This is the mail archive of the
mailing list for the GCC project.
Re: upcoming SSE/SSE2 support in 3.1
- From: Tim Prince <tprince at computer dot org>
- To: Jim Wray <wray at poseidon dot cs dot byu dot edu>, gcc at gcc dot gnu dot org
- Date: Thu, 9 May 2002 14:22:33 -0700
- Subject: Re: upcoming SSE/SSE2 support in 3.1
- References: <Pine.LNX.email@example.com>
- Reply-to: tprince at computer dot org
On Thursday 09 May 2002 14:07, Jim Wray wrote:
> I am getting very excited about the coming release of gcc 3.1 (as many
> are.) First, let me say, I considered sending this question to gcc-help,
> as it seemed more approriate, but I felt unsure that the proper audience
> would be there to answer question about upcoming products. My question
> deals with how automatic SSE(2) code generation is. I do research work in
> image vision, and am curious how well operations such as adding two images
> would be optimized. In other words, a typical operation would be to
> iteratively go through data that is consecutive in memory with two sources
> and a destination. Is this likely to get SSE(2) code generated
> automatically, or should I spend time looking at writing custom assembly.
> Thanks, for any help and all your hard work!
There is no auto-vectorization or auto-parallelization, so the compiler
simply generates scalar sse instructions rather than x87 instructions.
Intel's icc performs auto-vectorization, but I don't get the impression that
your application would take advantage of that.