| Re: [AD] malloc() vs align_malloc() |
[ Thread Index |
Date Index
| More lists.liballeg.org/allegro-developers Archives
]
Peter Wang wrote:
> I doubt it would improve performance with our current code, but it
should be benchmarked.
i doubt it will degrade performance (with our current code), yet offer
the user the ability to use SSE2 and/or developers to develop SSE code
for future use in allegro.
in the integer front, for instance, i can already see a SSE2 instruction
that may make masked_blit signicantly faster.
for floats there are *many* SSE instructions that could be used.