From a81162b0878e122496744241267fe05e4d06239b Mon Sep 17 00:00:00 2001 From: Jens Steube Date: Tue, 23 Feb 2016 12:18:47 +0100 Subject: [PATCH] Speed up -m 20 in -a 3 mode --- OpenCL/common.c | 550 ++++++++++++++++++++++++++++++++++++++++++++ OpenCL/m00020_a3.cl | 376 ++++++++++++------------------ 2 files changed, 697 insertions(+), 229 deletions(-) diff --git a/OpenCL/common.c b/OpenCL/common.c index 40b7b8d..2445e0f 100644 --- a/OpenCL/common.c +++ b/OpenCL/common.c @@ -5498,6 +5498,556 @@ static void overwrite_at_be (u32x sw[16], const u32x w0, const u32 salt_len) } } +static void overwrite_at_le_4x4 (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], const u32x wx, const u32 salt_len) +{ + #if defined cl_amd_media_ops + switch (salt_len) + { + case 0: w0[0] = wx; + break; + case 1: w0[0] = amd_bytealign (wx, w0[0] << 24, 3); + w0[1] = amd_bytealign (w0[1] >> 8, wx, 3); + break; + case 2: w0[0] = amd_bytealign (wx, w0[0] << 16, 2); + w0[1] = amd_bytealign (w0[1] >> 16, wx, 2); + break; + case 3: w0[0] = amd_bytealign (wx, w0[0] << 8, 1); + w0[1] = amd_bytealign (w0[1] >> 24, wx, 1); + break; + case 4: w0[1] = wx; + break; + case 5: w0[1] = amd_bytealign (wx, w0[1] << 24, 3); + w0[2] = amd_bytealign (w0[2] >> 8, wx, 3); + break; + case 6: w0[1] = amd_bytealign (wx, w0[1] << 16, 2); + w0[2] = amd_bytealign (w0[2] >> 16, wx, 2); + break; + case 7: w0[1] = amd_bytealign (wx, w0[1] << 8, 1); + w0[2] = amd_bytealign (w0[2] >> 24, wx, 1); + break; + case 8: w0[2] = wx; + break; + case 9: w0[2] = amd_bytealign (wx, w0[2] << 24, 3); + w0[3] = amd_bytealign (w0[3] >> 8, wx, 3); + break; + case 10: w0[2] = amd_bytealign (wx, w0[2] << 16, 2); + w0[3] = amd_bytealign (w0[3] >> 16, wx, 2); + break; + case 11: w0[2] = amd_bytealign (wx, w0[2] << 8, 1); + w0[3] = amd_bytealign (w0[3] >> 24, wx, 1); + break; + case 12: w0[3] = wx; + break; + case 13: w0[3] = amd_bytealign (wx, w0[3] << 24, 3); + w1[0] = amd_bytealign (w1[0] >> 8, wx, 3); + break; + case 14: w0[3] = amd_bytealign (wx, w0[3] << 16, 2); + w1[0] = amd_bytealign (w1[0] >> 16, wx, 2); + break; + case 15: w0[3] = amd_bytealign (wx, w0[3] << 8, 1); + w1[0] = amd_bytealign (w1[0] >> 24, wx, 1); + break; + case 16: w1[0] = wx; + break; + case 17: w1[0] = amd_bytealign (wx, w1[0] << 24, 3); + w1[1] = amd_bytealign (w1[1] >> 8, wx, 3); + break; + case 18: w1[0] = amd_bytealign (wx, w1[0] << 16, 2); + w1[1] = amd_bytealign (w1[1] >> 16, wx, 2); + break; + case 19: w1[0] = amd_bytealign (wx, w1[0] << 8, 1); + w1[1] = amd_bytealign (w1[1] >> 24, wx, 1); + break; + case 20: w1[1] = wx; + break; + case 21: w1[1] = amd_bytealign (wx, w1[1] << 24, 3); + w1[2] = amd_bytealign (w1[2] >> 8, wx, 3); + break; + case 22: w1[1] = amd_bytealign (wx, w1[1] << 16, 2); + w1[2] = amd_bytealign (w1[2] >> 16, wx, 2); + break; + case 23: w1[1] = amd_bytealign (wx, w1[1] << 8, 1); + w1[2] = amd_bytealign (w1[2] >> 24, wx, 1); + break; + case 24: w1[2] = wx; + break; + case 25: w1[2] = amd_bytealign (wx, w1[2] << 24, 3); + w1[3] = amd_bytealign (w1[3] >> 8, wx, 3); + break; + case 26: w1[2] = amd_bytealign (wx, w1[2] << 16, 2); + w1[3] = amd_bytealign (w1[3] >> 16, wx, 2); + break; + case 27: w1[2] = amd_bytealign (wx, w1[2] << 8, 1); + w1[3] = amd_bytealign (w1[3] >> 24, wx, 1); + break; + case 28: w1[3] = wx; + break; + case 29: w1[3] = amd_bytealign (wx, w1[3] << 24, 3); + w2[0] = amd_bytealign (w2[0] >> 8, wx, 3); + break; + case 30: w1[3] = amd_bytealign (wx, w1[3] << 16, 2); + w2[0] = amd_bytealign (w2[0] >> 16, wx, 2); + break; + case 31: w1[3] = amd_bytealign (wx, w1[3] << 8, 1); + w2[0] = amd_bytealign (w2[0] >> 24, wx, 1); + break; + case 32: w0[0] = wx; + break; + case 33: w2[0] = amd_bytealign (wx, w2[0] << 24, 3); + w2[1] = amd_bytealign (w2[1] >> 8, wx, 3); + break; + case 34: w2[0] = amd_bytealign (wx, w2[0] << 16, 2); + w2[1] = amd_bytealign (w2[1] >> 16, wx, 2); + break; + case 35: w2[0] = amd_bytealign (wx, w2[0] << 8, 1); + w2[1] = amd_bytealign (w2[1] >> 24, wx, 1); + break; + case 36: w2[1] = wx; + break; + case 37: w2[1] = amd_bytealign (wx, w2[1] << 24, 3); + w2[2] = amd_bytealign (w2[2] >> 8, wx, 3); + break; + case 38: w2[1] = amd_bytealign (wx, w2[1] << 16, 2); + w2[2] = amd_bytealign (w2[2] >> 16, wx, 2); + break; + case 39: w2[1] = amd_bytealign (wx, w2[1] << 8, 1); + w2[2] = amd_bytealign (w2[2] >> 24, wx, 1); + break; + case 40: w2[2] = wx; + break; + case 41: w2[2] = amd_bytealign (wx, w2[2] << 24, 3); + w2[3] = amd_bytealign (w2[3] >> 8, wx, 3); + break; + case 42: w2[2] = amd_bytealign (wx, w2[2] << 16, 2); + w2[3] = amd_bytealign (w2[3] >> 16, wx, 2); + break; + case 43: w2[2] = amd_bytealign (wx, w2[2] << 8, 1); + w2[3] = amd_bytealign (w2[3] >> 24, wx, 1); + break; + case 44: w2[3] = wx; + break; + case 45: w2[3] = amd_bytealign (wx, w2[3] << 24, 3); + w3[0] = amd_bytealign (w3[0] >> 8, wx, 3); + break; + case 46: w2[3] = amd_bytealign (wx, w2[3] << 16, 2); + w3[0] = amd_bytealign (w3[0] >> 16, wx, 2); + break; + case 47: w2[3] = amd_bytealign (wx, w2[3] << 8, 1); + w3[0] = amd_bytealign (w3[0] >> 24, wx, 1); + break; + case 48: w3[0] = wx; + break; + case 49: w3[0] = amd_bytealign (wx, w3[0] << 24, 3); + w3[1] = amd_bytealign (w3[1] >> 8, wx, 3); + break; + case 50: w3[0] = amd_bytealign (wx, w3[0] << 16, 2); + w3[1] = amd_bytealign (w3[1] >> 16, wx, 2); + break; + case 51: w3[0] = amd_bytealign (wx, w3[0] << 8, 1); + w3[1] = amd_bytealign (w3[1] >> 24, wx, 1); + break; + case 52: w3[1] = wx; + break; + case 53: w3[1] = amd_bytealign (wx, w3[1] << 24, 3); + w3[2] = amd_bytealign (w3[2] >> 8, wx, 3); + break; + case 54: w3[1] = amd_bytealign (wx, w3[1] << 16, 2); + w3[2] = amd_bytealign (w3[2] >> 16, wx, 2); + break; + case 55: w3[1] = amd_bytealign (wx, w3[1] << 8, 1); + w3[2] = amd_bytealign (w3[2] >> 24, wx, 1); + break; + case 56: w3[2] = wx; + break; + case 57: w3[2] = amd_bytealign (wx, w3[2] << 24, 3); + w3[3] = amd_bytealign (w3[3] >> 8, wx, 3); + break; + case 58: w3[2] = amd_bytealign (wx, w3[2] << 16, 2); + w3[3] = amd_bytealign (w3[3] >> 16, wx, 2); + break; + case 59: w3[2] = amd_bytealign (wx, w3[2] << 8, 1); + w3[3] = amd_bytealign (w3[3] >> 24, wx, 1); + break; + case 60: w3[3] = wx; + break; + case 61: w3[3] = amd_bytealign (wx, w3[3] << 24, 3); + //w4[0] = amd_bytealign (w4[0] >> 8, wx, 3); + break; + case 62: w3[3] = amd_bytealign (wx, w3[3] << 16, 2); + //w4[0] = amd_bytealign (w4[0] >> 16, wx, 2); + break; + case 63: w3[3] = amd_bytealign (wx, w3[3] << 8, 1); + //w4[0] = amd_bytealign (w4[0] >> 24, wx, 1); + break; + } + #else + switch (salt_len) + { + case 0: w0[0] = wx; + break; + case 1: w0[0] = (w0[0] & 0x000000ff) | (wx << 8); + w0[1] = (w0[1] & 0xffffff00) | (wx >> 24); + break; + case 2: w0[0] = (w0[0] & 0x0000ffff) | (wx << 16); + w0[1] = (w0[1] & 0xffff0000) | (wx >> 16); + break; + case 3: w0[0] = (w0[0] & 0x00ffffff) | (wx << 24); + w0[1] = (w0[1] & 0xff000000) | (wx >> 8); + break; + case 4: w0[1] = wx; + break; + case 5: w0[1] = (w0[1] & 0x000000ff) | (wx << 8); + w0[2] = (w0[2] & 0xffffff00) | (wx >> 24); + break; + case 6: w0[1] = (w0[1] & 0x0000ffff) | (wx << 16); + w0[2] = (w0[2] & 0xffff0000) | (wx >> 16); + break; + case 7: w0[1] = (w0[1] & 0x00ffffff) | (wx << 24); + w0[2] = (w0[2] & 0xff000000) | (wx >> 8); + break; + case 8: w0[2] = wx; + break; + case 9: w0[2] = (w0[2] & 0x000000ff) | (wx << 8); + w0[3] = (w0[3] & 0xffffff00) | (wx >> 24); + break; + case 10: w0[2] = (w0[2] & 0x0000ffff) | (wx << 16); + w0[3] = (w0[3] & 0xffff0000) | (wx >> 16); + break; + case 11: w0[2] = (w0[2] & 0x00ffffff) | (wx << 24); + w0[3] = (w0[3] & 0xff000000) | (wx >> 8); + break; + case 12: w0[3] = wx; + break; + case 13: w0[3] = (w0[3] & 0x000000ff) | (wx << 8); + w1[0] = (w1[0] & 0xffffff00) | (wx >> 24); + break; + case 14: w0[3] = (w0[3] & 0x0000ffff) | (wx << 16); + w1[0] = (w1[0] & 0xffff0000) | (wx >> 16); + break; + case 15: w0[3] = (w0[3] & 0x00ffffff) | (wx << 24); + w1[0] = (w1[0] & 0xff000000) | (wx >> 8); + break; + case 16: w1[0] = wx; + break; + case 17: w1[0] = (w1[0] & 0x000000ff) | (wx << 8); + w1[1] = (w1[1] & 0xffffff00) | (wx >> 24); + break; + case 18: w1[0] = (w1[0] & 0x0000ffff) | (wx << 16); + w1[1] = (w1[1] & 0xffff0000) | (wx >> 16); + break; + case 19: w1[0] = (w1[0] & 0x00ffffff) | (wx << 24); + w1[1] = (w1[1] & 0xff000000) | (wx >> 8); + break; + case 20: w1[1] = wx; + break; + case 21: w1[1] = (w1[1] & 0x000000ff) | (wx << 8); + w1[2] = (w1[2] & 0xffffff00) | (wx >> 24); + break; + case 22: w1[1] = (w1[1] & 0x0000ffff) | (wx << 16); + w1[2] = (w1[2] & 0xffff0000) | (wx >> 16); + break; + case 23: w1[1] = (w1[1] & 0x00ffffff) | (wx << 24); + w1[2] = (w1[2] & 0xff000000) | (wx >> 8); + break; + case 24: w1[2] = wx; + break; + case 25: w1[2] = (w1[2] & 0x000000ff) | (wx << 8); + w1[3] = (w1[3] & 0xffffff00) | (wx >> 24); + break; + case 26: w1[2] = (w1[2] & 0x0000ffff) | (wx << 16); + w1[3] = (w1[3] & 0xffff0000) | (wx >> 16); + break; + case 27: w1[2] = (w1[2] & 0x00ffffff) | (wx << 24); + w1[3] = (w1[3] & 0xff000000) | (wx >> 8); + break; + case 28: w1[3] = wx; + break; + case 29: w1[3] = (w1[3] & 0x000000ff) | (wx << 8); + w2[0] = (w2[0] & 0xffffff00) | (wx >> 24); + break; + case 30: w1[3] = (w1[3] & 0x0000ffff) | (wx << 16); + w2[0] = (w2[0] & 0xffff0000) | (wx >> 16); + break; + case 31: w1[3] = (w1[3] & 0x00ffffff) | (wx << 24); + w2[0] = (w2[0] & 0xff000000) | (wx >> 8); + break; + case 32: w2[0] = wx; + break; + case 33: w2[0] = (w2[0] & 0x000000ff) | (wx << 8); + w2[1] = (w2[1] & 0xffffff00) | (wx >> 24); + break; + case 34: w2[0] = (w2[0] & 0x0000ffff) | (wx << 16); + w2[1] = (w2[1] & 0xffff0000) | (wx >> 16); + break; + case 35: w2[0] = (w2[0] & 0x00ffffff) | (wx << 24); + w2[1] = (w2[1] & 0xff000000) | (wx >> 8); + break; + case 36: w2[1] = wx; + break; + case 37: w2[1] = (w2[1] & 0x000000ff) | (wx << 8); + w2[2] = (w2[2] & 0xffffff00) | (wx >> 24); + break; + case 38: w2[1] = (w2[1] & 0x0000ffff) | (wx << 16); + w2[2] = (w2[2] & 0xffff0000) | (wx >> 16); + break; + case 39: w2[1] = (w2[1] & 0x00ffffff) | (wx << 24); + w2[2] = (w2[2] & 0xff000000) | (wx >> 8); + break; + case 40: w2[2] = wx; + break; + case 41: w2[2] = (w2[2] & 0x000000ff) | (wx << 8); + w2[3] = (w2[3] & 0xffffff00) | (wx >> 24); + break; + case 42: w2[2] = (w2[2] & 0x0000ffff) | (wx << 16); + w2[3] = (w2[3] & 0xffff0000) | (wx >> 16); + break; + case 43: w2[2] = (w2[2] & 0x00ffffff) | (wx << 24); + w2[3] = (w2[3] & 0xff000000) | (wx >> 8); + break; + case 44: w2[3] = wx; + break; + case 45: w2[3] = (w2[3] & 0x000000ff) | (wx << 8); + w3[0] = (w3[0] & 0xffffff00) | (wx >> 24); + break; + case 46: w2[3] = (w2[3] & 0x0000ffff) | (wx << 16); + w3[0] = (w3[0] & 0xffff0000) | (wx >> 16); + break; + case 47: w2[3] = (w2[3] & 0x00ffffff) | (wx << 24); + w3[0] = (w3[0] & 0xff000000) | (wx >> 8); + break; + case 48: w3[0] = wx; + break; + case 49: w3[0] = (w3[0] & 0x000000ff) | (wx << 8); + w3[1] = (w3[1] & 0xffffff00) | (wx >> 24); + break; + case 50: w3[0] = (w3[0] & 0x0000ffff) | (wx << 16); + w3[1] = (w3[1] & 0xffff0000) | (wx >> 16); + break; + case 51: w3[0] = (w3[0] & 0x00ffffff) | (wx << 24); + w3[1] = (w3[1] & 0xff000000) | (wx >> 8); + break; + case 52: w3[1] = wx; + break; + case 53: w3[1] = (w3[1] & 0x000000ff) | (wx << 8); + w3[2] = (w3[2] & 0xffffff00) | (wx >> 24); + break; + case 54: w3[1] = (w3[1] & 0x0000ffff) | (wx << 16); + w3[2] = (w3[2] & 0xffff0000) | (wx >> 16); + break; + case 55: w3[1] = (w3[1] & 0x00ffffff) | (wx << 24); + w3[2] = (w3[2] & 0xff000000) | (wx >> 8); + break; + case 56: w3[2] = wx; + break; + case 57: w3[2] = (w3[2] & 0x000000ff) | (wx << 8); + w3[3] = (w3[3] & 0xffffff00) | (wx >> 24); + break; + case 58: w3[2] = (w3[2] & 0x0000ffff) | (wx << 16); + w3[3] = (w3[3] & 0xffff0000) | (wx >> 16); + break; + case 59: w3[2] = (w3[2] & 0x00ffffff) | (wx << 24); + w3[3] = (w3[3] & 0xff000000) | (wx >> 8); + break; + case 60: w3[3] = wx; + break; + case 61: w3[3] = (w3[3] & 0x000000ff) | (wx << 8); + //w4[0] = (w4[0] & 0xffffff00) | (wx >> 24); + break; + case 62: w3[3] = (w3[3] & 0x0000ffff) | (wx << 16); + //w4[0] = (w4[0] & 0xffff0000) | (wx >> 16); + break; + case 63: w3[3] = (w3[3] & 0x00ffffff) | (wx << 24); + //w4[0] = (w4[0] & 0xff000000) | (wx >> 8); + break; + } + #endif +} + +static void overwrite_at_be_4x4 (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], const u32x wx, const u32 salt_len) +{ + // would be nice to have optimization based on amd_bytealign as with _le counterpart + + switch (salt_len) + { + case 0: w0[0] = wx; + break; + case 1: w0[0] = (w0[0] & 0xff000000) | (wx >> 8); + w0[1] = (w0[1] & 0x00ffffff) | (wx << 24); + break; + case 2: w0[0] = (w0[0] & 0xffff0000) | (wx >> 16); + w0[1] = (w0[1] & 0x0000ffff) | (wx << 16); + break; + case 3: w0[0] = (w0[0] & 0xffffff00) | (wx >> 24); + w0[1] = (w0[1] & 0x000000ff) | (wx << 8); + break; + case 4: w0[1] = wx; + break; + case 5: w0[1] = (w0[1] & 0xff000000) | (wx >> 8); + w0[2] = (w0[2] & 0x00ffffff) | (wx << 24); + break; + case 6: w0[1] = (w0[1] & 0xffff0000) | (wx >> 16); + w0[2] = (w0[2] & 0x0000ffff) | (wx << 16); + break; + case 7: w0[1] = (w0[1] & 0xffffff00) | (wx >> 24); + w0[2] = (w0[2] & 0x000000ff) | (wx << 8); + break; + case 8: w0[2] = wx; + break; + case 9: w0[2] = (w0[2] & 0xff000000) | (wx >> 8); + w0[3] = (w0[3] & 0x00ffffff) | (wx << 24); + break; + case 10: w0[2] = (w0[2] & 0xffff0000) | (wx >> 16); + w0[3] = (w0[3] & 0x0000ffff) | (wx << 16); + break; + case 11: w0[2] = (w0[2] & 0xffffff00) | (wx >> 24); + w0[3] = (w0[3] & 0x000000ff) | (wx << 8); + break; + case 12: w0[3] = wx; + break; + case 13: w0[3] = (w0[3] & 0xff000000) | (wx >> 8); + w1[0] = (w1[0] & 0x00ffffff) | (wx << 24); + break; + case 14: w0[3] = (w0[3] & 0xffff0000) | (wx >> 16); + w1[0] = (w1[0] & 0x0000ffff) | (wx << 16); + break; + case 15: w0[3] = (w0[3] & 0xffffff00) | (wx >> 24); + w1[0] = (w1[0] & 0x000000ff) | (wx << 8); + break; + case 16: w1[0] = wx; + break; + case 17: w1[0] = (w1[0] & 0xff000000) | (wx >> 8); + w1[1] = (w1[1] & 0x00ffffff) | (wx << 24); + break; + case 18: w1[0] = (w1[0] & 0xffff0000) | (wx >> 16); + w1[1] = (w1[1] & 0x0000ffff) | (wx << 16); + break; + case 19: w1[0] = (w1[0] & 0xffffff00) | (wx >> 24); + w1[1] = (w1[1] & 0x000000ff) | (wx << 8); + break; + case 20: w1[1] = wx; + break; + case 21: w1[1] = (w1[1] & 0xff000000) | (wx >> 8); + w1[2] = (w1[2] & 0x00ffffff) | (wx << 24); + break; + case 22: w1[1] = (w1[1] & 0xffff0000) | (wx >> 16); + w1[2] = (w1[2] & 0x0000ffff) | (wx << 16); + break; + case 23: w1[1] = (w1[1] & 0xffffff00) | (wx >> 24); + w1[2] = (w1[2] & 0x000000ff) | (wx << 8); + break; + case 24: w1[2] = wx; + break; + case 25: w1[2] = (w1[2] & 0xff000000) | (wx >> 8); + w1[3] = (w1[3] & 0x00ffffff) | (wx << 24); + break; + case 26: w1[2] = (w1[2] & 0xffff0000) | (wx >> 16); + w1[3] = (w1[3] & 0x0000ffff) | (wx << 16); + break; + case 27: w1[2] = (w1[2] & 0xffffff00) | (wx >> 24); + w1[3] = (w1[3] & 0x000000ff) | (wx << 8); + break; + case 28: w1[3] = wx; + break; + case 29: w1[3] = (w1[3] & 0xff000000) | (wx >> 8); + w2[0] = (w2[0] & 0x00ffffff) | (wx << 24); + break; + case 30: w1[3] = (w1[3] & 0xffff0000) | (wx >> 16); + w2[0] = (w2[0] & 0x0000ffff) | (wx << 16); + break; + case 31: w1[3] = (w1[3] & 0xffffff00) | (wx >> 24); + w2[0] = (w2[0] & 0x000000ff) | (wx << 8); + break; + case 32: w2[0] = wx; + break; + case 33: w2[0] = (w2[0] & 0xff000000) | (wx >> 8); + w2[1] = (w2[1] & 0x00ffffff) | (wx << 24); + break; + case 34: w2[0] = (w2[0] & 0xffff0000) | (wx >> 16); + w2[1] = (w2[1] & 0x0000ffff) | (wx << 16); + break; + case 35: w2[0] = (w2[0] & 0xffffff00) | (wx >> 24); + w2[1] = (w2[1] & 0x000000ff) | (wx << 8); + break; + case 36: w2[1] = wx; + break; + case 37: w2[1] = (w2[1] & 0xff000000) | (wx >> 8); + w2[2] = (w2[2] & 0x00ffffff) | (wx << 24); + break; + case 38: w2[1] = (w2[1] & 0xffff0000) | (wx >> 16); + w2[2] = (w2[2] & 0x0000ffff) | (wx << 16); + break; + case 39: w2[1] = (w2[1] & 0xffffff00) | (wx >> 24); + w2[2] = (w2[2] & 0x000000ff) | (wx << 8); + break; + case 40: w2[2] = wx; + break; + case 41: w2[2] = (w2[2] & 0xff000000) | (wx >> 8); + w2[3] = (w2[3] & 0x00ffffff) | (wx << 24); + break; + case 42: w2[2] = (w2[2] & 0xffff0000) | (wx >> 16); + w2[3] = (w2[3] & 0x0000ffff) | (wx << 16); + break; + case 43: w2[2] = (w2[2] & 0xffffff00) | (wx >> 24); + w2[3] = (w2[3] & 0x000000ff) | (wx << 8); + break; + case 44: w2[3] = wx; + break; + case 45: w2[3] = (w2[3] & 0xff000000) | (wx >> 8); + w3[0] = (w3[0] & 0x00ffffff) | (wx << 24); + break; + case 46: w2[3] = (w2[3] & 0xffff0000) | (wx >> 16); + w3[0] = (w3[0] & 0x0000ffff) | (wx << 16); + break; + case 47: w2[3] = (w2[3] & 0xffffff00) | (wx >> 24); + w3[0] = (w3[0] & 0x000000ff) | (wx << 8); + break; + case 48: w3[0] = wx; + break; + case 49: w3[0] = (w3[0] & 0xff000000) | (wx >> 8); + w3[1] = (w3[1] & 0x00ffffff) | (wx << 24); + break; + case 50: w3[0] = (w3[0] & 0xffff0000) | (wx >> 16); + w3[1] = (w3[1] & 0x0000ffff) | (wx << 16); + break; + case 51: w3[0] = (w3[0] & 0xffffff00) | (wx >> 24); + w3[1] = (w3[1] & 0x000000ff) | (wx << 8); + break; + case 52: w3[1] = wx; + break; + case 53: w3[1] = (w3[1] & 0xff000000) | (wx >> 8); + w3[2] = (w3[2] & 0x00ffffff) | (wx << 24); + break; + case 54: w3[1] = (w3[1] & 0xffff0000) | (wx >> 16); + w3[2] = (w3[2] & 0x0000ffff) | (wx << 16); + break; + case 55: w3[1] = (w3[1] & 0xffffff00) | (wx >> 24); + w3[2] = (w3[2] & 0x000000ff) | (wx << 8); + break; + case 56: w3[2] = wx; + break; + case 57: w3[2] = (w3[2] & 0xff000000) | (wx >> 8); + w3[3] = (w3[3] & 0x00ffffff) | (wx << 24); + break; + case 58: w3[2] = (w3[2] & 0xffff0000) | (wx >> 16); + w3[3] = (w3[3] & 0x0000ffff) | (wx << 16); + break; + case 59: w3[2] = (w3[2] & 0xffffff00) | (wx >> 24); + w3[3] = (w3[3] & 0x000000ff) | (wx << 8); + break; + case 60: w3[3] = wx; + break; + case 61: w3[3] = (w3[3] & 0xff000000) | (wx >> 8); + //w4[0] = (w4[0] & 0x00ffffff) | (wx << 24); + break; + case 62: w3[3] = (w3[3] & 0xffff0000) | (wx >> 16); + //w4[0] = (w4[0] & 0x0000ffff) | (wx << 16); + break; + case 63: w3[3] = (w3[3] & 0xffffff00) | (wx >> 24); + //w4[0] = (w4[0] & 0x000000ff) | (wx << 8); + break; + } +} + /** * vector functions as scalar (for outer loop usage) */ diff --git a/OpenCL/m00020_a3.cl b/OpenCL/m00020_a3.cl index 1c6d8b1..742905c 100644 --- a/OpenCL/m00020_a3.cl +++ b/OpenCL/m00020_a3.cl @@ -20,7 +20,7 @@ #include "OpenCL/common.c" #include "OpenCL/simd.c" -static void m00020m (u32 t0[4], u32 t1[4], u32 t2[4], u32 t3[4], const u32 pw_len, __global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global void *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global void *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 bfs_cnt, const u32 digests_cnt, const u32 digests_offset) +static void m00020m (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 pw_len, __global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global void *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global void *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 bfs_cnt, const u32 digests_cnt, const u32 digests_offset) { /** * modifier @@ -63,101 +63,60 @@ static void m00020m (u32 t0[4], u32 t1[4], u32 t2[4], u32 t3[4], const u32 pw_le * prepend salt */ - u32 w0_t[4]; - u32 w1_t[4]; - u32 w2_t[4]; - u32 w3_t[4]; - - w0_t[0] = t0[0]; - w0_t[1] = t0[1]; - w0_t[2] = t0[2]; - w0_t[3] = t0[3]; - w1_t[0] = t1[0]; - w1_t[1] = t1[1]; - w1_t[2] = t1[2]; - w1_t[3] = t1[3]; - w2_t[0] = t2[0]; - w2_t[1] = t2[1]; - w2_t[2] = t2[2]; - w2_t[3] = t2[3]; - w3_t[0] = t3[0]; - w3_t[1] = t3[1]; - w3_t[2] = t3[2]; - w3_t[3] = t3[3]; - - switch_buffer_by_offset_le_S (w0_t, w1_t, w2_t, w3_t, salt_len); - - w0_t[0] |= salt_buf0[0]; - w0_t[1] |= salt_buf0[1]; - w0_t[2] |= salt_buf0[2]; - w0_t[3] |= salt_buf0[3]; - w1_t[0] |= salt_buf1[0]; - w1_t[1] |= salt_buf1[1]; - w1_t[2] |= salt_buf1[2]; - w1_t[3] |= salt_buf1[3]; - w2_t[0] |= salt_buf2[0]; - w2_t[1] |= salt_buf2[1]; - w2_t[2] |= salt_buf2[2]; - w2_t[3] |= salt_buf2[3]; - w3_t[0] |= salt_buf3[0]; - w3_t[1] |= salt_buf3[1]; - w3_t[2] |= salt_buf3[2]; - w3_t[3] |= salt_buf3[3]; + const u32 w0l = w0[0]; + + switch_buffer_by_offset_le_S (w0, w1, w2, w3, salt_len); + + w0[0] |= salt_buf0[0]; + w0[1] |= salt_buf0[1]; + w0[2] |= salt_buf0[2]; + w0[3] |= salt_buf0[3]; + w1[0] |= salt_buf1[0]; + w1[1] |= salt_buf1[1]; + w1[2] |= salt_buf1[2]; + w1[3] |= salt_buf1[3]; + w2[0] |= salt_buf2[0]; + w2[1] |= salt_buf2[1]; + w2[2] |= salt_buf2[2]; + w2[3] |= salt_buf2[3]; + w3[0] |= salt_buf3[0]; + w3[1] |= salt_buf3[1]; + w3[2] |= salt_buf3[2]; + w3[3] |= salt_buf3[3]; /** * loop */ - u32 w0l = t0[0]; - for (u32 il_pos = 0; il_pos < bfs_cnt; il_pos += VECT_SIZE) { const u32x w0r = ix_create_bft (bfs_buf, il_pos); const u32x w0lr = w0l | w0r; - u32x wx[16]; - - wx[ 0] = w0_t[0]; - wx[ 1] = w0_t[1]; - wx[ 2] = w0_t[2]; - wx[ 3] = w0_t[3]; - wx[ 4] = w1_t[0]; - wx[ 5] = w1_t[1]; - wx[ 6] = w1_t[2]; - wx[ 7] = w1_t[3]; - wx[ 8] = w2_t[0]; - wx[ 9] = w2_t[1]; - wx[10] = w2_t[2]; - wx[11] = w2_t[3]; - wx[12] = w3_t[0]; - wx[13] = w3_t[1]; - wx[14] = w3_t[2]; - wx[15] = w3_t[3]; - - overwrite_at_le (wx, w0lr, salt_len); - - u32x w0[4]; - u32x w1[4]; - u32x w2[4]; - u32x w3[4]; - - w0[0] = wx[ 0]; - w0[1] = wx[ 1]; - w0[2] = wx[ 2]; - w0[3] = wx[ 3]; - w1[0] = wx[ 4]; - w1[1] = wx[ 5]; - w1[2] = wx[ 6]; - w1[3] = wx[ 7]; - w2[0] = wx[ 8]; - w2[1] = wx[ 9]; - w2[2] = wx[10]; - w2[3] = wx[11]; - w3[0] = wx[12]; - w3[1] = wx[13]; - w3[2] = pw_salt_len * 8; - w3[3] = 0; + u32x w0_t[4]; + u32x w1_t[4]; + u32x w2_t[4]; + u32x w3_t[4]; + + w0_t[0] = w0[0]; + w0_t[1] = w0[1]; + w0_t[2] = w0[2]; + w0_t[3] = w0[3]; + w1_t[0] = w1[0]; + w1_t[1] = w1[1]; + w1_t[2] = w1[2]; + w1_t[3] = w1[3]; + w2_t[0] = w2[0]; + w2_t[1] = w2[1]; + w2_t[2] = w2[2]; + w2_t[3] = w2[3]; + w3_t[0] = w3[0]; + w3_t[1] = w3[1]; + w3_t[2] = pw_salt_len * 8; + w3_t[3] = 0; + + overwrite_at_le_4x4 (w0_t, w1_t, w2_t, w3_t, w0lr, salt_len); /** * md5 @@ -168,73 +127,73 @@ static void m00020m (u32 t0[4], u32 t1[4], u32 t2[4], u32 t3[4], const u32 pw_le u32x c = MD5M_C; u32x d = MD5M_D; - MD5_STEP (MD5_Fo, a, b, c, d, w0[0], MD5C00, MD5S00); - MD5_STEP (MD5_Fo, d, a, b, c, w0[1], MD5C01, MD5S01); - MD5_STEP (MD5_Fo, c, d, a, b, w0[2], MD5C02, MD5S02); - MD5_STEP (MD5_Fo, b, c, d, a, w0[3], MD5C03, MD5S03); - MD5_STEP (MD5_Fo, a, b, c, d, w1[0], MD5C04, MD5S00); - MD5_STEP (MD5_Fo, d, a, b, c, w1[1], MD5C05, MD5S01); - MD5_STEP (MD5_Fo, c, d, a, b, w1[2], MD5C06, MD5S02); - MD5_STEP (MD5_Fo, b, c, d, a, w1[3], MD5C07, MD5S03); - MD5_STEP (MD5_Fo, a, b, c, d, w2[0], MD5C08, MD5S00); - MD5_STEP (MD5_Fo, d, a, b, c, w2[1], MD5C09, MD5S01); - MD5_STEP (MD5_Fo, c, d, a, b, w2[2], MD5C0a, MD5S02); - MD5_STEP (MD5_Fo, b, c, d, a, w2[3], MD5C0b, MD5S03); - MD5_STEP (MD5_Fo, a, b, c, d, w3[0], MD5C0c, MD5S00); - MD5_STEP (MD5_Fo, d, a, b, c, w3[1], MD5C0d, MD5S01); - MD5_STEP (MD5_Fo, c, d, a, b, w3[2], MD5C0e, MD5S02); - MD5_STEP (MD5_Fo, b, c, d, a, w3[3], MD5C0f, MD5S03); - - MD5_STEP (MD5_Go, a, b, c, d, w0[1], MD5C10, MD5S10); - MD5_STEP (MD5_Go, d, a, b, c, w1[2], MD5C11, MD5S11); - MD5_STEP (MD5_Go, c, d, a, b, w2[3], MD5C12, MD5S12); - MD5_STEP (MD5_Go, b, c, d, a, w0[0], MD5C13, MD5S13); - MD5_STEP (MD5_Go, a, b, c, d, w1[1], MD5C14, MD5S10); - MD5_STEP (MD5_Go, d, a, b, c, w2[2], MD5C15, MD5S11); - MD5_STEP (MD5_Go, c, d, a, b, w3[3], MD5C16, MD5S12); - MD5_STEP (MD5_Go, b, c, d, a, w1[0], MD5C17, MD5S13); - MD5_STEP (MD5_Go, a, b, c, d, w2[1], MD5C18, MD5S10); - MD5_STEP (MD5_Go, d, a, b, c, w3[2], MD5C19, MD5S11); - MD5_STEP (MD5_Go, c, d, a, b, w0[3], MD5C1a, MD5S12); - MD5_STEP (MD5_Go, b, c, d, a, w2[0], MD5C1b, MD5S13); - MD5_STEP (MD5_Go, a, b, c, d, w3[1], MD5C1c, MD5S10); - MD5_STEP (MD5_Go, d, a, b, c, w0[2], MD5C1d, MD5S11); - MD5_STEP (MD5_Go, c, d, a, b, w1[3], MD5C1e, MD5S12); - MD5_STEP (MD5_Go, b, c, d, a, w3[0], MD5C1f, MD5S13); - - MD5_STEP (MD5_H , a, b, c, d, w1[1], MD5C20, MD5S20); - MD5_STEP (MD5_H , d, a, b, c, w2[0], MD5C21, MD5S21); - MD5_STEP (MD5_H , c, d, a, b, w2[3], MD5C22, MD5S22); - MD5_STEP (MD5_H , b, c, d, a, w3[2], MD5C23, MD5S23); - MD5_STEP (MD5_H , a, b, c, d, w0[1], MD5C24, MD5S20); - MD5_STEP (MD5_H , d, a, b, c, w1[0], MD5C25, MD5S21); - MD5_STEP (MD5_H , c, d, a, b, w1[3], MD5C26, MD5S22); - MD5_STEP (MD5_H , b, c, d, a, w2[2], MD5C27, MD5S23); - MD5_STEP (MD5_H , a, b, c, d, w3[1], MD5C28, MD5S20); - MD5_STEP (MD5_H , d, a, b, c, w0[0], MD5C29, MD5S21); - MD5_STEP (MD5_H , c, d, a, b, w0[3], MD5C2a, MD5S22); - MD5_STEP (MD5_H , b, c, d, a, w1[2], MD5C2b, MD5S23); - MD5_STEP (MD5_H , a, b, c, d, w2[1], MD5C2c, MD5S20); - MD5_STEP (MD5_H , d, a, b, c, w3[0], MD5C2d, MD5S21); - MD5_STEP (MD5_H , c, d, a, b, w3[3], MD5C2e, MD5S22); - MD5_STEP (MD5_H , b, c, d, a, w0[2], MD5C2f, MD5S23); - - MD5_STEP (MD5_I , a, b, c, d, w0[0], MD5C30, MD5S30); - MD5_STEP (MD5_I , d, a, b, c, w1[3], MD5C31, MD5S31); - MD5_STEP (MD5_I , c, d, a, b, w3[2], MD5C32, MD5S32); - MD5_STEP (MD5_I , b, c, d, a, w1[1], MD5C33, MD5S33); - MD5_STEP (MD5_I , a, b, c, d, w3[0], MD5C34, MD5S30); - MD5_STEP (MD5_I , d, a, b, c, w0[3], MD5C35, MD5S31); - MD5_STEP (MD5_I , c, d, a, b, w2[2], MD5C36, MD5S32); - MD5_STEP (MD5_I , b, c, d, a, w0[1], MD5C37, MD5S33); - MD5_STEP (MD5_I , a, b, c, d, w2[0], MD5C38, MD5S30); - MD5_STEP (MD5_I , d, a, b, c, w3[3], MD5C39, MD5S31); - MD5_STEP (MD5_I , c, d, a, b, w1[2], MD5C3a, MD5S32); - MD5_STEP (MD5_I , b, c, d, a, w3[1], MD5C3b, MD5S33); - MD5_STEP (MD5_I , a, b, c, d, w1[0], MD5C3c, MD5S30); - MD5_STEP (MD5_I , d, a, b, c, w2[3], MD5C3d, MD5S31); - MD5_STEP (MD5_I , c, d, a, b, w0[2], MD5C3e, MD5S32); - MD5_STEP (MD5_I , b, c, d, a, w2[1], MD5C3f, MD5S33); + MD5_STEP (MD5_Fo, a, b, c, d, w0_t[0], MD5C00, MD5S00); + MD5_STEP (MD5_Fo, d, a, b, c, w0_t[1], MD5C01, MD5S01); + MD5_STEP (MD5_Fo, c, d, a, b, w0_t[2], MD5C02, MD5S02); + MD5_STEP (MD5_Fo, b, c, d, a, w0_t[3], MD5C03, MD5S03); + MD5_STEP (MD5_Fo, a, b, c, d, w1_t[0], MD5C04, MD5S00); + MD5_STEP (MD5_Fo, d, a, b, c, w1_t[1], MD5C05, MD5S01); + MD5_STEP (MD5_Fo, c, d, a, b, w1_t[2], MD5C06, MD5S02); + MD5_STEP (MD5_Fo, b, c, d, a, w1_t[3], MD5C07, MD5S03); + MD5_STEP (MD5_Fo, a, b, c, d, w2_t[0], MD5C08, MD5S00); + MD5_STEP (MD5_Fo, d, a, b, c, w2_t[1], MD5C09, MD5S01); + MD5_STEP (MD5_Fo, c, d, a, b, w2_t[2], MD5C0a, MD5S02); + MD5_STEP (MD5_Fo, b, c, d, a, w2_t[3], MD5C0b, MD5S03); + MD5_STEP (MD5_Fo, a, b, c, d, w3_t[0], MD5C0c, MD5S00); + MD5_STEP (MD5_Fo, d, a, b, c, w3_t[1], MD5C0d, MD5S01); + MD5_STEP (MD5_Fo, c, d, a, b, w3_t[2], MD5C0e, MD5S02); + MD5_STEP (MD5_Fo, b, c, d, a, w3_t[3], MD5C0f, MD5S03); + + MD5_STEP (MD5_Go, a, b, c, d, w0_t[1], MD5C10, MD5S10); + MD5_STEP (MD5_Go, d, a, b, c, w1_t[2], MD5C11, MD5S11); + MD5_STEP (MD5_Go, c, d, a, b, w2_t[3], MD5C12, MD5S12); + MD5_STEP (MD5_Go, b, c, d, a, w0_t[0], MD5C13, MD5S13); + MD5_STEP (MD5_Go, a, b, c, d, w1_t[1], MD5C14, MD5S10); + MD5_STEP (MD5_Go, d, a, b, c, w2_t[2], MD5C15, MD5S11); + MD5_STEP (MD5_Go, c, d, a, b, w3_t[3], MD5C16, MD5S12); + MD5_STEP (MD5_Go, b, c, d, a, w1_t[0], MD5C17, MD5S13); + MD5_STEP (MD5_Go, a, b, c, d, w2_t[1], MD5C18, MD5S10); + MD5_STEP (MD5_Go, d, a, b, c, w3_t[2], MD5C19, MD5S11); + MD5_STEP (MD5_Go, c, d, a, b, w0_t[3], MD5C1a, MD5S12); + MD5_STEP (MD5_Go, b, c, d, a, w2_t[0], MD5C1b, MD5S13); + MD5_STEP (MD5_Go, a, b, c, d, w3_t[1], MD5C1c, MD5S10); + MD5_STEP (MD5_Go, d, a, b, c, w0_t[2], MD5C1d, MD5S11); + MD5_STEP (MD5_Go, c, d, a, b, w1_t[3], MD5C1e, MD5S12); + MD5_STEP (MD5_Go, b, c, d, a, w3_t[0], MD5C1f, MD5S13); + + MD5_STEP (MD5_H , a, b, c, d, w1_t[1], MD5C20, MD5S20); + MD5_STEP (MD5_H , d, a, b, c, w2_t[0], MD5C21, MD5S21); + MD5_STEP (MD5_H , c, d, a, b, w2_t[3], MD5C22, MD5S22); + MD5_STEP (MD5_H , b, c, d, a, w3_t[2], MD5C23, MD5S23); + MD5_STEP (MD5_H , a, b, c, d, w0_t[1], MD5C24, MD5S20); + MD5_STEP (MD5_H , d, a, b, c, w1_t[0], MD5C25, MD5S21); + MD5_STEP (MD5_H , c, d, a, b, w1_t[3], MD5C26, MD5S22); + MD5_STEP (MD5_H , b, c, d, a, w2_t[2], MD5C27, MD5S23); + MD5_STEP (MD5_H , a, b, c, d, w3_t[1], MD5C28, MD5S20); + MD5_STEP (MD5_H , d, a, b, c, w0_t[0], MD5C29, MD5S21); + MD5_STEP (MD5_H , c, d, a, b, w0_t[3], MD5C2a, MD5S22); + MD5_STEP (MD5_H , b, c, d, a, w1_t[2], MD5C2b, MD5S23); + MD5_STEP (MD5_H , a, b, c, d, w2_t[1], MD5C2c, MD5S20); + MD5_STEP (MD5_H , d, a, b, c, w3_t[0], MD5C2d, MD5S21); + MD5_STEP (MD5_H , c, d, a, b, w3_t[3], MD5C2e, MD5S22); + MD5_STEP (MD5_H , b, c, d, a, w0_t[2], MD5C2f, MD5S23); + + MD5_STEP (MD5_I , a, b, c, d, w0_t[0], MD5C30, MD5S30); + MD5_STEP (MD5_I , d, a, b, c, w1_t[3], MD5C31, MD5S31); + MD5_STEP (MD5_I , c, d, a, b, w3_t[2], MD5C32, MD5S32); + MD5_STEP (MD5_I , b, c, d, a, w1_t[1], MD5C33, MD5S33); + MD5_STEP (MD5_I , a, b, c, d, w3_t[0], MD5C34, MD5S30); + MD5_STEP (MD5_I , d, a, b, c, w0_t[3], MD5C35, MD5S31); + MD5_STEP (MD5_I , c, d, a, b, w2_t[2], MD5C36, MD5S32); + MD5_STEP (MD5_I , b, c, d, a, w0_t[1], MD5C37, MD5S33); + MD5_STEP (MD5_I , a, b, c, d, w2_t[0], MD5C38, MD5S30); + MD5_STEP (MD5_I , d, a, b, c, w3_t[3], MD5C39, MD5S31); + MD5_STEP (MD5_I , c, d, a, b, w1_t[2], MD5C3a, MD5S32); + MD5_STEP (MD5_I , b, c, d, a, w3_t[1], MD5C3b, MD5S33); + MD5_STEP (MD5_I , a, b, c, d, w1_t[0], MD5C3c, MD5S30); + MD5_STEP (MD5_I , d, a, b, c, w2_t[3], MD5C3d, MD5S31); + MD5_STEP (MD5_I , c, d, a, b, w0_t[2], MD5C3e, MD5S32); + MD5_STEP (MD5_I , b, c, d, a, w2_t[1], MD5C3f, MD5S33); COMPARE_M_SIMD (a, d, c, b); } @@ -295,102 +254,61 @@ static void m00020s (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], const u32 pw_le * prepend salt */ - u32 w0_t[4]; - u32 w1_t[4]; - u32 w2_t[4]; - u32 w3_t[4]; - - w0_t[0] = w0[0]; - w0_t[1] = w0[1]; - w0_t[2] = w0[2]; - w0_t[3] = w0[3]; - w1_t[0] = w1[0]; - w1_t[1] = w1[1]; - w1_t[2] = w1[2]; - w1_t[3] = w1[3]; - w2_t[0] = w2[0]; - w2_t[1] = w2[1]; - w2_t[2] = w2[2]; - w2_t[3] = w2[3]; - w3_t[0] = w3[0]; - w3_t[1] = w3[1]; - w3_t[2] = w3[2]; - w3_t[3] = w3[3]; - - switch_buffer_by_offset_le_S (w0_t, w1_t, w2_t, w3_t, salt_len); - - w0_t[0] |= salt_buf0[0]; - w0_t[1] |= salt_buf0[1]; - w0_t[2] |= salt_buf0[2]; - w0_t[3] |= salt_buf0[3]; - w1_t[0] |= salt_buf1[0]; - w1_t[1] |= salt_buf1[1]; - w1_t[2] |= salt_buf1[2]; - w1_t[3] |= salt_buf1[3]; - w2_t[0] |= salt_buf2[0]; - w2_t[1] |= salt_buf2[1]; - w2_t[2] |= salt_buf2[2]; - w2_t[3] |= salt_buf2[3]; - w3_t[0] |= salt_buf3[0]; - w3_t[1] |= salt_buf3[1]; - w3_t[2] |= salt_buf3[2]; - w3_t[3] |= salt_buf3[3]; + const u32 w0l = w0[0]; + + switch_buffer_by_offset_le_S (w0, w1, w2, w3, salt_len); + + w0[0] |= salt_buf0[0]; + w0[1] |= salt_buf0[1]; + w0[2] |= salt_buf0[2]; + w0[3] |= salt_buf0[3]; + w1[0] |= salt_buf1[0]; + w1[1] |= salt_buf1[1]; + w1[2] |= salt_buf1[2]; + w1[3] |= salt_buf1[3]; + w2[0] |= salt_buf2[0]; + w2[1] |= salt_buf2[1]; + w2[2] |= salt_buf2[2]; + w2[3] |= salt_buf2[3]; + w3[0] |= salt_buf3[0]; + w3[1] |= salt_buf3[1]; + w3[2] |= salt_buf3[2]; + w3[3] |= salt_buf3[3]; /** * loop */ - u32 w0l = w0[0]; - for (u32 il_pos = 0; il_pos < bfs_cnt; il_pos += VECT_SIZE) { const u32x w0r = ix_create_bft (bfs_buf, il_pos); const u32x w0lr = w0l | w0r; - u32x wx[16]; - - wx[ 0] = w0_t[0]; - wx[ 1] = w0_t[1]; - wx[ 2] = w0_t[2]; - wx[ 3] = w0_t[3]; - wx[ 4] = w1_t[0]; - wx[ 5] = w1_t[1]; - wx[ 6] = w1_t[2]; - wx[ 7] = w1_t[3]; - wx[ 8] = w2_t[0]; - wx[ 9] = w2_t[1]; - wx[10] = w2_t[2]; - wx[11] = w2_t[3]; - wx[12] = w3_t[0]; - wx[13] = w3_t[1]; - wx[14] = w3_t[2]; - wx[15] = w3_t[3]; - - overwrite_at_le (wx, w0lr, salt_len); - u32x w0_t[4]; u32x w1_t[4]; u32x w2_t[4]; u32x w3_t[4]; - w0_t[0] = wx[ 0]; - w0_t[1] = wx[ 1]; - w0_t[2] = wx[ 2]; - w0_t[3] = wx[ 3]; - w1_t[0] = wx[ 4]; - w1_t[1] = wx[ 5]; - w1_t[2] = wx[ 6]; - w1_t[3] = wx[ 7]; - w2_t[0] = wx[ 8]; - w2_t[1] = wx[ 9]; - w2_t[2] = wx[10]; - w2_t[3] = wx[11]; - w3_t[0] = wx[12]; - w3_t[1] = wx[13]; + w0_t[0] = w0[0]; + w0_t[1] = w0[1]; + w0_t[2] = w0[2]; + w0_t[3] = w0[3]; + w1_t[0] = w1[0]; + w1_t[1] = w1[1]; + w1_t[2] = w1[2]; + w1_t[3] = w1[3]; + w2_t[0] = w2[0]; + w2_t[1] = w2[1]; + w2_t[2] = w2[2]; + w2_t[3] = w2[3]; + w3_t[0] = w3[0]; + w3_t[1] = w3[1]; w3_t[2] = pw_salt_len * 8; w3_t[3] = 0; + overwrite_at_le_4x4 (w0_t, w1_t, w2_t, w3_t, w0lr, salt_len); + /** * md5 */ -- 2.25.1