15/07/21 02:59:01.93 HmEEJFC8x
MaxwellのLOP3.LUT命令はやっぱり効果抜群ですね。
上がJohnDoeさんのS-Box、下がJohn the Ripperのだけど、
ゲートカウントが全然違うもんなあ。
// Bitslice DES S-boxes with LOP3.LUT instructions
// For NVIDIA Maxwell architecture and CUDA 7.5 RC
// by DeepLearningJohnDoe, version 0.1.6, 2015/07/19
//
// Gate counts: 25 24 25 18 25 24 24 23
// Average: 23.5
* Bitslice DES S-boxes making use of a vector conditional select operation
* (e.g., vsel on PowerPC with AltiVec).
*
* Gate counts: 36 33 33 26 35 34 34 32
* Average: 32.875