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

2 files changed, 71 insertions, 67 deletions

diff --git a/src/hash/fork256/fork256.cpp b/src/hash/fork256/fork256.cpp
index 5ac779cab..dc023004d 100644
--- a/src/hash/fork256/fork256.cpp
+++ b/src/hash/fork256/fork256.cpp
@@ -40,7 +40,7 @@ inline void step(u32bit& A, u32bit& B, u32bit& C, u32bit& D,
 /*************************************************
 * FORK-256 Compression Function                   *
 *************************************************/
-void FORK_256::hash(const byte input[])
+void FORK_256::compress_n(const byte input[], u32bit blocks)
    {
    const u32bit DELTA[16] = {
       0x428A2F98, 0x71374491, 0xB5C0FBCF, 0xE9B5DBA5, 0x3956C25B, 0x59F111F1,
@@ -48,71 +48,75 @@ void FORK_256::hash(const byte input[])
       0x72BE5D74, 0x80DEB1FE, 0x9BDC06A7, 0xC19BF174
    };
 
-   u32bit A1, B1, C1, D1, E1, F1, G1, H1;
-   u32bit A2, B2, C2, D2, E2, F2, G2, H2;
-   u32bit A3, B3, C3, D3, E3, F3, G3, H3;
-   u32bit A4, B4, C4, D4, E4, F4, G4, H4;
-
-   A1 = A2 = A3 = A4 = digest[0];
-   B1 = B2 = B3 = B4 = digest[1];
-   C1 = C2 = C3 = C4 = digest[2];
-   D1 = D2 = D3 = D4 = digest[3];
-   E1 = E2 = E3 = E4 = digest[4];
-   F1 = F2 = F3 = F4 = digest[5];
-   G1 = G2 = G3 = G4 = digest[6];
-   H1 = H2 = H3 = H4 = digest[7];
-
-   for(u32bit j = 0; j != 16; ++j)
-      M[j] = load_be<u32bit>(input, j);
-
-   step(A1, B1, C1, D1, E1, F1, G1, H1, M[ 0], M[ 1], DELTA[ 0], DELTA[ 1]);
-   step(A2, B2, C2, D2, E2, F2, G2, H2, M[14], M[15], DELTA[15], DELTA[14]);
-   step(A3, B3, C3, D3, E3, F3, G3, H3, M[ 7], M[ 6], DELTA[ 1], DELTA[ 0]);
-   step(A4, B4, C4, D4, E4, F4, G4, H4, M[ 5], M[12], DELTA[14], DELTA[15]);
-
-   step(H1, A1, B1, C1, D1, E1, F1, G1, M[ 2], M[ 3], DELTA[ 2], DELTA[ 3]);
-   step(H2, A2, B2, C2, D2, E2, F2, G2, M[11], M[ 9], DELTA[13], DELTA[12]);
-   step(H3, A3, B3, C3, D3, E3, F3, G3, M[10], M[14], DELTA[ 3], DELTA[ 2]);
-   step(H4, A4, B4, C4, D4, E4, F4, G4, M[ 1], M[ 8], DELTA[12], DELTA[13]);
-
-   step(G1, H1, A1, B1, C1, D1, E1, F1, M[ 4], M[ 5], DELTA[ 4], DELTA[ 5]);
-   step(G2, H2, A2, B2, C2, D2, E2, F2, M[ 8], M[10], DELTA[11], DELTA[10]);
-   step(G3, H3, A3, B3, C3, D3, E3, F3, M[13], M[ 2], DELTA[ 5], DELTA[ 4]);
-   step(G4, H4, A4, B4, C4, D4, E4, F4, M[15], M[ 0], DELTA[10], DELTA[11]);
-
-   step(F1, G1, H1, A1, B1, C1, D1, E1, M[ 6], M[ 7], DELTA[ 6], DELTA[ 7]);
-   step(F2, G2, H2, A2, B2, C2, D2, E2, M[ 3], M[ 4], DELTA[ 9], DELTA[ 8]);
-   step(F3, G3, H3, A3, B3, C3, D3, E3, M[ 9], M[12], DELTA[ 7], DELTA[ 6]);
-   step(F4, G4, H4, A4, B4, C4, D4, E4, M[13], M[11], DELTA[ 8], DELTA[ 9]);
-
-   step(E1, F1, G1, H1, A1, B1, C1, D1, M[ 8], M[ 9], DELTA[ 8], DELTA[ 9]);
-   step(E2, F2, G2, H2, A2, B2, C2, D2, M[ 2], M[13], DELTA[ 7], DELTA[ 6]);
-   step(E3, F3, G3, H3, A3, B3, C3, D3, M[11], M[ 4], DELTA[ 9], DELTA[ 8]);
-   step(E4, F4, G4, H4, A4, B4, C4, D4, M[ 3], M[10], DELTA[ 6], DELTA[ 7]);
-
-   step(D1, E1, F1, G1, H1, A1, B1, C1, M[10], M[11], DELTA[10], DELTA[11]);
-   step(D2, E2, F2, G2, H2, A2, B2, C2, M[ 0], M[ 5], DELTA[ 5], DELTA[ 4]);
-   step(D3, E3, F3, G3, H3, A3, B3, C3, M[15], M[ 8], DELTA[11], DELTA[10]);
-   step(D4, E4, F4, G4, H4, A4, B4, C4, M[ 9], M[ 2], DELTA[ 4], DELTA[ 5]);
-
-   step(C1, D1, E1, F1, G1, H1, A1, B1, M[12], M[13], DELTA[12], DELTA[13]);
-   step(C2, D2, E2, F2, G2, H2, A2, B2, M[ 6], M[ 7], DELTA[ 3], DELTA[ 2]);
-   step(C3, D3, E3, F3, G3, H3, A3, B3, M[ 5], M[ 0], DELTA[13], DELTA[12]);
-   step(C4, D4, E4, F4, G4, H4, A4, B4, M[ 7], M[14], DELTA[ 2], DELTA[ 3]);
-
-   step(B1, C1, D1, E1, F1, G1, H1, A1, M[14], M[15], DELTA[14], DELTA[15]);
-   step(B2, C2, D2, E2, F2, G2, H2, A2, M[12], M[ 1], DELTA[ 1], DELTA[ 0]);
-   step(B3, C3, D3, E3, F3, G3, H3, A3, M[ 1], M[ 3], DELTA[15], DELTA[14]);
-   step(B4, C4, D4, E4, F4, G4, H4, A4, M[ 4], M[ 6], DELTA[ 0], DELTA[ 1]);
-
-   digest[0] += (A1 + A2) ^ (A3 + A4);
-   digest[1] += (B1 + B2) ^ (B3 + B4);
-   digest[2] += (C1 + C2) ^ (C3 + C4);
-   digest[3] += (D1 + D2) ^ (D3 + D4);
-   digest[4] += (E1 + E2) ^ (E3 + E4);
-   digest[5] += (F1 + F2) ^ (F3 + F4);
-   digest[6] += (G1 + G2) ^ (G3 + G4);
-   digest[7] += (H1 + H2) ^ (H3 + H4);
+   for(u32bit i = 0; i != blocks; ++i)
+      {
+      u32bit A1, B1, C1, D1, E1, F1, G1, H1;
+      u32bit A2, B2, C2, D2, E2, F2, G2, H2;
+      u32bit A3, B3, C3, D3, E3, F3, G3, H3;
+      u32bit A4, B4, C4, D4, E4, F4, G4, H4;
+
+      A1 = A2 = A3 = A4 = digest[0];
+      B1 = B2 = B3 = B4 = digest[1];
+      C1 = C2 = C3 = C4 = digest[2];
+      D1 = D2 = D3 = D4 = digest[3];
+      E1 = E2 = E3 = E4 = digest[4];
+      F1 = F2 = F3 = F4 = digest[5];
+      G1 = G2 = G3 = G4 = digest[6];
+      H1 = H2 = H3 = H4 = digest[7];
+
+      for(u32bit j = 0; j != 16; ++j)
+         M[j] = load_be<u32bit>(input, j);
+      input += HASH_BLOCK_SIZE;
+
+      step(A1, B1, C1, D1, E1, F1, G1, H1, M[ 0], M[ 1], DELTA[ 0], DELTA[ 1]);
+      step(A2, B2, C2, D2, E2, F2, G2, H2, M[14], M[15], DELTA[15], DELTA[14]);
+      step(A3, B3, C3, D3, E3, F3, G3, H3, M[ 7], M[ 6], DELTA[ 1], DELTA[ 0]);
+      step(A4, B4, C4, D4, E4, F4, G4, H4, M[ 5], M[12], DELTA[14], DELTA[15]);
+
+      step(H1, A1, B1, C1, D1, E1, F1, G1, M[ 2], M[ 3], DELTA[ 2], DELTA[ 3]);
+      step(H2, A2, B2, C2, D2, E2, F2, G2, M[11], M[ 9], DELTA[13], DELTA[12]);
+      step(H3, A3, B3, C3, D3, E3, F3, G3, M[10], M[14], DELTA[ 3], DELTA[ 2]);
+      step(H4, A4, B4, C4, D4, E4, F4, G4, M[ 1], M[ 8], DELTA[12], DELTA[13]);
+
+      step(G1, H1, A1, B1, C1, D1, E1, F1, M[ 4], M[ 5], DELTA[ 4], DELTA[ 5]);
+      step(G2, H2, A2, B2, C2, D2, E2, F2, M[ 8], M[10], DELTA[11], DELTA[10]);
+      step(G3, H3, A3, B3, C3, D3, E3, F3, M[13], M[ 2], DELTA[ 5], DELTA[ 4]);
+      step(G4, H4, A4, B4, C4, D4, E4, F4, M[15], M[ 0], DELTA[10], DELTA[11]);
+
+      step(F1, G1, H1, A1, B1, C1, D1, E1, M[ 6], M[ 7], DELTA[ 6], DELTA[ 7]);
+      step(F2, G2, H2, A2, B2, C2, D2, E2, M[ 3], M[ 4], DELTA[ 9], DELTA[ 8]);
+      step(F3, G3, H3, A3, B3, C3, D3, E3, M[ 9], M[12], DELTA[ 7], DELTA[ 6]);
+      step(F4, G4, H4, A4, B4, C4, D4, E4, M[13], M[11], DELTA[ 8], DELTA[ 9]);
+
+      step(E1, F1, G1, H1, A1, B1, C1, D1, M[ 8], M[ 9], DELTA[ 8], DELTA[ 9]);
+      step(E2, F2, G2, H2, A2, B2, C2, D2, M[ 2], M[13], DELTA[ 7], DELTA[ 6]);
+      step(E3, F3, G3, H3, A3, B3, C3, D3, M[11], M[ 4], DELTA[ 9], DELTA[ 8]);
+      step(E4, F4, G4, H4, A4, B4, C4, D4, M[ 3], M[10], DELTA[ 6], DELTA[ 7]);
+
+      step(D1, E1, F1, G1, H1, A1, B1, C1, M[10], M[11], DELTA[10], DELTA[11]);
+      step(D2, E2, F2, G2, H2, A2, B2, C2, M[ 0], M[ 5], DELTA[ 5], DELTA[ 4]);
+      step(D3, E3, F3, G3, H3, A3, B3, C3, M[15], M[ 8], DELTA[11], DELTA[10]);
+      step(D4, E4, F4, G4, H4, A4, B4, C4, M[ 9], M[ 2], DELTA[ 4], DELTA[ 5]);
+
+      step(C1, D1, E1, F1, G1, H1, A1, B1, M[12], M[13], DELTA[12], DELTA[13]);
+      step(C2, D2, E2, F2, G2, H2, A2, B2, M[ 6], M[ 7], DELTA[ 3], DELTA[ 2]);
+      step(C3, D3, E3, F3, G3, H3, A3, B3, M[ 5], M[ 0], DELTA[13], DELTA[12]);
+      step(C4, D4, E4, F4, G4, H4, A4, B4, M[ 7], M[14], DELTA[ 2], DELTA[ 3]);
+
+      step(B1, C1, D1, E1, F1, G1, H1, A1, M[14], M[15], DELTA[14], DELTA[15]);
+      step(B2, C2, D2, E2, F2, G2, H2, A2, M[12], M[ 1], DELTA[ 1], DELTA[ 0]);
+      step(B3, C3, D3, E3, F3, G3, H3, A3, M[ 1], M[ 3], DELTA[15], DELTA[14]);
+      step(B4, C4, D4, E4, F4, G4, H4, A4, M[ 4], M[ 6], DELTA[ 0], DELTA[ 1]);
+
+      digest[0] += (A1 + A2) ^ (A3 + A4);
+      digest[1] += (B1 + B2) ^ (B3 + B4);
+      digest[2] += (C1 + C2) ^ (C3 + C4);
+      digest[3] += (D1 + D2) ^ (D3 + D4);
+      digest[4] += (E1 + E2) ^ (E3 + E4);
+      digest[5] += (F1 + F2) ^ (F3 + F4);
+      digest[6] += (G1 + G2) ^ (G3 + G4);
+      digest[7] += (H1 + H2) ^ (H3 + H4);
+      }
    }
 
 /*************************************************
diff --git a/src/hash/fork256/fork256.h b/src/hash/fork256/fork256.h
index 8ad9c1f8d..d27e8693f 100644
--- a/src/hash/fork256/fork256.h
+++ b/src/hash/fork256/fork256.h
@@ -21,7 +21,7 @@ class BOTAN_DLL FORK_256 : public MDx_HashFunction
       HashFunction* clone() const { return new FORK_256; }
       FORK_256() : MDx_HashFunction(32, 64, true, true) { clear(); }
    private:
-      void hash(const byte[]);
+      void compress_n(const byte[], u32bit blocks);
       void copy_out(byte[]);
 
       SecureBuffer<u32bit, 8> digest;