From d65046e34693c21067dc580e71485a8e703fc284 Mon Sep 17 00:00:00 2001
From: Paul Duncan <pabs@pablotron.org>
Date: Mon, 18 Sep 2023 19:27:12 -0400
Subject: tests/permute: increase test permutation count, add timing note,
 faster avx512 iota

---
 tests/permute/permute.c | 66 +++++++++++++++++++++++++------------------------
 1 file changed, 34 insertions(+), 32 deletions(-)

diff --git a/tests/permute/permute.c b/tests/permute/permute.c
index 7c8afa6..3a236a0 100644
--- a/tests/permute/permute.c
+++ b/tests/permute/permute.c
@@ -444,12 +444,28 @@ void permute_avx512(uint64_t a[static 25], const size_t num_rounds) {
 // - probably more unnecessary register usage
 void permute_avx512_fast(uint64_t s[static 25], const size_t num_rounds) {
   // unaligned load mask and permutation indices
-  uint8_t mask = 0x1f;
-  const __mmask8 m = _load_mask8(&mask);
+  uint8_t mask = 0x1f,
+          m0b = 0x01;
+  const __mmask8 m = _load_mask8(&mask),
+                 m0 = _load_mask8(&m0b);
+
+  // round constants (used in iota)
+  static const uint64_t RCS[] = {
+    0x0000000000000001ULL, 0x0000000000008082ULL, 0x800000000000808aULL, 0x8000000080008000ULL,
+    0x000000000000808bULL, 0x0000000080000001ULL, 0x8000000080008081ULL, 0x8000000000008009ULL,
+    0x000000000000008aULL, 0x0000000000000088ULL, 0x0000000080008009ULL, 0x000000008000000aULL,
+    0x000000008000808bULL, 0x800000000000008bULL, 0x8000000000008089ULL, 0x8000000000008003ULL,
+    0x8000000000008002ULL, 0x8000000000000080ULL, 0x000000000000800aULL, 0x800000008000000aULL,
+    0x8000000080008081ULL, 0x8000000000008080ULL, 0x0000000080000001ULL, 0x8000000080008008ULL,
+  };
+
+  // load round constant
+  // note: this will bomb if num_rounds < 8 or num_rounds > 24.
+  __m512i rc = _mm512_loadu_epi64((void*) (RCS + 24 - num_rounds));
 
-  // load rc permutation (TODO)
-  // static const uint64_t rc_ps[8] = { 1, 2, 3, 4, 5, 6, 7, 0 },
-  // const __m512i rc_p = _mm512_loadu_epi64(m, (void*) rc_ps);
+  // load rc permutation
+  static const uint64_t rc_ps[8] = { 1, 2, 3, 4, 5, 6, 7, 0 };
+  const __m512i rc_p = _mm512_loadu_epi64((void*) rc_ps);
 
   // load rows
   __m512i r0 = _mm512_maskz_loadu_epi64(m, (void*) (s)),
@@ -639,32 +655,15 @@ void permute_avx512_fast(uint64_t s[static 25], const size_t num_rounds) {
 
     // iota
     {
-      // round constants (ambiguous in spec)
-      static const uint64_t RCS[] = {
-        0x0000000000000001ULL, 0x0000000000008082ULL, 0x800000000000808aULL, 0x8000000080008000ULL,
-        0x000000000000808bULL, 0x0000000080000001ULL, 0x8000000080008081ULL, 0x8000000000008009ULL,
-        0x000000000000008aULL, 0x0000000000000088ULL, 0x0000000080008009ULL, 0x000000008000000aULL,
-        0x000000008000808bULL, 0x800000000000008bULL, 0x8000000000008089ULL, 0x8000000000008003ULL,
-        0x8000000000008002ULL, 0x8000000000000080ULL, 0x000000000000800aULL, 0x800000008000000aULL,
-        0x8000000080008081ULL, 0x8000000000008080ULL, 0x0000000080000001ULL, 0x8000000080008008ULL,
-      };
-
-      // TODO
-      // if ((24 - num_rounds + i + 1) % 8) {
-      //   rc = _mm512_permutexvar_epi64(p0, r4),
-      // } else {
-      // }
-
-      // get rc address
-      const uint64_t *rc = RCS + (24 - num_rounds + i);
-
-      // load mask
-      uint8_t m0b = 0x01;
-      const __mmask8 m0 = _load_mask8(&m0b);
-
-      // mask/store row
-      const __m512i c0 = _mm512_maskz_loadu_epi64(m0, (void*) rc);
-      r0 = _mm512_xor_epi64(r0, c0);
+      // xor round constant, shuffle rc register
+      r0 = _mm512_mask_xor_epi64(r0, m0, r0, rc);
+      rc = _mm512_permutexvar_epi64(rc_p, rc);
+
+      if (((24 - num_rounds + i + 1) % 8) == 0 && i != 23) {
+        // load next set of round constants
+        // note: this will bomb if num_rounds < 8 or num_rounds > 24.
+        rc = _mm512_loadu_epi64((void*) (RCS + 24 - num_rounds + (i + 1)));
+      }
     }
   }
 
@@ -816,7 +815,7 @@ static void test_permute_fast(void) {
 }
 
 // number of times to run permutation in timing tests below
-#define NUM_TIME_PERMUTES 10000000
+#define NUM_TIME_PERMUTES 20000000
 
 // time scalar keccak permutation
 static void time_permute_scalar(void) {
@@ -882,8 +881,11 @@ int main() {
   test_iota();
   test_permute_slow();
   test_permute_fast();
+
+  printf("timing permute, please wait...\n");
   time_permute_scalar();
   time_permute_avx512_slow();
   time_permute_avx512_fast();
+
   return 0;
 }
-- 
cgit v1.2.3