[BUGS] hardware changing time .. moving forward
Andrew Reilly
areilly at bigpond.net.au
Tue Jan 29 09:48:32 EST 2008
On Tue, Jan 29, 2008 at 09:14:57AM +1100, Jerahmy Pocott wrote:
>
> On 28/01/2008, at 9:06 PM, harrywwc wrote:
>
> > my turn...
> >
> > ref: DEC Alpha chip and all that.
>
> My experience with the Alpha, was that it had absolutely TERRIBLE
> floating point performance..
How did you manage that? Did you have all of the "precise
exceptions" and "go slow" switches turned on?
I have to admit that I never got to seriously use an Alpha,
although I sat close to several people who did. Everyone seemed
very happy with their floating point performance. The MIPS
boxes that I did get to use were easily twice as fast as the PCs
of the day, though (on floating point).
In the present timeframe, though, Intel Core-2 processors
execute as many flops/clock as Itaniums (if you line 'em up
right) and clock twice as fast. Athlons/Opterons aren't far
off. Only the vast caches on Itania and PowerPC-6 really give
them any edge. MIPS has packed-up and gone embedded. SPARC
is chasing multi-thread web-server benchmarks and has given
floating point a rest for a while. Alpha and PARisc have shut
up shop. One has to get one's flops where one can find 'em,
and frankly, I can't see anything much to complain about. (And
now VIA/Centaur have announced an x86_64 processor that can
apparently do eight flops/clock with really low latency... who'd
a thunk it?)
Cheers,
--
Andrew
More information about the BUGS
mailing list