This is the mail archive of the
mailing list for the GCC project.
Re: Problems with FP compares / mixed SSE and I387 code
- From: Jakub Jelinek <jakub at redhat dot com>
- To: Uros Bizjak <uros at kss-loka dot si>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Fri, 7 Jan 2005 04:39:04 -0500
- Subject: Re: Problems with FP compares / mixed SSE and I387 code
- References: <41DE57B8.firstname.lastname@example.org>
- Reply-to: Jakub Jelinek <jakub at redhat dot com>
On Fri, Jan 07, 2005 at 10:34:48AM +0100, Uros Bizjak wrote:
> Some progress was achieved by separating SSE and x87 floating point
> operators by TARGET_SSE_MATH, and it was again shown that a small
> (ehm...) oversight in i386.md produced another big performance problem .
> Regarding these reports, I would suggest this solution:
> - with TARGET_SSE_MATH, _all_ SFmode values should be processed in SSE
> - with TARGET_SSE_MATH && TARGET_SSE2, _all_ SFmode and DFmode values
> should be processed in SSE registers.
Well, if you have __asm with "f", "t" or "u" constraints, you should still
put SFmode resp. DFmode values into the stacked registers.