[Date Prev][Date Next] [Thread Prev][Thread Next]
[Thread Index]
[Date Index]
[Author Index]
Re: lapack, blas, gemm under linux/axp
- From: Greg Lindahl <lindahl cs virginia edu>
- To: axp-list redhat com
- Subject: Re: lapack, blas, gemm under linux/axp
- Date: Wed, 16 Sep 1998 10:20:44 -0400 (EDT)
> It also dies making slarfx.o. I found out that one of three optimization
> options has to be turned off, then g77 -O6 -mcpu=ev56 -Wa,-mev56 is
> successful:
>
> -fno-rerun-cse-after-loop OR -fno-schedule-insns OR -fno-strenght-reduce
BTW, it would be very helpful if you could install egcs-1.1b, verify
that the bugs still exist, and then report them to
egcs-bugs@cygnus.com. If you have more time, also try throwing in
-mieee to see if that hurts anything.
> Any comments, which one hurts least? Also, any gurus willing to explain
> what these options really do?
I'd guess that Haifa is actually the guilty party, and also that Haifa
produces the least benefit...
Do these RISC blas things have tunable parameters for block sizes?
-- g
[Date Prev][Date Next] [Thread Prev][Thread Next]
[Thread Index]
[Date Index]
[Author Index]
[]