FAST vector multiply routine

FAST vector multiply routine

Post by Ernest Hucka » Sat, 09 Oct 1999 04:00:00

I was wondering if there is a fast vector multiply
routine analogous to Goto's fast DAXPY routine. I
did a few tests and found that DAXPY is about
3 times as fast
as an F90 vector multiply or vector add.

real(kind=8), dimension(1:N) :: u,v,w

u = v*w



1. fast multiply for (long long) on UltraSPARC-II

Does anyone know of a way to optimize a long long multiplication on a
UltraSPARC-II?  If not, is there a way to detect an overflow when
multiplying two integers.


Brian Holland
The MITRE Corporation MS E095
202 Burlington Road
Bedford, Ma. 01730-1420

2. modem drivers

3. Fast vector operations on a Sun?

4. stdarg.h question

5. Fast opmized BLAS(Level 1) routine is available

6. rexec() - getc blocks - why?

7. Fast memory search routine wanted

8. Etherwave Cards

9. Ported fast Cray libm routines now available !

10. Fast BLAS routine(Level 3)

11. Version 0.21 of free fast math routines (libffm, preliminary version) now released !

12. fast GEMM routine

13. HELP: need pointers to fast fixed precision routines