[[XviD-devel] nasm code to permute Simple_IDCT parm list

[trbarry at trbarry.com]

Hi -

At Nic's suggestion I added nasm/mmx code in MPEG2DEC3 to
permute-in-place the parm list to Simple_IDCT.  This allows the IDCT
results to not have to be copied again on the way out and, on my
machine, speeds up MPEG2 decoding by about 5-8% when using Simple_IDCT.

Supposedly this simple_idct_mmx.asm function and the needed parm
permuting are also used in Xvid.  So my small change might also be
useful here.

I'm not currently set up with Xvid CVS or source but, if it can help,
please see the new self-permuting simple_idct_mmx_P function at the
bottom of source member simple_idct_mmx.asm at:
www.trbarry.com/MPEG2DEC3_1.0.7T.zip .

Cheers,

- Tom