Add longestCommonPrefix bench

low bits of lcg have low periods

CMakeLists.txt

@@ -1,7 +1,7 @@
 cmake_minimum_required(VERSION 3.18)
 project(
   conflict-set
-  VERSION 0.0.12
+  VERSION 0.0.14
   DESCRIPTION
     "A data structure for optimistic concurrency control on ranges of bitwise-lexicographically-ordered keys."
   HOMEPAGE_URL "https://git.weaselab.dev/weaselab/conflict-set"

ConflictSet.cpp

@@ -17,9 +17,9 @@ limitations under the License.
 #include "ConflictSet.h"
 #include "Internal.h"
 #include "LongestCommonPrefix.h"
+#include "Metrics.h"
 
 #include <algorithm>
-#include <atomic>
 #include <bit>
 #include <cassert>
 #include <cstddef>
@@ -87,22 +87,42 @@ constexpr int64_t kMaxCorrectVersionWindow =
     std::numeric_limits<int32_t>::max();
 static_assert(kNominalVersionWindow <= kMaxCorrectVersionWindow);
 
+#ifndef USE_64_BIT
+#define USE_64_BIT 0
+#endif
+
 struct InternalVersionT {
   constexpr InternalVersionT() = default;
   constexpr explicit InternalVersionT(int64_t value) : value(value) {}
   constexpr int64_t toInt64() const { return value; } // GCOVR_EXCL_LINE
   constexpr auto operator<=>(const InternalVersionT &rhs) const {
+#if USE_64_BIT
+    return value <=> rhs.value;
+#else
     // Maintains ordering after overflow, as long as the full-precision versions
     // are within `kMaxCorrectVersionWindow` of eachother.
     return int32_t(value - rhs.value) <=> 0;
+#endif
   }
   constexpr bool operator==(const InternalVersionT &) const = default;
+#if USE_64_BIT
+  static const InternalVersionT zero;
+#else
   static thread_local InternalVersionT zero;
+#endif
 
 private:
+#if USE_64_BIT
+  int64_t value;
+#else
   uint32_t value;
+#endif
 };
+#if USE_64_BIT
+const InternalVersionT InternalVersionT::zero{0};
+#else
 thread_local InternalVersionT InternalVersionT::zero;
+#endif
 
 struct Entry {
   InternalVersionT pointVersion;
@@ -518,8 +538,13 @@ std::string getSearchPath(Node *n);
 // Each node with an entry present gets a budget of kBytesPerKey. Node0 always
 // has an entry present.
 // Induction hypothesis is that each node's surplus is >= kMinNodeSurplus
+#if USE_64_BIT
+constexpr int kBytesPerKey = 144;
+constexpr int kMinNodeSurplus = 104;
+#else
 constexpr int kBytesPerKey = 112;
 constexpr int kMinNodeSurplus = 80;
+#endif
 // Cound the entry itself as a child
 constexpr int kMinChildrenNode0 = 1;
 constexpr int kMinChildrenNode3 = 2;
@@ -553,39 +578,6 @@ static_assert(kBytesPerKey - sizeof(Node0) >= kMinNodeSurplus);
 
 constexpr int64_t kFreeListMaxMemory = 1 << 20;
 
-struct Metric {
-  Metric *prev;
-  const char *name;
-  const char *help;
-  ConflictSet::MetricsV1::Type type;
-  std::atomic<int64_t> value;
-
-protected:
-  Metric(ConflictSet::Impl *impl, const char *name, const char *help,
-         ConflictSet::MetricsV1::Type type);
-};
-
-struct Gauge : private Metric {
-  Gauge(ConflictSet::Impl *impl, const char *name, const char *help)
-      : Metric(impl, name, help, ConflictSet::MetricsV1::Gauge) {}
-
-  void set(int64_t value) {
-    this->value.store(value, std::memory_order_relaxed);
-  }
-};
-
-struct Counter : private Metric {
-  Counter(ConflictSet::Impl *impl, const char *name, const char *help)
-      : Metric(impl, name, help, ConflictSet::MetricsV1::Counter) {}
-  // Expensive. Accumulate locally and then call add instead of repeatedly
-  // calling add.
-  void add(int64_t value) {
-    assert(value >= 0);
-    static_assert(std::atomic<int64_t>::is_always_lock_free);
-    this->value.fetch_add(value, std::memory_order_relaxed);
-  }
-};
-
 template <class T> struct BoundedFreeListAllocator {
 
   static_assert(sizeof(T) >= sizeof(void *));
@@ -724,9 +716,13 @@ struct WriteContext {
     int64_t write_bytes;
   } accum;
 
+#if USE_64_BIT
+  static constexpr InternalVersionT zero{0};
+#else
   // Cache a copy of InternalVersionT::zero, so we don't need to do the TLS
   // lookup as often.
   InternalVersionT zero;
+#endif
 
   WriteContext() { memset(&accum, 0, sizeof(accum)); }
 
@@ -1563,6 +1559,9 @@ void rezero16(InternalVersionT *vs, InternalVersionT zero) {
   }
 }
 
+#if USE_64_BIT
+void rezero(Node *, InternalVersionT) {}
+#else
 void rezero(Node *n, InternalVersionT z) {
 #if DEBUG_VERBOSE && !defined(NDEBUG)
   fprintf(stderr, "rezero to %" PRId64 ": %s\n", z.toInt64(),
@@ -1605,6 +1604,7 @@ void rezero(Node *n, InternalVersionT z) {
     __builtin_unreachable(); // GCOVR_EXCL_LINE
   }
 }
+#endif
 
 void mergeWithChild(Node *&self, WriteContext *tls, ConflictSet::Impl *impl,
                     Node *&dontInvalidate, Node3 *self3) {
@@ -1987,7 +1987,14 @@ downLeftSpine:
 }
 
 #ifdef HAS_AVX
-uint32_t compare16_32bit(const InternalVersionT *vs, InternalVersionT rv) {
+uint32_t compare16(const InternalVersionT *vs, InternalVersionT rv) {
+#if USE_64_BIT
+  uint32_t compared = 0;
+  for (int i = 0; i < 16; ++i) {
+    compared |= (vs[i] > rv) << i;
+  }
+  return compared;
+#else
   uint32_t compared = 0;
   __m128i w[4]; // GCOVR_EXCL_LINE
   memcpy(w, vs, sizeof(w));
@@ -2001,15 +2008,26 @@ uint32_t compare16_32bit(const InternalVersionT *vs, InternalVersionT rv) {
                 << (i * 4);
   }
   return compared;
+#endif
 }
 
 __attribute__((target("avx512f"))) uint32_t
-compare16_32bit_avx512(const InternalVersionT *vs, InternalVersionT rv) {
+compare16_avx512(const InternalVersionT *vs, InternalVersionT rv) {
+#if USE_64_BIT
+  int64_t r;
+  memcpy(&r, &rv, sizeof(r));
+  uint32_t low =
+      _mm512_cmpgt_epi64_mask(_mm512_loadu_epi64(vs), _mm512_set1_epi64(r));
+  uint32_t high =
+      _mm512_cmpgt_epi64_mask(_mm512_loadu_epi64(vs + 8), _mm512_set1_epi64(r));
+  return low | (high << 8);
+#else
   uint32_t r;
   memcpy(&r, &rv, sizeof(r));
   return _mm512_cmpgt_epi32_mask(
       _mm512_sub_epi32(_mm512_loadu_epi32(vs), _mm512_set1_epi32(r)),
       _mm512_setzero_epi32());
+#endif
 }
 #endif
 
@@ -2066,9 +2084,9 @@ bool scan16(const InternalVersionT *vs, const uint8_t *is, int begin, int end,
 
   uint32_t compared = 0;
   if constexpr (kAVX512) {
-    compared = compare16_32bit_avx512(vs, readVersion); // GCOVR_EXCL_LINE
+    compared = compare16_avx512(vs, readVersion);
   } else {
-    compared = compare16_32bit(vs, readVersion); // GCOVR_EXCL_LINE
+    compared = compare16(vs, readVersion);
   }
   return !(compared & mask);
 
@@ -2127,9 +2145,9 @@ bool scan16(const InternalVersionT *vs, int begin, int end,
 #elif defined(HAS_AVX)
   uint32_t conflict;
   if constexpr (kAVX512) {
-    conflict = compare16_32bit_avx512(vs, readVersion); // GCOVR_EXCL_LINE
+    conflict = compare16_avx512(vs, readVersion);
   } else {
-    conflict = compare16_32bit(vs, readVersion); // GCOVR_EXCL_LINE
+    conflict = compare16(vs, readVersion);
   }
   conflict &= (1 << end) - 1;
   conflict >>= begin;
@@ -2264,12 +2282,9 @@ bool checkMaxBetweenExclusiveImpl(Node *n, int begin, int end,
 
     uint32_t compared = 0;
     if constexpr (kAVX512) {
-      compared =                                        // GCOVR_EXCL_LINE
-          compare16_32bit_avx512(self->childMaxVersion, // GCOVR_EXCL_LINE
-                                 readVersion);          // GCOVR_EXCL_LINE
+      compared = compare16_avx512(self->childMaxVersion, readVersion);
     } else {
-      compared = compare16_32bit(self->childMaxVersion,
-                                 readVersion); // GCOVR_EXCL_LINE
+      compared = compare16(self->childMaxVersion, readVersion);
     }
     return !(compared & mask) && firstRangeOk;
 
@@ -2495,38 +2510,19 @@ downLeftSpine:
 namespace {
 // Return true if the max version among all keys that start with key[:prefixLen]
 // that are >= key is <= readVersion
-struct CheckRangeLeftSide {
-  CheckRangeLeftSide(Node *n, std::span<const uint8_t> key, int prefixLen,
-                     InternalVersionT readVersion, ReadContext *tls)
-      : n(n), remaining(key), prefixLen(prefixLen), readVersion(readVersion),
-        impl(tls->impl), tls(tls) {
-#if DEBUG_VERBOSE && !defined(NDEBUG)
-    fprintf(stderr, "Check range left side from %s for keys starting with %s\n",
-            printable(key).c_str(),
-            printable(key.subspan(0, prefixLen)).c_str());
-#endif
-  }
-
-  Node *n;
-  std::span<const uint8_t> remaining;
-  int prefixLen;
-  InternalVersionT readVersion;
-  ConflictSet::Impl *impl;
-  ReadContext *tls;
+bool checkRangeLeftSide(Node *n, std::span<const uint8_t> key, int prefixLen,
+                        InternalVersionT readVersion, ReadContext *tls) {
+  auto remaining = key;
   int searchPathLen = 0;
-  bool ok;
-
-  bool step() {
+  for (;; ++tls->range_read_iterations_accum) {
     if (remaining.size() == 0) {
       assert(searchPathLen >= prefixLen);
-      ok = maxVersion(n) <= readVersion;
-      return true;
+      return maxVersion(n) <= readVersion;
     }
 
     if (searchPathLen >= prefixLen) {
       if (!checkMaxBetweenExclusive(n, remaining[0], 256, readVersion, tls)) {
-        ok = false;
-        return true;
+        return false;
       }
     }
 
@@ -2536,18 +2532,16 @@ struct CheckRangeLeftSide {
       if (c != nullptr) {
         if (searchPathLen < prefixLen) {
           n = c;
-          return downLeftSpine();
+          goto downLeftSpine;
         }
         n = c;
-        ok = maxVersion(n) <= readVersion;
-        return true;
+        return maxVersion(n) <= readVersion;
       } else {
         n = nextSibling(n);
         if (n == nullptr) {
-          ok = true;
           return true;
         }
-        return downLeftSpine();
+        goto downLeftSpine;
       }
     }
 
@@ -2563,21 +2557,18 @@ struct CheckRangeLeftSide {
         auto c = n->partialKey()[i] <=> remaining[i];
         if (c > 0) {
           if (searchPathLen < prefixLen) {
-            return downLeftSpine();
+            goto downLeftSpine;
           }
           if (n->entryPresent && n->entry.rangeVersion > readVersion) {
-            ok = false;
-            return true;
+            return false;
           }
-          ok = maxVersion(n) <= readVersion;
-          return true;
+          return maxVersion(n) <= readVersion;
         } else {
           n = nextSibling(n);
           if (n == nullptr) {
-            ok = true;
             return true;
           }
-          return downLeftSpine();
+          goto downLeftSpine;
         }
       }
       if (commonLen == n->partialKeyLen) {
@@ -2586,83 +2577,47 @@ struct CheckRangeLeftSide {
       } else if (n->partialKeyLen > int(remaining.size())) {
         assert(searchPathLen >= prefixLen);
         if (n->entryPresent && n->entry.rangeVersion > readVersion) {
-          ok = false;
-          return true;
+          return false;
         }
-        ok = maxVersion(n) <= readVersion;
-        return true;
+        return maxVersion(n) <= readVersion;
       }
     }
     if (maxV <= readVersion) {
-      ok = true;
       return true;
     }
-    return false;
   }
-
-  bool downLeftSpine() {
-    for (; !n->entryPresent; n = getFirstChildExists(n)) {
-    }
-    ok = n->entry.rangeVersion <= readVersion;
-    return true;
+downLeftSpine:
+  for (; !n->entryPresent; n = getFirstChildExists(n)) {
   }
-};
+  return n->entry.rangeVersion <= readVersion;
+}
 
 // Return true if the max version among all keys that start with key[:prefixLen]
 // that are < key is <= readVersion
-struct CheckRangeRightSide {
-  CheckRangeRightSide(Node *n, std::span<const uint8_t> key, int prefixLen,
-                      InternalVersionT readVersion, ReadContext *tls)
-      : n(n), key(key), remaining(key), prefixLen(prefixLen),
-        readVersion(readVersion), impl(tls->impl), tls(tls) {
-#if DEBUG_VERBOSE && !defined(NDEBUG)
-    fprintf(stderr, "Check range right side to %s for keys starting with %s\n",
-            printable(key).c_str(),
-            printable(key.subspan(0, prefixLen)).c_str());
-#endif
-  }
-
-  Node *n;
-  std::span<const uint8_t> key;
-  std::span<const uint8_t> remaining;
-  int prefixLen;
-  InternalVersionT readVersion;
-  ConflictSet::Impl *impl;
-  ReadContext *tls;
+bool checkRangeRightSide(Node *n, std::span<const uint8_t> key, int prefixLen,
+                         InternalVersionT readVersion, ReadContext *tls) {
+  auto remaining = key;
   int searchPathLen = 0;
-  bool ok;
-
-  bool step() {
-#if DEBUG_VERBOSE && !defined(NDEBUG)
-    fprintf(stderr,
-            "Search path: %s, searchPathLen: %d, prefixLen: %d, remaining: "
-            "%s\n",
-            getSearchPathPrintable(n).c_str(), searchPathLen, prefixLen,
-            printable(remaining).c_str());
-#endif
 
+  for (;; ++tls->range_read_iterations_accum) {
     assert(searchPathLen <= int(key.size()));
-
     if (remaining.size() == 0) {
-      return downLeftSpine();
+      goto downLeftSpine;
     }
 
     if (searchPathLen >= prefixLen) {
       if (n->entryPresent && n->entry.pointVersion > readVersion) {
-        ok = false;
-        return true;
+        return false;
       }
 
       if (!checkMaxBetweenExclusive(n, -1, remaining[0], readVersion, tls)) {
-        ok = false;
-        return true;
+        return false;
       }
     }
 
     if (searchPathLen > prefixLen && n->entryPresent &&
         n->entry.rangeVersion > readVersion) {
-      ok = false;
-      return true;
+      return false;
     }
 
     auto *child = getChild(n, remaining[0]);
@@ -2670,9 +2625,9 @@ struct CheckRangeRightSide {
       auto c = getChildGeq(n, remaining[0]);
       if (c != nullptr) {
         n = c;
-        return downLeftSpine();
+        goto downLeftSpine;
       } else {
-        return backtrack();
+        goto backtrack;
       }
     }
 
@@ -2688,57 +2643,48 @@ struct CheckRangeRightSide {
         ++searchPathLen;
         auto c = n->partialKey()[i] <=> remaining[i];
         if (c > 0) {
-          return downLeftSpine();
+          goto downLeftSpine;
         } else {
           if (searchPathLen > prefixLen && n->entryPresent &&
               n->entry.rangeVersion > readVersion) {
-            ok = false;
-            return true;
+            return false;
           }
-          return backtrack();
+          goto backtrack;
         }
       }
       if (commonLen == n->partialKeyLen) {
         // partial key matches
         remaining = remaining.subspan(commonLen, remaining.size() - commonLen);
       } else if (n->partialKeyLen > int(remaining.size())) {
-        return downLeftSpine();
-      }
-    }
-    return false;
-  }
-
-  bool backtrack() {
-    for (;;) {
-      // searchPathLen > prefixLen implies n is not the root
-      if (searchPathLen > prefixLen && maxVersion(n) > readVersion) {
-        ok = false;
-        return true;
-      }
-      if (n->parent == nullptr) {
-        ok = true;
-        return true;
-      }
-      auto next = getChildGeq(n->parent, n->parentsIndex + 1);
-      if (next == nullptr) {
-        searchPathLen -= 1 + n->partialKeyLen;
-        n = n->parent;
-      } else {
-        searchPathLen -= n->partialKeyLen;
-        n = next;
-        searchPathLen += n->partialKeyLen;
-        return downLeftSpine();
+        goto downLeftSpine;
       }
     }
   }
-
-  bool downLeftSpine() {
-    for (; !n->entryPresent; n = getFirstChildExists(n)) {
+backtrack:
+  for (;;) {
+    // searchPathLen > prefixLen implies n is not the root
+    if (searchPathLen > prefixLen && maxVersion(n) > readVersion) {
+      return false;
+    }
+    if (n->parent == nullptr) {
+      return true;
+    }
+    auto next = getChildGeq(n->parent, n->parentsIndex + 1);
+    if (next == nullptr) {
+      searchPathLen -= 1 + n->partialKeyLen;
+      n = n->parent;
+    } else {
+      searchPathLen -= n->partialKeyLen;
+      n = next;
+      searchPathLen += n->partialKeyLen;
+      goto downLeftSpine;
     }
-    ok = n->entry.rangeVersion <= readVersion;
-    return true;
   }
-};
+downLeftSpine:
+  for (; !n->entryPresent; n = getFirstChildExists(n)) {
+  }
+  return n->entry.rangeVersion <= readVersion;
+}
 } // namespace
 
 bool checkRangeRead(Node *n, std::span<const uint8_t> begin,
@@ -2759,8 +2705,8 @@ bool checkRangeRead(Node *n, std::span<const uint8_t> begin,
 
   auto remaining = begin.subspan(0, lcp);
   Arena arena;
-  // If the common prefix isn't a prefix of any physical entry in the tree, we
-  // can go to "downLeftSpine"
+
+  // Advance down common prefix, but stay on a physical path in the tree
   for (;; ++tls->range_read_iterations_accum) {
     assert(getSearchPath(arena, n) <=>
                begin.subspan(0, lcp - remaining.size()) ==
@@ -2801,47 +2747,17 @@ bool checkRangeRead(Node *n, std::span<const uint8_t> begin,
   lcp -= consumed;
 
   if (lcp == int(begin.size())) {
-    CheckRangeRightSide checkRangeRightSide{n, end, lcp, readVersion, tls};
-    while (!checkRangeRightSide.step())
-      ;
-    return checkRangeRightSide.ok;
+    return checkRangeRightSide(n, end, lcp, readVersion, tls);
   }
 
-  if (!checkRangeStartsWith(n, begin.subspan(0, lcp), begin[lcp], end[lcp],
-                            readVersion, tls)) {
-    return false;
-  }
-
-  // This makes it safe to check maxVersion within CheckRangeLeftSide. If this
+  // This makes it safe to check maxVersion within checkRangeLeftSide. If this
   // were false, then we would have returned above since lcp == begin.size().
   assert(!(n->parent == nullptr && begin.size() == 0));
-  CheckRangeLeftSide checkRangeLeftSide{n, begin, lcp + 1, readVersion, tls};
-  CheckRangeRightSide checkRangeRightSide{n, end, lcp + 1, readVersion, tls};
 
-  for (;;) {
-    bool leftDone = checkRangeLeftSide.step();
-    bool rightDone = checkRangeRightSide.step();
-    if (!leftDone && !rightDone) {
-      tls->range_read_iterations_accum += 2;
-      continue;
-    }
-    if (leftDone && rightDone) {
-      break;
-    } else if (leftDone) {
-      while (!checkRangeRightSide.step()) {
-        ++tls->range_read_iterations_accum;
-      }
-      break;
-    } else {
-      assert(rightDone);
-      while (!checkRangeLeftSide.step()) {
-        ++tls->range_read_iterations_accum;
-      }
-    }
-    break;
-  }
-
-  return checkRangeLeftSide.ok && checkRangeRightSide.ok;
+  return checkRangeStartsWith(n, begin.subspan(0, lcp), begin[lcp], end[lcp],
+                              readVersion, tls) &&
+         checkRangeLeftSide(n, begin, lcp + 1, readVersion, tls) &&
+         checkRangeRightSide(n, end, lcp + 1, readVersion, tls);
 }
 
 #ifdef __x86_64__
@@ -2864,9 +2780,8 @@ checkMaxBetweenExclusiveImpl<true>(Node *n, int begin, int end,
 // of the result will have `maxVersion` set to `writeVersion` as a
 // postcondition. Nodes along the search path may be invalidated. Callers must
 // ensure that the max version of the self argument is updated.
-[[nodiscard]]
-Node **insert(Node **self, std::span<const uint8_t> key,
-              InternalVersionT writeVersion, WriteContext *tls) {
+[[nodiscard]] Node **insert(Node **self, std::span<const uint8_t> key,
+                            InternalVersionT writeVersion, WriteContext *tls) {
 
   for (; key.size() != 0; ++tls->accum.insert_iterations) {
     self = &getOrCreateChild(*self, key, writeVersion, tls);
@@ -3101,6 +3016,8 @@ struct __attribute__((visibility("hidden"))) ConflictSet::Impl {
     tls.impl = this;
     int64_t check_byte_accum = 0;
     for (int i = 0; i < count; ++i) {
+      assert(reads[i].readVersion >= 0);
+      assert(reads[i].readVersion <= newestVersionFullPrecision);
       const auto &r = reads[i];
       check_byte_accum += r.begin.len + r.end.len;
       auto begin = std::span<const uint8_t>(r.begin.p, r.begin.len);
@@ -3137,10 +3054,12 @@ struct __attribute__((visibility("hidden"))) ConflictSet::Impl {
   }
 
   void addWrites(const WriteRange *writes, int count, int64_t writeVersion) {
+#if !USE_64_BIT
     // There could be other conflict sets in the same thread. We need
     // InternalVersionT::zero to be correct for this conflict set for the
     // lifetime of the current call frame.
     InternalVersionT::zero = tls.zero = oldestVersion;
+#endif
 
     assert(writeVersion >= newestVersionFullPrecision);
     assert(tls.accum.entries_erased == 0);
@@ -3155,7 +3074,10 @@ struct __attribute__((visibility("hidden"))) ConflictSet::Impl {
 
       newestVersionFullPrecision = writeVersion;
       newest_version.set(newestVersionFullPrecision);
-      setOldestVersion(newestVersionFullPrecision - kNominalVersionWindow);
+      if (newestVersionFullPrecision - kNominalVersionWindow >
+          oldestVersionFullPrecision) {
+        setOldestVersion(newestVersionFullPrecision - kNominalVersionWindow);
+      }
       while (oldestExtantVersion <
              newestVersionFullPrecision - kMaxCorrectVersionWindow) {
         gcScanStep(1000);
@@ -3163,7 +3085,10 @@ struct __attribute__((visibility("hidden"))) ConflictSet::Impl {
     } else {
       newestVersionFullPrecision = writeVersion;
       newest_version.set(newestVersionFullPrecision);
-      setOldestVersion(newestVersionFullPrecision - kNominalVersionWindow);
+      if (newestVersionFullPrecision - kNominalVersionWindow >
+          oldestVersionFullPrecision) {
+        setOldestVersion(newestVersionFullPrecision - kNominalVersionWindow);
+      }
     }
 
     for (int i = 0; i < count; ++i) {
@@ -3185,7 +3110,6 @@ struct __attribute__((visibility("hidden"))) ConflictSet::Impl {
                           0) *
         2;
 
-    memory_bytes.set(totalBytes);
     point_writes_total.add(tls.accum.point_writes);
     range_writes_total.add(tls.accum.range_writes);
     nodes_allocated_total.add(tls.accum.nodes_allocated);
@@ -3248,14 +3172,22 @@ struct __attribute__((visibility("hidden"))) ConflictSet::Impl {
     return fuel;
   }
 
-  void setOldestVersion(int64_t o) {
-    if (o <= oldestVersionFullPrecision) {
+  void setOldestVersion(int64_t newOldestVersion) {
+    assert(newOldestVersion >= 0);
+    assert(newOldestVersion <= newestVersionFullPrecision);
+    // If addWrites advances oldestVersion to keep within valid window, a
+    // subsequent setOldestVersion can be legitimately called with a version
+    // older than `oldestVersionFullPrecision`. < instead of <= so that we can
+    // do garbage collection work without advancing the oldest version.
+    if (newOldestVersion < oldestVersionFullPrecision) {
       return;
     }
-    InternalVersionT oldestVersion{o};
-    this->oldestVersionFullPrecision = o;
+    InternalVersionT oldestVersion{newOldestVersion};
+    this->oldestVersionFullPrecision = newOldestVersion;
     this->oldestVersion = oldestVersion;
+#if !USE_64_BIT
     InternalVersionT::zero = tls.zero = oldestVersion;
+#endif
 #ifdef NDEBUG
     // This is here for performance reasons, since we want to amortize the cost
     // of storing the search path as a string. In tests, we want to exercise the
@@ -3266,7 +3198,6 @@ struct __attribute__((visibility("hidden"))) ConflictSet::Impl {
 #endif
     keyUpdates = gcScanStep(keyUpdates);
 
-    memory_bytes.set(totalBytes);
     nodes_allocated_total.add(std::exchange(tls.accum.nodes_allocated, 0));
     nodes_released_total.add(std::exchange(tls.accum.nodes_released, 0));
     entries_inserted_total.add(std::exchange(tls.accum.entries_inserted, 0));
@@ -3304,14 +3235,17 @@ struct __attribute__((visibility("hidden"))) ConflictSet::Impl {
     root->entry.pointVersion = this->oldestVersion;
     root->entry.rangeVersion = this->oldestVersion;
 
+#if !USE_64_BIT
     InternalVersionT::zero = tls.zero = this->oldestVersion;
+#endif
 
     // Intentionally not resetting totalBytes
   }
 
   explicit Impl(int64_t oldestVersion) {
+    assert(oldestVersion >= 0);
     init(oldestVersion);
-    initMetrics();
+    metrics = initMetrics(metricsList, metricsCount);
   }
   ~Impl() {
     eraseTree(root, &tls);
@@ -3334,23 +3268,12 @@ struct __attribute__((visibility("hidden"))) ConflictSet::Impl {
 
   MetricsV1 *metrics;
   int metricsCount = 0;
-  void initMetrics() {
-    metrics = (MetricsV1 *)safe_malloc(metricsCount * sizeof(metrics[0]));
-    for (auto [i, m] = std::make_tuple(metricsCount - 1, metricList); i >= 0;
-         --i, m = m->prev) {
-      metrics[i].name = m->name;
-      metrics[i].help = m->help;
-      metrics[i].p = m;
-      metrics[i].type = m->type;
-    }
-  }
-
-  Metric *metricList = nullptr;
+  Metric *metricsList = nullptr;
 
 #define GAUGE(name, help)                                                      \
-  Gauge name { this, #name, help }
+  Gauge name { metricsList, metricsCount, #name, help }
 #define COUNTER(name, help)                                                    \
-  Counter name { this, #name, help }
+  Counter name { metricsList, metricsCount, #name, help }
   // ==================== METRICS DEFINITIONS ====================
   COUNTER(point_read_total, "Total number of point reads checked");
   COUNTER(point_read_short_circuit_total,
@@ -3416,13 +3339,6 @@ struct __attribute__((visibility("hidden"))) ConflictSet::Impl {
   }
 };
 
-Metric::Metric(ConflictSet::Impl *impl, const char *name, const char *help,
-               ConflictSet::MetricsV1::Type type)
-    : prev(std::exchange(impl->metricList, this)), name(name), help(help),
-      type(type), value(0) {
-  ++impl->metricsCount;
-}
-
 Node *&getInTree(Node *n, ConflictSet::Impl *impl) {
   return n->parent == nullptr ? impl->root
                               : getChildExists(n->parent, n->parentsIndex);
@@ -3441,6 +3357,7 @@ void internal_addWrites(ConflictSet::Impl *impl,
   mallocBytesDelta = 0;
   impl->addWrites(writes, count, writeVersion);
   impl->totalBytes += mallocBytesDelta;
+  impl->memory_bytes.set(impl->totalBytes);
 #if SHOW_MEMORY
   if (impl->totalBytes != mallocBytes) {
     abort();
@@ -3452,6 +3369,7 @@ void internal_setOldestVersion(ConflictSet::Impl *impl, int64_t oldestVersion) {
   mallocBytesDelta = 0;
   impl->setOldestVersion(oldestVersion);
   impl->totalBytes += mallocBytesDelta;
+  impl->memory_bytes.set(impl->totalBytes);
 #if SHOW_MEMORY
   if (impl->totalBytes != mallocBytes) {
     abort();
@@ -3704,13 +3622,13 @@ std::string getSearchPath(Node *n) {
         fprintf(file,
                 " k_%p [label=\"m=%" PRId64 " p=%" PRId64 " r=%" PRId64
                 "\n%s\", pos=\"%d,%d!\"];\n",
-                (void *)n, maxVersion(n).toInt64(),
+                (void *)n, n->parent == nullptr ? -1 : maxVersion(n).toInt64(),
                 n->entry.pointVersion.toInt64(),
                 n->entry.rangeVersion.toInt64(),
                 getPartialKeyPrintable(n).c_str(), x, y);
       } else {
         fprintf(file, " k_%p [label=\"m=%" PRId64 "\n%s\", pos=\"%d,%d!\"];\n",
-                (void *)n, maxVersion(n).toInt64(),
+                (void *)n, n->parent == nullptr ? -1 : maxVersion(n).toInt64(),
                 getPartialKeyPrintable(n).c_str(), x, y);
       }
       x += kSeparation;
@@ -3751,6 +3669,9 @@ Node *firstGeq(Node *n, std::string_view key) {
       n, std::span<const uint8_t>((const uint8_t *)key.data(), key.size()));
 }
 
+#if USE_64_BIT
+void checkVersionsGeqOldestExtant(Node *, InternalVersionT) {}
+#else
 void checkVersionsGeqOldestExtant(Node *n,
                                   InternalVersionT oldestExtantVersion) {
   if (n->entryPresent) {
@@ -3794,6 +3715,7 @@ void checkVersionsGeqOldestExtant(Node *n,
     abort();
   }
 }
+#endif
 
 [[maybe_unused]] InternalVersionT
 checkMaxVersion(Node *root, Node *node, InternalVersionT oldestVersion,
@@ -4001,6 +3923,73 @@ struct __attribute__((visibility("default"))) PeakPrinter {
 
 #ifdef ENABLE_MAIN
 
+#define ANKERL_NANOBENCH_IMPLEMENT
+#include "third_party/nanobench.h"
+
+template <int kN> void benchRezero() {
+  static_assert(kN % 16 == 0);
+  ankerl::nanobench::Bench bench;
+  InternalVersionT vs[kN];
+  InternalVersionT zero;
+  bench.run("rezero" + std::to_string(kN), [&]() {
+    bench.doNotOptimizeAway(vs);
+    bench.doNotOptimizeAway(zero);
+    for (int i = 0; i < kN; i += 16) {
+      rezero16(vs + i, zero);
+    }
+  });
+}
+
+template <int kN> void benchScan1() {
+  static_assert(kN % 16 == 0);
+  ankerl::nanobench::Bench bench;
+  InternalVersionT vs[kN];
+  uint8_t is[kN];
+  uint8_t begin;
+  uint8_t end;
+  InternalVersionT v;
+  bench.run("scan" + std::to_string(kN), [&]() {
+    bench.doNotOptimizeAway(vs);
+    bench.doNotOptimizeAway(is);
+    bench.doNotOptimizeAway(begin);
+    bench.doNotOptimizeAway(end);
+    bench.doNotOptimizeAway(v);
+    for (int i = 0; i < kN; i += 16) {
+      scan16</*kAVX512=*/true>(vs + i, is + i, begin, end, v);
+    }
+  });
+}
+
+template <int kN> void benchScan2() {
+  static_assert(kN % 16 == 0);
+  ankerl::nanobench::Bench bench;
+  InternalVersionT vs[kN];
+  uint8_t is[kN];
+  uint8_t begin;
+  uint8_t end;
+  InternalVersionT v;
+  bench.run("scan" + std::to_string(kN), [&]() {
+    bench.doNotOptimizeAway(vs);
+    bench.doNotOptimizeAway(begin);
+    bench.doNotOptimizeAway(end);
+    bench.doNotOptimizeAway(v);
+    for (int i = 0; i < kN; i += 16) {
+      scan16</*kAVX512=*/true>(vs + i, begin, end, v);
+    }
+  });
+}
+
+void benchLCP(int len) {
+  ankerl::nanobench::Bench bench;
+  std::vector<uint8_t> lhs(len);
+  std::vector<uint8_t> rhs(len);
+  bench.run("lcp " + std::to_string(len), [&]() {
+    bench.doNotOptimizeAway(lhs);
+    bench.doNotOptimizeAway(rhs);
+    bench.doNotOptimizeAway(longestCommonPrefix(lhs.data(), rhs.data(), len));
+  });
+}
+
 void printTree() {
   int64_t writeVersion = 0;
   ConflictSet::Impl cs{writeVersion};
@@ -4022,7 +4011,11 @@ void printTree() {
   debugPrintDot(stdout, cs.root, &cs);
 }
 
-int main(void) { printTree(); }
+int main(void) {
+  for (int i = 0; i < 256; ++i) {
+    benchLCP(i);
+  }
+}
 #endif
 
 #ifdef ENABLE_FUZZ

Internal.h

@@ -20,7 +20,6 @@ using namespace weaselab;
 #include <thread>
 #include <unordered_set>
 #include <utility>
-#include <vector>
 
 #include <callgrind.h>
 

Jenkinsfile

@@ -48,6 +48,17 @@ pipeline {
                 recordIssues(tools: [clang()])
             }
         }
+        stage('64 bit versions') {
+            agent {
+                dockerfile {
+                    args '-v /home/jenkins/ccache:/ccache'
+                    reuseNode true
+                }
+            }
+            steps {
+                CleanBuildAndTest("-DCMAKE_CXX_FLAGS=-DUSE_64_BIT=1")
+            }
+        }
         stage('Debug') {
             agent {
                 dockerfile {
@@ -118,7 +129,7 @@ pipeline {
             }
             steps {
                 script {
-                    filter_args = "-f ConflictSet.cpp -f LongestCommonPrefix.h"
+                    filter_args = "-f ConflictSet.cpp -f LongestCommonPrefix.h -f Metrics.h"
                 }
                 CleanBuildAndTest("-DCMAKE_C_COMPILER=gcc -DCMAKE_CXX_COMPILER=g++ -DCMAKE_C_FLAGS=--coverage -DCMAKE_CXX_FLAGS=--coverage -DCMAKE_BUILD_TYPE=Debug -DDISABLE_TSAN=ON")
                 sh """

Metrics.h

@@ -0,0 +1,64 @@
+#pragma once
+
+#include "ConflictSet.h"
+#include "Internal.h"
+
+#include <assert.h>
+#include <atomic>
+#include <tuple>
+
+struct Metric {
+  Metric *prev;
+  const char *name;
+  const char *help;
+  weaselab::ConflictSet::MetricsV1::Type type;
+  std::atomic<int64_t> value;
+
+protected:
+  Metric(Metric *&metricList, int &metricsCount, const char *name,
+         const char *help, weaselab::ConflictSet::MetricsV1::Type type)
+      : prev(std::exchange(metricList, this)), name(name), help(help),
+        type(type), value(0) {
+    ++metricsCount;
+  }
+};
+
+struct Gauge : private Metric {
+  Gauge(Metric *&metricList, int &metricsCount, const char *name,
+        const char *help)
+      : Metric(metricList, metricsCount, name, help,
+               weaselab::ConflictSet::MetricsV1::Gauge) {}
+
+  void set(int64_t value) {
+    this->value.store(value, std::memory_order_relaxed);
+  }
+};
+
+struct Counter : private Metric {
+  Counter(Metric *&metricList, int &metricsCount, const char *name,
+          const char *help)
+      : Metric(metricList, metricsCount, name, help,
+               weaselab::ConflictSet::MetricsV1::Counter) {}
+  // Expensive. Accumulate locally and then call add instead of repeatedly
+  // calling add.
+  void add(int64_t value) {
+    assert(value >= 0);
+    static_assert(std::atomic<int64_t>::is_always_lock_free);
+    this->value.fetch_add(value, std::memory_order_relaxed);
+  }
+};
+
+inline weaselab::ConflictSet::MetricsV1 *initMetrics(Metric *metricsList,
+                                                     int metricsCount) {
+  weaselab::ConflictSet::MetricsV1 *metrics =
+      (weaselab::ConflictSet::MetricsV1 *)safe_malloc(metricsCount *
+                                                      sizeof(metrics[0]));
+  for (auto [i, m] = std::make_tuple(metricsCount - 1, metricsList); i >= 0;
+       --i, m = m->prev) {
+    metrics[i].name = m->name;
+    metrics[i].help = m->help;
+    metrics[i].p = m;
+    metrics[i].type = m->type;
+  }
+  return metrics;
+}

README.md

@@ -24,15 +24,15 @@ Hardware for all benchmarks is an AMD Ryzen 9 7900 with (2x32GB) 5600MT/s CL28-3
 
 |               ns/op |                op/s |    err% |          ins/op |          cyc/op |    IPC |         bra/op |   miss% |     total | benchmark
 |--------------------:|--------------------:|--------:|----------------:|----------------:|-------:|---------------:|--------:|----------:|:----------
-|               11.18 |       89,455,125.34 |    0.6% |          185.37 |           57.08 |  3.248 |          41.51 |    0.4% |      0.01 | `point reads`
-|               14.53 |       68,800,688.89 |    0.4% |          282.41 |           74.80 |  3.776 |          55.06 |    0.3% |      0.01 | `prefix reads`
-|               36.54 |       27,367,576.87 |    0.2% |          798.06 |          188.90 |  4.225 |         141.69 |    0.2% |      0.01 | `range reads`
-|               16.69 |       59,912,106.02 |    0.6% |          314.57 |           86.29 |  3.645 |          39.84 |    0.4% |      0.01 | `point writes`
-|               30.09 |       33,235,744.07 |    0.5% |          591.33 |          155.92 |  3.793 |          82.69 |    0.2% |      0.01 | `prefix writes`
-|               35.77 |       27,956,388.03 |    1.4% |          682.25 |          187.63 |  3.636 |          96.12 |    0.1% |      0.01 | `range writes`
-|               74.04 |       13,505,408.41 |    2.7% |        1,448.95 |          392.10 |  3.695 |         260.53 |    0.1% |      0.01 | `monotonic increasing point writes`
-|          330,984.50 |            3,021.29 |    1.9% |    3,994,153.50 |    1,667,309.00 |  2.396 |     806,019.50 |    0.0% |      0.01 | `worst case for radix tree`
-|               92.46 |       10,814,961.65 |    0.5% |        1,800.00 |          463.41 |  3.884 |         297.00 |    0.0% |      0.01 | `create and destroy`
+|               12.88 |       77,653,350.77 |    0.5% |          185.37 |           64.45 |  2.876 |          41.51 |    0.4% |      0.01 | `point reads`
+|               14.67 |       68,179,354.49 |    0.1% |          271.44 |           73.40 |  3.698 |          53.70 |    0.3% |      0.01 | `prefix reads`
+|               34.84 |       28,701,444.36 |    0.3% |          715.74 |          175.27 |  4.084 |         127.30 |    0.2% |      0.01 | `range reads`
+|               17.12 |       58,422,988.28 |    0.2% |          314.30 |           86.11 |  3.650 |          39.82 |    0.4% |      0.01 | `point writes`
+|               31.42 |       31,830,804.65 |    0.1% |          591.06 |          158.07 |  3.739 |          82.67 |    0.2% |      0.01 | `prefix writes`
+|               37.37 |       26,759,432.70 |    2.2% |          681.98 |          188.95 |  3.609 |          96.10 |    0.1% |      0.01 | `range writes`
+|               76.72 |       13,035,140.63 |    2.3% |        1,421.28 |          387.17 |  3.671 |         257.76 |    0.1% |      0.01 | `monotonic increasing point writes`
+|          297,452.00 |            3,361.89 |    0.9% |    3,508,083.00 |    1,500,834.67 |  2.337 |     727,525.33 |    0.1% |      0.01 | `worst case for radix tree`
+|               87.70 |       11,402,490.60 |    1.0% |        1,795.00 |          442.09 |  4.060 |         297.00 |    0.0% |      0.01 | `create and destroy`
 
 # "Real data" test
 

ServerBench.cpp

@@ -21,7 +21,7 @@
 
 std::atomic<int64_t> transactions;
 
-constexpr int kBaseSearchDepth = 32;
+constexpr int kBaseSearchDepth = 115;
 constexpr int kWindowSize = 10000000;
 
 std::string numToKey(int64_t num) {

SkipList.cpp

@@ -22,9 +22,11 @@
 
 #include "ConflictSet.h"
 #include "Internal.h"
+#include "Metrics.h"
 
 #include <algorithm>
 #include <span>
+#include <vector>
 
 std::span<const uint8_t> keyAfter(Arena &arena, std::span<const uint8_t> key) {
   auto result =
@@ -115,15 +117,6 @@ bool operator==(const KeyInfo &lhs, const KeyInfo &rhs) {
   return !(lhs < rhs || rhs < lhs);
 }
 
-void swapSort(std::vector<KeyInfo> &points, int a, int b) {
-  if (points[b] < points[a]) {
-    KeyInfo temp;
-    temp = points[a];
-    points[a] = points[b];
-    points[b] = temp;
-  }
-}
-
 struct SortTask {
   int begin;
   int size;
@@ -183,13 +176,6 @@ void sortPoints(std::vector<KeyInfo> &points) {
   }
 }
 
-static thread_local uint32_t g_seed = 0;
-
-static inline int skfastrand() {
-  g_seed = g_seed * 1664525L + 1013904223L;
-  return g_seed;
-}
-
 static int compare(const StringRef &a, const StringRef &b) {
   int c = memcmp(a.data(), b.data(), std::min(a.size(), b.size()));
   if (c < 0)
@@ -215,20 +201,24 @@ struct ReadConflictRange {
   }
 };
 
+static constexpr int MaxLevels = 26;
+
+struct RandomLevel {
+  explicit RandomLevel(uint32_t seed) : seed(seed) {}
+
+  int next() {
+    int result = __builtin_clz(seed | (uint32_t(-1) >> (MaxLevels - 1)));
+    seed = seed * 1664525L + 1013904223L;
+    return result;
+  }
+
+private:
+  uint32_t seed;
+};
+
 class SkipList {
 private:
-  static constexpr int MaxLevels = 26;
-
-  int randomLevel() const {
-    uint32_t i = uint32_t(skfastrand()) >> (32 - (MaxLevels - 1));
-    int level = 0;
-    while (i & 1) {
-      i >>= 1;
-      level++;
-    }
-    assert(level < MaxLevels);
-    return level;
-  }
+  RandomLevel randomLevel{0};
 
   // Represent a node in the SkipList. The node has multiple (i.e., level)
   // pointers to other nodes, and keeps a record of the max versions for each
@@ -426,27 +416,33 @@ public:
   }
   void swap(SkipList &other) { std::swap(header, other.header); }
 
-  void addConflictRanges(const Finger *fingers, int rangeCount,
-                         Version version) {
+  // Returns the change in the number of entries
+  int64_t addConflictRanges(const Finger *fingers, int rangeCount,
+                            Version version) {
+    int64_t result = rangeCount;
     for (int r = rangeCount - 1; r >= 0; r--) {
       const Finger &startF = fingers[r * 2];
       const Finger &endF = fingers[r * 2 + 1];
 
-      if (endF.found() == nullptr)
+      if (endF.found() == nullptr) {
+        ++result;
         insert(endF, endF.finger[0]->getMaxVersion(0));
+      }
 
-      remove(startF, endF);
+      result -= remove(startF, endF);
       insert(startF, version);
     }
+    return result;
   }
 
-  void detectConflicts(ReadConflictRange *ranges, int count,
-                       ConflictSet::Result *transactionConflictStatus) const {
+  // Return number of iterations of main loop
+  int detectConflicts(ReadConflictRange *ranges, int count,
+                      ConflictSet::Result *transactionConflictStatus) const {
     const int M = 16;
     int nextJob[M];
     CheckMax inProgress[M];
     if (!count)
-      return;
+      return 0;
 
     int started = std::min(M, count);
     for (int i = 0; i < started; i++) {
@@ -457,8 +453,9 @@ public:
 
     int prevJob = started - 1;
     int job = 0;
+    int iters = 0;
     // vtune: 340 parts
-    while (true) {
+    for (;; ++iters) {
       if (inProgress[job].advance()) {
         if (started == count) {
           if (prevJob == job)
@@ -474,6 +471,7 @@ public:
       prevJob = job;
       job = nextJob[job];
     }
+    return iters;
   }
 
   void find(const StringRef *values, Finger *results, int *temp, int count) {
@@ -567,9 +565,10 @@ public:
   }
 
 private:
-  void remove(const Finger &start, const Finger &end) {
+  // Returns the number of entries removed
+  int64_t remove(const Finger &start, const Finger &end) {
     if (start.finger[0] == end.finger[0])
-      return;
+      return 0;
 
     Node *x = start.finger[0]->getNext(0);
 
@@ -578,17 +577,20 @@ private:
       if (start.finger[i] != end.finger[i])
         start.finger[i]->setNext(i, end.finger[i]->getNext(i));
 
+    int64_t result = 0;
     while (true) {
       Node *next = x->getNext(0);
       x->destroy();
+      ++result;
       if (x == end.finger[0])
         break;
       x = next;
     }
+    return result;
   }
 
   void insert(const Finger &f, Version version) {
-    int level = randomLevel();
+    int level = randomLevel.next();
     // std::cout << std::string((const char*)value,length) << " level: " <<
     // level << std::endl;
     Node *x = Node::create(f.value, level);
@@ -704,17 +706,27 @@ private:
   };
 };
 
-struct SkipListConflictSet {};
+struct ReadContext {
+  int64_t commits_accum = 0;
+  int64_t conflicts_accum = 0;
+  int64_t too_olds_accum = 0;
+  int64_t check_bytes_accum = 0;
+};
 
 struct __attribute__((visibility("hidden"))) ConflictSet::Impl {
   Impl(int64_t oldestVersion)
       : oldestVersion(oldestVersion), newestVersion(oldestVersion),
-        skipList(oldestVersion) {}
+        skipList(oldestVersion) {
+    metrics = initMetrics(metricsList, metricsCount);
+  }
+  ~Impl() { safe_free(metrics, metricsCount * sizeof(metrics[0])); }
   void check(const ConflictSet::ReadRange *reads, ConflictSet::Result *results,
-             int count) const {
+             int count) {
+    ReadContext tls;
     Arena arena;
     auto *ranges = new (arena) ReadConflictRange[count];
     for (int i = 0; i < count; ++i) {
+      tls.check_bytes_accum += reads[i].begin.len + reads[i].end.len;
       ranges[i].begin = {reads[i].begin.p, size_t(reads[i].begin.len)};
       ranges[i].end = reads[i].end.len > 0
                           ? StringRef{reads[i].end.p, size_t(reads[i].end.len)}
@@ -722,13 +734,22 @@ struct __attribute__((visibility("hidden"))) ConflictSet::Impl {
       ranges[i].version = reads[i].readVersion;
       results[i] = ConflictSet::Commit;
     }
-    skipList.detectConflicts(ranges, count, results);
+    int iters = skipList.detectConflicts(ranges, count, results);
     for (int i = 0; i < count; ++i) {
       if (reads[i].readVersion < oldestVersion ||
           reads[i].readVersion < newestVersion - 2e9) {
         results[i] = TooOld;
       }
+      tls.commits_accum += results[i] == Commit;
+      tls.conflicts_accum += results[i] == Conflict;
+      tls.too_olds_accum += results[i] == TooOld;
     }
+    range_read_iterations_total.add(iters);
+    range_read_total.add(count);
+    commits_total.add(tls.commits_accum);
+    conflicts_total.add(tls.conflicts_accum);
+    too_olds_total.add(tls.too_olds_accum);
+    check_bytes_total.add(tls.check_bytes_accum);
   }
 
   void addWrites(const ConflictSet::WriteRange *writes, int count,
@@ -775,27 +796,33 @@ struct __attribute__((visibility("hidden"))) ConflictSet::Impl {
     StringRef values[stripeSize];
     int64_t writeVersions[stripeSize / 2];
     int ss = stringCount - (stripes - 1) * stripeSize;
+    int64_t entryDelta = 0;
     for (int s = stripes - 1; s >= 0; s--) {
       for (int i = 0; i * 2 < ss; ++i) {
         const auto &w = combinedWriteConflictRanges[s * stripeSize / 2 + i];
         values[i * 2] = w.first;
         values[i * 2 + 1] = w.second;
-        keyUpdates += 3;
       }
       skipList.find(values, fingers, temp, ss);
-      skipList.addConflictRanges(fingers, ss / 2, writeVersion);
+      entryDelta += skipList.addConflictRanges(fingers, ss / 2, writeVersion);
       ss = stripeSize;
     }
+
+    // Run gc at least 200% the rate we're inserting entries
+    keyUpdates += std::max<int64_t>(entryDelta, 0) * 2;
   }
 
   void setOldestVersion(int64_t oldestVersion) {
+    // This isn't 100% accurate. It overcounts if you hit the end
+    gc_iterations_total.add(keyUpdates);
+
     assert(oldestVersion >= this->oldestVersion);
     this->oldestVersion = oldestVersion;
     SkipList::Finger finger;
     int temp;
     std::span<const uint8_t> key = removalKey;
     skipList.find(&key, &finger, &temp, 1);
-    skipList.removeBefore(oldestVersion, finger, std::exchange(keyUpdates, 10));
+    skipList.removeBefore(oldestVersion, finger, std::exchange(keyUpdates, 0));
     removalArena = Arena();
     removalKey = copyToArena(
         removalArena, {finger.getValue().data(), finger.getValue().size()});
@@ -803,8 +830,56 @@ struct __attribute__((visibility("hidden"))) ConflictSet::Impl {
 
   int64_t totalBytes = 0;
 
+  MetricsV1 *metrics;
+  int metricsCount = 0;
+  Metric *metricsList = nullptr;
+
+#define GAUGE(name, help)                                                      \
+  Gauge name { metricsList, metricsCount, #name, help }
+#define COUNTER(name, help)                                                    \
+  Counter name { metricsList, metricsCount, #name, help }
+  // ==================== METRICS DEFINITIONS ====================
+  COUNTER(range_read_total, "Total number of range reads checked");
+  COUNTER(range_read_iterations_total,
+          "Total number of iterations of the main loops for range read checks");
+  COUNTER(commits_total,
+          "Total number of checks where the result is \"commit\"");
+  COUNTER(conflicts_total,
+          "Total number of checks where the result is \"conflict\"");
+  COUNTER(too_olds_total,
+          "Total number of checks where the result is \"too old\"");
+  COUNTER(check_bytes_total, "Total number of key bytes checked");
+  GAUGE(memory_bytes, "Total number of bytes in use");
+  COUNTER(nodes_allocated_total,
+          "The total number of physical tree nodes allocated");
+  COUNTER(nodes_released_total,
+          "The total number of physical tree nodes released");
+  COUNTER(insert_iterations_total,
+          "The total number of iterations of the main loop for insertion. "
+          "Includes searches where the entry already existed, and so insertion "
+          "did not take place");
+  COUNTER(entries_inserted_total,
+          "The total number of entries inserted in the tree");
+  COUNTER(entries_erased_total,
+          "The total number of entries erased from the tree");
+  COUNTER(
+      gc_iterations_total,
+      "The total number of iterations of the main loop for garbage collection");
+  COUNTER(write_bytes_total, "Total number of key bytes in calls to addWrites");
+  GAUGE(oldest_version,
+        "The lowest version that doesn't result in \"TooOld\" for checks");
+  GAUGE(newest_version, "The version of the most recent call to addWrites");
+  // ==================== END METRICS DEFINITIONS ====================
+#undef GAUGE
+#undef COUNTER
+
+  void getMetricsV1(MetricsV1 **metrics, int *count) {
+    *metrics = this->metrics;
+    *count = metricsCount;
+  }
+
 private:
-  int64_t keyUpdates = 10;
+  int64_t keyUpdates = 0;
   Arena removalArena;
   std::span<const uint8_t> removalKey;
   int64_t oldestVersion;
@@ -825,6 +900,7 @@ void internal_addWrites(ConflictSet::Impl *impl,
   mallocBytesDelta = 0;
   impl->addWrites(writes, count, writeVersion);
   impl->totalBytes += mallocBytesDelta;
+  impl->memory_bytes.set(impl->totalBytes);
 #if SHOW_MEMORY
   if (impl->totalBytes != mallocBytes) {
     abort();
@@ -836,6 +912,7 @@ void internal_setOldestVersion(ConflictSet::Impl *impl, int64_t oldestVersion) {
   mallocBytesDelta = 0;
   impl->setOldestVersion(oldestVersion);
   impl->totalBytes += mallocBytesDelta;
+  impl->memory_bytes.set(impl->totalBytes);
 #if SHOW_MEMORY
   if (impl->totalBytes != mallocBytes) {
     abort();
@@ -859,12 +936,11 @@ int64_t internal_getBytes(ConflictSet::Impl *impl) { return impl->totalBytes; }
 
 void internal_getMetricsV1(ConflictSet::Impl *impl,
                            ConflictSet::MetricsV1 **metrics, int *count) {
-  *metrics = nullptr;
-  *count = 0;
+  return impl->getMetricsV1(metrics, count);
 }
 
 double internal_getMetricValue(const ConflictSet::MetricsV1 *metric) {
-  return 0;
+  return ((Metric *)metric->p)->value.load(std::memory_order_relaxed);
 }
 
 void ConflictSet::check(const ReadRange *reads, Result *results,