Anyone benchmarked SSE2 bitsliced Serpent?
From: Paul Rubin (//phr.cx_at_NOSPAM.invalid)
Date: 03/26/05
- Next message: Douglas A. Gwyn: "Re: Inaccuracy of ANSI clock() as ?-RNG"
- Previous message: Dave Turner: "Bruce Schneier's Applied Cryptography"
- Next in thread: Tom St Denis: "Re: Anyone benchmarked SSE2 bitsliced Serpent?"
- Reply: Tom St Denis: "Re: Anyone benchmarked SSE2 bitsliced Serpent?"
- Reply: Grumble: "Re: Anyone benchmarked SSE2 bitsliced Serpent?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Date: 26 Mar 2005 00:45:55 -0800
Among ciphers that have stood up to serious analysis, that might be
the fastest one yet. It might do even better on the Athlon 64 than on
the P4, since the A64 has sixteen SSE2 registers instead of eight, so
the A64 can hold more intermediate results if needed.
Using SSE2 would allow doing 4 encryptions in parallel, or 128,
depending on how one counts them.
- Next message: Douglas A. Gwyn: "Re: Inaccuracy of ANSI clock() as ?-RNG"
- Previous message: Dave Turner: "Bruce Schneier's Applied Cryptography"
- Next in thread: Tom St Denis: "Re: Anyone benchmarked SSE2 bitsliced Serpent?"
- Reply: Tom St Denis: "Re: Anyone benchmarked SSE2 bitsliced Serpent?"
- Reply: Grumble: "Re: Anyone benchmarked SSE2 bitsliced Serpent?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Relevant Pages
|