[Ffmpeg-devel] SSE load and store doubts

[Dave Dodge]

On Thu, Oct 06, 2005 at 02:33:31PM +0200, Roberto Pariset wrote:
> by the way, is deassembler the only way to learn such things, or
> maybe there's a guide? i mean, intel talks alot about what asm
> instructions do, but i haven't found much about the intrinsics and
> how to code sse in c.

I'm not sure which Intel manuals you're looking at already.  A quick
bit of searching finds some that might be useful.

The Intel C/C++ compiler manual at least lists the available
intrinsics and gives some very brief introductory information.  Search
for "intrinsics" in the index:

  http://www.intel.com/software/products/compilers/clin/docs/main_cls/index.htm

Note: the compiler manual references an SSE2 "external architecture
specification" that does not seem to exist.

Here's a short introductory manual on using SSE and SSE2 (with both
intrinsics and inline assembly):

  http://cache-www.intel.com/cd/00/00/01/77/17741_getting_started.pdf

There's some specific examples here:

  http://www.intel.com/design/pentiumiii/sml/

A short magazine article:

  http://www.intel.com/technology/magazine/computing/sw03011.pdf

Some online tutorials with examples; the first one may require some
sort of registration process:

  https://shale.intel.com/SoftwareCollege/CourseDetails.asp?courseID=24
  http://www.intel.com/software/products/college/sse2_3d/index.htm

An introduction to SSE programming:

  http://www.codeproject.com/cpp/sseintro.asp

And of course there's the instruction set reference manuals, which I
think is where they've put the SSE stuff these days:

  http://www.intel.com/design/pentium4/manuals/index_new.htm

                                                  -Dave Dodge