3 sample options and more benchmarking changes

GraphStreamingProject · Mar 13, 2024 · 52491cb · 52491cb
1 parent 2c2bf59
commit 52491cb
Show file tree

Hide file tree

Showing 2 changed files with 89 additions and 7 deletions.
diff --git a/src/sketch.cpp b/src/sketch.cpp
@@ -91,6 +91,45 @@ void Sketch::zero_contents() {
   reset_sample_state();
 }
 
+#if 1
+SketchSample Sketch::sample() {
+  if (sample_idx >= num_samples) {
+    throw OutOfSamplesException(seed, num_samples, sample_idx);
+  }
+
+  size_t idx = sample_idx++;
+  size_t first_column = idx * cols_per_sample;
+
+  if (Bucket_Boruvka::is_empty(buckets[num_buckets - 1]))
+    return {0, ZERO};  // the "first" bucket is deterministic so if all zero then no edges to return
+
+  if (Bucket_Boruvka::is_good(buckets[num_buckets - 1], checksum_seed()))
+    return {buckets[num_buckets - 1].alpha, GOOD};
+
+  for (size_t col = first_column; col < first_column + cols_per_sample; ++col) {
+    // start from the bottom of the column and iterate up until non-empty found
+    int prev = bkt_per_col - 1;
+    int row = prev - 4;
+    while (Bucket_Boruvka::is_empty(buckets[col * bkt_per_col + row]) && row > 0) {
+      prev = row;
+      row -= 4;
+    }
+    row = prev;
+    while (Bucket_Boruvka::is_empty(buckets[col * bkt_per_col + row]) && row > 0) {
+      --row;
+    }
+
+    // now that we've found a non-zero bucket check next if next 4 buckets good
+    int stop = std::max(row - 5, 0);
+    for (; row >= stop; row--) {
+      if (Bucket_Boruvka::is_good(buckets[col * bkt_per_col + row], checksum_seed()))
+        return {buckets[col * bkt_per_col + row].alpha, GOOD};
+    }
+  }
+  return {0, FAIL};
+}
+#else
+#if 1
 SketchSample Sketch::sample() {
   if (sample_idx >= num_samples) {
     throw OutOfSamplesException(seed, num_samples, sample_idx);
@@ -113,14 +152,39 @@ SketchSample Sketch::sample() {
     }
 
     // now that we've found a non-zero bucket check next if next 4 buckets good
-    int stop = std::max(row - 4, 0);
+    int stop = std::max(row - 5, 0);
     for (; row >= stop; row--) {
       if (Bucket_Boruvka::is_good(buckets[col * bkt_per_col + row], checksum_seed()))
         return {buckets[col * bkt_per_col + row].alpha, GOOD};
     }
   }
   return {0, FAIL};
 }
+#else
+SketchSample Sketch::sample() {
+  if (sample_idx >= num_samples) {
+    throw OutOfSamplesException(seed, num_samples, sample_idx);
+  }
+
+  size_t idx = sample_idx++;
+  size_t first_column = idx * cols_per_sample;
+
+  if (Bucket_Boruvka::is_empty(buckets[num_buckets - 1]))
+    return {0, ZERO};  // the "first" bucket is deterministic so if all zero then no edges to return
+
+  if (Bucket_Boruvka::is_good(buckets[num_buckets - 1], checksum_seed()))
+    return {buckets[num_buckets - 1].alpha, GOOD};
+
+  for (size_t col = first_column; col < first_column + cols_per_sample; ++col) {
+    for (size_t row = 0; row < bkt_per_col; row++) {
+      if (Bucket_Boruvka::is_good(buckets[col * bkt_per_col + row], checksum_seed()))
+        return {buckets[col * bkt_per_col + row].alpha, GOOD};
+    }
+  }
+  return {0, FAIL};
+}
+#endif
+#endif
 
 ExhaustiveSketchSample Sketch::exhaustive_sample() {
   if (sample_idx >= num_samples) {

diff --git a/tools/benchmark/graphcc_bench.cpp b/tools/benchmark/graphcc_bench.cpp
@@ -250,7 +250,7 @@ static void BM_Sketch_Update(benchmark::State& state) {
 BENCHMARK(BM_Sketch_Update)->RangeMultiplier(4)->Ranges({{KB << 4, MB << 4}});
 
 // Benchmark the speed of querying sketches
-static constexpr size_t sample_vec_size = MB;
+static constexpr size_t sample_vec_size = size_t(1) << 40;
 static void BM_Sketch_Sample(benchmark::State& state) {
   constexpr size_t num_sketches = 400;
 
@@ -281,14 +281,14 @@ static void BM_Sketch_Sample(benchmark::State& state) {
       benchmark::Counter(state.iterations() * num_sketches, benchmark::Counter::kIsRate);
   state.counters["Successes"] = double(successes) / (state.iterations() * num_sketches);
 }
-BENCHMARK(BM_Sketch_Sample)->RangeMultiplier(4)->Range(1, sample_vec_size / 2);
+BENCHMARK(BM_Sketch_Sample)->RangeMultiplier(4)->Range(1, sample_vec_size / (1 << 18));
 
 static void BM_Sketch_Merge(benchmark::State& state) {
   size_t n = state.range(0);
-  size_t upds = n / 100;
+  size_t upds = n / (1 << 14);
   size_t seed = get_seed();
-  Sketch s1(n, seed);
-  Sketch s2(n, seed);
+  Sketch s1(n, seed, 5);
+  Sketch s2(n, seed, 5);
 
   for (size_t i = 0; i < upds; i++) {
     s1.update(static_cast<vec_t>(concat_pairing_fn(rand() % n, rand() % n)));
@@ -299,7 +299,25 @@ static void BM_Sketch_Merge(benchmark::State& state) {
     s1.merge(s2);
   }
 }
-BENCHMARK(BM_Sketch_Merge)->RangeMultiplier(10)->Range(1e3, 1e6);
+BENCHMARK(BM_Sketch_Merge)->RangeMultiplier(8)->Range(1 << 18, 1 << 30);
+
+static void BM_Sketch_RangeMerge(benchmark::State& state) {
+  size_t n = state.range(0);
+  size_t upds = n / (1 << 14);
+  size_t seed = get_seed();
+  Sketch s1(n, seed, 5);
+  Sketch s2(n, seed, 5);
+
+  for (size_t i = 0; i < upds; i++) {
+    s1.update(static_cast<vec_t>(concat_pairing_fn(rand() % n, rand() % n)));
+    s2.update(static_cast<vec_t>(concat_pairing_fn(rand() % n, rand() % n)));
+  }
+
+  for (auto _ : state) {
+    s1.range_merge(s2, 3, 1);
+  }
+}
+BENCHMARK(BM_Sketch_RangeMerge)->RangeMultiplier(8)->Range(1 << 18, 1 << 30);
 
 static void BM_Sketch_Serialize(benchmark::State& state) {
   size_t n = state.range(0);