I had not anticipated this being really worthwhile, but it turns out

to have been so! Change MDx_HashFunction::hash to a new compress_n
which hashes an arbitrary number of blocks. I had a thought this might
reduce a bit of loop overhead but the results were far better than I
anticipated. Speedup across the board of about 2%, and very
noticable (+10%) increases for MD4 and Tiger (probably b/c both
of those have so few instructions in each iteration of the
compression function).

Before:
SHA-1:
  amd64: 211.9 MiB/s
  core: 210.0 MiB/s
  sse2: 295.2 MiB/s

MD4: 476.2 MiB/s
MD5: 355.2 MiB/s
SHA-256: 99.8 MiB/s
SHA-512: 151.4 MiB/s
RIPEMD-128: 326.9 MiB/s
RIPEMD-160: 225.1 MiB/s
Tiger: 214.8 MiB/s
Whirlpool: 38.4 MiB/s

After:
SHA-1:
  amd64: 215.6 MiB/s
  core: 213.8 MiB/s
  sse2: 299.9 MiB/s

MD4: 528.4 MiB/s
MD5: 368.8 MiB/s
SHA-256: 103.9 MiB/s
SHA-512: 156.8 MiB/s
RIPEMD-128: 334.8 MiB/s
RIPEMD-160: 229.7 MiB/s
Tiger: 240.7 MiB/s
Whirlpool: 38.6 MiB/s

1 files changed, 56 insertions, 52 deletions

diff --git a/src/hash/sha1/sha160.cpp b/src/hash/sha1/sha160.cpp
index 142cddcd3..302151921 100644
--- a/src/hash/sha1/sha160.cpp
+++ b/src/hash/sha1/sha160.cpp
@@ -52,61 +52,65 @@ inline void F4(u32bit A, u32bit& B, u32bit C, u32bit D, u32bit& E, u32bit msg)
 /*************************************************
 * SHA-160 Compression Function                   *
 *************************************************/
-void SHA_160::hash(const byte input[])
+void SHA_160::compress_n(const byte input[], u32bit blocks)
    {
-   for(u32bit j = 0; j != 16; j += 4)
+   for(u32bit i = 0; i != blocks; ++i)
       {
-      W[j  ] = load_be<u32bit>(input, j);
-      W[j+1] = load_be<u32bit>(input, j+1);
-      W[j+2] = load_be<u32bit>(input, j+2);
-      W[j+3] = load_be<u32bit>(input, j+3);
+      for(u32bit j = 0; j != 16; j += 4)
+         {
+         W[j  ] = load_be<u32bit>(input, j);
+         W[j+1] = load_be<u32bit>(input, j+1);
+         W[j+2] = load_be<u32bit>(input, j+2);
+         W[j+3] = load_be<u32bit>(input, j+3);
+         }
+      input += HASH_BLOCK_SIZE;
+
+      for(u32bit j = 16; j != 80; j += 4)
+         {
+         W[j  ] = rotate_left((W[j-3] ^ W[j-8] ^ W[j-14] ^ W[j-16]), 1);
+         W[j+1] = rotate_left((W[j-2] ^ W[j-7] ^ W[j-13] ^ W[j-15]), 1);
+         W[j+2] = rotate_left((W[j-1] ^ W[j-6] ^ W[j-12] ^ W[j-14]), 1);
+         W[j+3] = rotate_left((W[j  ] ^ W[j-5] ^ W[j-11] ^ W[j-13]), 1);
+         }
+
+      u32bit A = digest[0], B = digest[1], C = digest[2],
+             D = digest[3], E = digest[4];
+
+      F1(A,B,C,D,E,W[ 0]);   F1(E,A,B,C,D,W[ 1]);   F1(D,E,A,B,C,W[ 2]);
+      F1(C,D,E,A,B,W[ 3]);   F1(B,C,D,E,A,W[ 4]);   F1(A,B,C,D,E,W[ 5]);
+      F1(E,A,B,C,D,W[ 6]);   F1(D,E,A,B,C,W[ 7]);   F1(C,D,E,A,B,W[ 8]);
+      F1(B,C,D,E,A,W[ 9]);   F1(A,B,C,D,E,W[10]);   F1(E,A,B,C,D,W[11]);
+      F1(D,E,A,B,C,W[12]);   F1(C,D,E,A,B,W[13]);   F1(B,C,D,E,A,W[14]);
+      F1(A,B,C,D,E,W[15]);   F1(E,A,B,C,D,W[16]);   F1(D,E,A,B,C,W[17]);
+      F1(C,D,E,A,B,W[18]);   F1(B,C,D,E,A,W[19]);
+
+      F2(A,B,C,D,E,W[20]);   F2(E,A,B,C,D,W[21]);   F2(D,E,A,B,C,W[22]);
+      F2(C,D,E,A,B,W[23]);   F2(B,C,D,E,A,W[24]);   F2(A,B,C,D,E,W[25]);
+      F2(E,A,B,C,D,W[26]);   F2(D,E,A,B,C,W[27]);   F2(C,D,E,A,B,W[28]);
+      F2(B,C,D,E,A,W[29]);   F2(A,B,C,D,E,W[30]);   F2(E,A,B,C,D,W[31]);
+      F2(D,E,A,B,C,W[32]);   F2(C,D,E,A,B,W[33]);   F2(B,C,D,E,A,W[34]);
+      F2(A,B,C,D,E,W[35]);   F2(E,A,B,C,D,W[36]);   F2(D,E,A,B,C,W[37]);
+      F2(C,D,E,A,B,W[38]);   F2(B,C,D,E,A,W[39]);
+
+      F3(A,B,C,D,E,W[40]);   F3(E,A,B,C,D,W[41]);   F3(D,E,A,B,C,W[42]);
+      F3(C,D,E,A,B,W[43]);   F3(B,C,D,E,A,W[44]);   F3(A,B,C,D,E,W[45]);
+      F3(E,A,B,C,D,W[46]);   F3(D,E,A,B,C,W[47]);   F3(C,D,E,A,B,W[48]);
+      F3(B,C,D,E,A,W[49]);   F3(A,B,C,D,E,W[50]);   F3(E,A,B,C,D,W[51]);
+      F3(D,E,A,B,C,W[52]);   F3(C,D,E,A,B,W[53]);   F3(B,C,D,E,A,W[54]);
+      F3(A,B,C,D,E,W[55]);   F3(E,A,B,C,D,W[56]);   F3(D,E,A,B,C,W[57]);
+      F3(C,D,E,A,B,W[58]);   F3(B,C,D,E,A,W[59]);
+
+      F4(A,B,C,D,E,W[60]);   F4(E,A,B,C,D,W[61]);   F4(D,E,A,B,C,W[62]);
+      F4(C,D,E,A,B,W[63]);   F4(B,C,D,E,A,W[64]);   F4(A,B,C,D,E,W[65]);
+      F4(E,A,B,C,D,W[66]);   F4(D,E,A,B,C,W[67]);   F4(C,D,E,A,B,W[68]);
+      F4(B,C,D,E,A,W[69]);   F4(A,B,C,D,E,W[70]);   F4(E,A,B,C,D,W[71]);
+      F4(D,E,A,B,C,W[72]);   F4(C,D,E,A,B,W[73]);   F4(B,C,D,E,A,W[74]);
+      F4(A,B,C,D,E,W[75]);   F4(E,A,B,C,D,W[76]);   F4(D,E,A,B,C,W[77]);
+      F4(C,D,E,A,B,W[78]);   F4(B,C,D,E,A,W[79]);
+
+      digest[0] += A; digest[1] += B; digest[2] += C;
+      digest[3] += D; digest[4] += E;
       }
-
-   for(u32bit j = 16; j != 80; j += 4)
-      {
-      W[j  ] = rotate_left((W[j-3] ^ W[j-8] ^ W[j-14] ^ W[j-16]), 1);
-      W[j+1] = rotate_left((W[j-2] ^ W[j-7] ^ W[j-13] ^ W[j-15]), 1);
-      W[j+2] = rotate_left((W[j-1] ^ W[j-6] ^ W[j-12] ^ W[j-14]), 1);
-      W[j+3] = rotate_left((W[j  ] ^ W[j-5] ^ W[j-11] ^ W[j-13]), 1);
-      }
-
-   u32bit A = digest[0], B = digest[1], C = digest[2],
-          D = digest[3], E = digest[4];
-
-   F1(A,B,C,D,E,W[ 0]);   F1(E,A,B,C,D,W[ 1]);   F1(D,E,A,B,C,W[ 2]);
-   F1(C,D,E,A,B,W[ 3]);   F1(B,C,D,E,A,W[ 4]);   F1(A,B,C,D,E,W[ 5]);
-   F1(E,A,B,C,D,W[ 6]);   F1(D,E,A,B,C,W[ 7]);   F1(C,D,E,A,B,W[ 8]);
-   F1(B,C,D,E,A,W[ 9]);   F1(A,B,C,D,E,W[10]);   F1(E,A,B,C,D,W[11]);
-   F1(D,E,A,B,C,W[12]);   F1(C,D,E,A,B,W[13]);   F1(B,C,D,E,A,W[14]);
-   F1(A,B,C,D,E,W[15]);   F1(E,A,B,C,D,W[16]);   F1(D,E,A,B,C,W[17]);
-   F1(C,D,E,A,B,W[18]);   F1(B,C,D,E,A,W[19]);
-
-   F2(A,B,C,D,E,W[20]);   F2(E,A,B,C,D,W[21]);   F2(D,E,A,B,C,W[22]);
-   F2(C,D,E,A,B,W[23]);   F2(B,C,D,E,A,W[24]);   F2(A,B,C,D,E,W[25]);
-   F2(E,A,B,C,D,W[26]);   F2(D,E,A,B,C,W[27]);   F2(C,D,E,A,B,W[28]);
-   F2(B,C,D,E,A,W[29]);   F2(A,B,C,D,E,W[30]);   F2(E,A,B,C,D,W[31]);
-   F2(D,E,A,B,C,W[32]);   F2(C,D,E,A,B,W[33]);   F2(B,C,D,E,A,W[34]);
-   F2(A,B,C,D,E,W[35]);   F2(E,A,B,C,D,W[36]);   F2(D,E,A,B,C,W[37]);
-   F2(C,D,E,A,B,W[38]);   F2(B,C,D,E,A,W[39]);
-
-   F3(A,B,C,D,E,W[40]);   F3(E,A,B,C,D,W[41]);   F3(D,E,A,B,C,W[42]);
-   F3(C,D,E,A,B,W[43]);   F3(B,C,D,E,A,W[44]);   F3(A,B,C,D,E,W[45]);
-   F3(E,A,B,C,D,W[46]);   F3(D,E,A,B,C,W[47]);   F3(C,D,E,A,B,W[48]);
-   F3(B,C,D,E,A,W[49]);   F3(A,B,C,D,E,W[50]);   F3(E,A,B,C,D,W[51]);
-   F3(D,E,A,B,C,W[52]);   F3(C,D,E,A,B,W[53]);   F3(B,C,D,E,A,W[54]);
-   F3(A,B,C,D,E,W[55]);   F3(E,A,B,C,D,W[56]);   F3(D,E,A,B,C,W[57]);
-   F3(C,D,E,A,B,W[58]);   F3(B,C,D,E,A,W[59]);
-
-   F4(A,B,C,D,E,W[60]);   F4(E,A,B,C,D,W[61]);   F4(D,E,A,B,C,W[62]);
-   F4(C,D,E,A,B,W[63]);   F4(B,C,D,E,A,W[64]);   F4(A,B,C,D,E,W[65]);
-   F4(E,A,B,C,D,W[66]);   F4(D,E,A,B,C,W[67]);   F4(C,D,E,A,B,W[68]);
-   F4(B,C,D,E,A,W[69]);   F4(A,B,C,D,E,W[70]);   F4(E,A,B,C,D,W[71]);
-   F4(D,E,A,B,C,W[72]);   F4(C,D,E,A,B,W[73]);   F4(B,C,D,E,A,W[74]);
-   F4(A,B,C,D,E,W[75]);   F4(E,A,B,C,D,W[76]);   F4(D,E,A,B,C,W[77]);
-   F4(C,D,E,A,B,W[78]);   F4(B,C,D,E,A,W[79]);
-
-   digest[0] += A; digest[1] += B; digest[2] += C;
-   digest[3] += D; digest[4] += E;
    }
 
 /*************************************************