Hi,
When compiling a simple matrix dot product example on both
alpha & ia64 using -O3 and -fmodulo-sched options,
I am unable to find a software pipelined loop in the
generated assembly.
I think the advanced optimizations dont seem to occur
anymore in Gcc although they are documented
in the manuals.
Is this a known issue ?
I think that apart from powerpc, no other architecture
is optimized properly by gcc. This is probably due to
the separate contrib tree from IBM vs the rest of the world.
The IBM contrib works whereas later contribs dont.
The IBM contrib from Haifa works but I am unaware
as to why newer contribs seem to fail !
None of the selective sched, modulo sched contributions
show up at all!
Can anyone help?
Sincerely,
Seima Rao.
