What do you guys thing about the P4

Richard Walsh rbw at networkcs.com
Fri Apr 13 15:45:47 EDT 2001


Phillip Matz wrote:

>Which brings up another question.  My current impression was that SSE/SSE2
>is only single-precision while G98 (and probably most computational
>chemistry programs) requires double-precision -> if this is correct then SSE
>doesn't help, rather, it invalidates ones G98 results.  Would anyone care to
>set me straight?

I think SSE---> 2 is what makes the difference.

Intel's write up on the P4 shows a packed double precision (IEEE 64-bit)
floating point data type (2 per XMM register) and goes on to say

"The SSE2 instructions have two primary functions.
        ^
        |
        |
    Permit packed [read small-vector] and scalar double-precision
    [read 64-bit] floating-point operations to be carried out in 
    the XMM [that's MMX backwards ... ;-)] registers.

    ....

"

I read that as a go for compiler developers wanting to improve their
compiler's double-precision (64-bit) performance on a P4. Bob Feller's
G98 benchmarks in an earlier posting confirm the excellent Gaussian
performance on the P4. 

Regards,

rbw
#---------------------------------------------------
#
# Richard Walsh
# Project Manager, Financial Markets
# netASPx, Inc.
# 1200 Washington Ave. So.
# Minneapolis, MN 55415
# VOX:    612-337-3467
# FAX:    612-337-3400
# EMAIL:  rbw at networkcs.com, richard.walsh at netaspx.com
#
#---------------------------------------------------
# "What you can do, or dream you can, begin it;
#  Boldness has genius, power, and magic in it."
#                                  -Goethe
#---------------------------------------------------
# "Without mystery, there can be no authority."
#                                  -Charles DeGaulle
#---------------------------------------------------


_______________________________________________
Beowulf mailing list, Beowulf at beowulf.org
To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf



More information about the Beowulf mailing list