down programmer productivity 
> by making it more difficult to write correct and efficient programs. 
> Computer science students' limited exposure to parallel programming, lack 
> of popular languages with native support for parallelism
> ...."
> 
> ---
> 
> 
> 





More information about the erlang-questions mailing list