Slashdot Mirror

← Back to Stories (view on slashdot.org)

Grand Unified Theory of SIMD

Posted by Hemos on from the the-string-theory-of-SIMD dept.

Glen Low writes " All of a sudden, there's going to be an Altivec unit in every pot: the Mac Mini, the Cell processor, the Xbox2. Yet programming for the PowerPC Altivec and Intel MMX/SSE SIMD (single instruction multiple data) units remains the black art of assembly language magicians. The macstl project tries to unify the architectures in a simple C++ template library. It just reached its 0.2 milestone and claims a 3.6x to 16.2x speed-up over hand-coded scalar loops. And of course it's all OSI-approved RPL goodness. "

2 of 223 comments (clear)

  1. faster? Bogus.... by Anonymous Coward · · Score: 0, Troll

    Excuse my ignorance, but how can a C++ template library be faster than hand coded assembler? ever.... no really - with a straight face. Given of course that "hand coded" implies it's hand coded for the task at hand an not something "like" it. If this was an article about a SIMD library why does it go all koolaid? Is this today's "mac-mini" astroturf?

  2. Re:Altivec by Lord+Kano · · Score: 0, Troll

    Altivec is good stuff but, I think it's funny how for years Mac Zealots were talking about how RISC was more efficient, but Altivec proves that speed can be gained by having more instructions in the chip.

    LK

    --
    "Hi. This is my friend, Jack Shit, and you don't know him." - Lord Kano