avx512 implementations for fixupMaxVersion

ConflictSet.cpp

@@ -1604,7 +1604,6 @@ __attribute__((target("avx512f"))) void rezero16(InternalVersionT *vs,
       _mm512_sub_epi32(_mm512_loadu_epi32(vs), zvec), _mm512_setzero_epi32());
   _mm512_mask_storeu_epi32(vs, m, zvec);
 }
-
 __attribute__((target("default")))
 #endif
 
@@ -2471,6 +2470,7 @@ checkMaxBetweenExclusive(Node *n, int begin, int end,
 }
 __attribute__((target("default")))
 #endif
+
 bool checkMaxBetweenExclusive(Node *n, int begin, int end,
                          InternalVersionT readVersion, ReadContext *tls) {
   return checkMaxBetweenExclusiveImpl<false>(n, begin, end, readVersion, tls);
@@ -2910,6 +2910,71 @@ void addPointWrite(Node *&root, std::span<const uint8_t> key,
   }
 }
 
+#if defined(HAS_AVX) && !defined(__SANITIZE_THREAD__)
+__attribute__((target("avx512f"))) InternalVersionT
+horizontalMaxUpTo16(InternalVersionT *vs, InternalVersionT z, int len) {
+  assume(len <= 16);
+#if USE_64_BIT
+  // Hope it gets vectorized
+  InternalVersionT max = vs[0];
+  for (int i = 1; i < len; ++i) {
+    max = std::max(vs[i], max);
+  }
+  return max;
+#else
+  uint32_t zero;
+  memcpy(&zero, &z, sizeof(zero));
+  auto zeroVec = _mm512_set1_epi32(zero);
+  return InternalVersionT(
+      zero +
+      _mm512_reduce_max_epu32(_mm512_sub_epi32(
+          _mm512_mask_loadu_epi32(zeroVec, _mm512_int2mask((1 << len) - 1), vs),
+          zeroVec)));
+#endif
+}
+__attribute__((target("default")))
+#endif
+
+InternalVersionT
+horizontalMaxUpTo16(InternalVersionT *vs, InternalVersionT, int len) {
+  assume(len <= 16);
+  InternalVersionT max = vs[0];
+  for (int i = 1; i < len; ++i) {
+    max = std::max(vs[i], max);
+  }
+  return max;
+}
+
+#if defined(HAS_AVX) && !defined(__SANITIZE_THREAD__)
+__attribute__((target("avx512f"))) InternalVersionT
+horizontalMax16(InternalVersionT *vs, InternalVersionT z) {
+#if USE_64_BIT
+  // Hope it gets vectorized
+  InternalVersionT max = vs[0];
+  for (int i = 1; i < 16; ++i) {
+    max = std::max(vs[i], max);
+  }
+  return max;
+#else
+  uint32_t zero;
+  memcpy(&zero, &z, sizeof(zero));
+  auto zeroVec = _mm512_set1_epi32(zero);
+  return InternalVersionT(zero + _mm512_reduce_max_epu32(_mm512_sub_epi32(
+                                     _mm512_loadu_epi32(vs), zeroVec)));
+#endif
+}
+__attribute__((target("default")))
+#endif
+
+InternalVersionT
+horizontalMax16(InternalVersionT *vs, InternalVersionT) {
+  InternalVersionT max = vs[0];
+  for (int i = 1; i < 16; ++i) {
+    max = std::max(vs[i], max);
+  }
+  return max;
+}
+
 // Precondition: `node->entryPresent`, and node is not the root
 void fixupMaxVersion(Node *node, WriteContext *tls) {
   assert(node->parent);
@@ -2921,15 +2986,13 @@ void fixupMaxVersion(Node *node, WriteContext *tls) {
     break;
   case Type_Node3: {
     auto *self3 = static_cast<Node3 *>(node);
-    for (int i = 0; i < self3->numChildren; ++i) {
+    max = std::max(max, horizontalMaxUpTo16(self3->childMaxVersion, tls->zero,
-      max = std::max(self3->childMaxVersion[i], max);
+                                            self3->numChildren));
-    }
   } break;
   case Type_Node16: {
     auto *self16 = static_cast<Node16 *>(node);
-    for (int i = 0; i < self16->numChildren; ++i) {
+    max = std::max(max, horizontalMaxUpTo16(self16->childMaxVersion, tls->zero,
-      max = std::max(self16->childMaxVersion[i], max);
+                                            self16->numChildren));
-    }
   } break;
   case Type_Node48: {
     auto *self48 = static_cast<Node48 *>(node);
@@ -2939,9 +3002,7 @@ void fixupMaxVersion(Node *node, WriteContext *tls) {
   } break;
   case Type_Node256: {
     auto *self256 = static_cast<Node256 *>(node);
-    for (auto v : self256->maxOfMax) {
+    max = std::max(max, horizontalMax16(self256->childMaxVersion, tls->zero));
-      max = std::max(v, max);
-    }
   } break;
   default:                   // GCOVR_EXCL_LINE
     __builtin_unreachable(); // GCOVR_EXCL_LINE
@@ -4033,6 +4094,24 @@ template <int kN> void benchScan2() {
   });
 }
 
+void benchHorizontal16() {
+  ankerl::nanobench::Bench bench;
+  InternalVersionT vs[16];
+  for (int i = 0; i < 16; ++i) {
+    vs[i] = InternalVersionT(rand() % 1000 + 1000);
+  }
+#if !USE_64_BIT
+  InternalVersionT::zero = InternalVersionT(rand() % 1000);
+#endif
+  bench.run("horizontal16", [&]() {
+    bench.doNotOptimizeAway(horizontalMax16(vs, InternalVersionT::zero));
+  });
+  int x = rand() % 15 + 1;
+  bench.run("horizontalUpTo16", [&]() {
+    bench.doNotOptimizeAway(horizontalMaxUpTo16(vs, InternalVersionT::zero, x));
+  });
+}
+
 void benchLCP(int len) {
   ankerl::nanobench::Bench bench;
   std::vector<uint8_t> lhs(len);
@@ -4065,11 +4144,7 @@ void printTree() {
   debugPrintDot(stdout, cs.root, &cs);
 }
 
-int main(void) {
+int main(void) { benchHorizontal16(); }
-  for (int i = 0; i < 256; ++i) {
-    benchLCP(i);
-  }
-}
 #endif
 
 #ifdef ENABLE_FUZZ