DISQUS

DISQUS Hello!  The comments on this profile are unclaimed and thus are unverified.

Do they belong to you? Claim these comments.

Asm's picture

Unregistered

Feeds

aliases

  • Asm

Asm

9 months ago

in Searching arrays in X86 assembler with a bloom filter pt 3 on Phil Dawes' Stuff
Variable shifts have the count in %cl.

But why bother? bt r,r and bt r,i are only 1 uop on a Core2. Check the instruction timings here: http://www.agner.org/optimize/

Use mov (%eax,%edi,4), %edi and drop the shl $0x2, %edi.

9 months ago

in Searching arrays in X86 assembler with a bloom filter pt 2 on Phil Dawes' Stuff
BT is slow for m, r operands. 12 uops on a P4. 10 uops on a Core2. Avoid.
Returning? Login