mccalpin@MASIG1.OCEAN.FSU.EDU ("John D. McCalpin") (03/17/89)
Thanks to:
Len Lattanzi (len@Synthesis.com) Synthesis Software Solutions, Inc.
for information about the feedback optimization.
Some more questions:
(1) Can both of these options (cache reorganization and inlining)
be used at the same time?
(2) The -O3 with the -feedback option worked. I got no speedup
on this particular code, but that is not necessarily a problem.
(3) The -cord option does NOT work. The compiler gets almost to
the end before bombing --- it could not find /usr/bin/ftoc.
What does ftoc do, and should it be there?
By the way, I am not running dhrystone :-). I am running LINPACK and
a variety of floating-point intensive finite-difference PDE codes.
So far it looks like loop unrolling buys a lot on this machine.
On 32-bit LINPACK (order 100 case), with full optimization -O3,
unrolling the innermost loops (the BLAS subroutines) gives a speedup
from 1.4 to 1.9 MFLOPS (unrolled to a depth of 16). I still can't
recover the 3.0 MFLOPS in the LINPACK published results for the MIPS
M-800 (which should be the same CPU and clock speed).
--
---------------------- John D. McCalpin -------------------------
Mesoscale Air-Sea Interaction Group & Department of Oceanography
& Supercomputer Computations Research Institute - Fl State Univ.
mccalpin@masig1.ocean.fsu.edu mccalpin@nu.cs.fsu.edu
mccalpin@fsu (BITNET or MFENET) SCRI::MCCALPIN (SPAN)
------------------------------------------------------------------