[GitHub] [arrow] wesm commented on a change in pull request #7417: ARROW-9079: [C++] Write benchmark for arithmetic kernels

[GitBox <gi...@apache.org>]

wesm commented on a change in pull request #7417:
URL: https://github.com/apache/arrow/pull/7417#discussion_r439431785



##########
File path: cpp/src/arrow/compute/kernels/scalar_arithmetic_benchmark.cc
##########
@@ -0,0 +1,92 @@
+// Licensed to the Apache Software Foundation (ASF) under one
+// or more contributor license agreements.  See the NOTICE file
+// distributed with this work for additional information
+// regarding copyright ownership.  The ASF licenses this file
+// to you under the Apache License, Version 2.0 (the
+// "License"); you may not use this file except in compliance
+// with the License.  You may obtain a copy of the License at
+//
+//   http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing,
+// software distributed under the License is distributed on an
+// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+// KIND, either express or implied.  See the License for the
+// specific language governing permissions and limitations
+// under the License.
+
+#include "benchmark/benchmark.h"
+
+#include <vector>
+
+#include "arrow/compute/api_scalar.h"
+#include "arrow/compute/benchmark_util.h"
+#include "arrow/compute/kernels/test_util.h"
+#include "arrow/testing/gtest_util.h"
+#include "arrow/testing/random.h"
+
+namespace arrow {
+namespace compute {
+
+constexpr auto kSeed = 0x94378165;
+
+template <typename ArrowType, typename CType = typename ArrowType::c_type>
+static void AddArrayScalarKernel(benchmark::State& state) {
+  RegressionArgs args(state);
+
+  const int64_t array_size = args.size / sizeof(CType);
+  auto min = std::numeric_limits<CType>::lowest();
+  auto max = std::numeric_limits<CType>::max();
+
+  auto rand = random::RandomArrayGenerator(kSeed);
+  auto lhs = std::static_pointer_cast<NumericArray<ArrowType>>(
+      rand.Numeric<ArrowType>(array_size, min, max, args.null_proportion));
+
+  for (auto _ : state) {
+    ABORT_NOT_OK(Add(lhs, Datum(CType(15))).status());
+  }
+}
+
+template <typename ArrowType, typename CType = typename ArrowType::c_type>
+static void AddArrayArrayKernel(benchmark::State& state) {
+  RegressionArgs args(state);
+
+  const int64_t array_size = args.size / sizeof(CType);
+  auto min = std::numeric_limits<CType>::lowest();
+  auto max = std::numeric_limits<CType>::max();
+
+  auto rand = random::RandomArrayGenerator(kSeed);
+  auto lhs = std::static_pointer_cast<NumericArray<ArrowType>>(
+      rand.Numeric<ArrowType>(array_size, min, max, args.null_proportion));
+  auto rhs = std::static_pointer_cast<NumericArray<ArrowType>>(
+      rand.Numeric<ArrowType>(array_size, min, max, args.null_proportion));
+
+  for (auto _ : state) {
+    ABORT_NOT_OK(Add(lhs, rhs).status());
+  }

Review comment:
       can you add the items processed here also? (should be `state.iterations() * lhs->length()`)

##########
File path: cpp/src/arrow/compute/kernels/scalar_arithmetic_benchmark.cc
##########
@@ -0,0 +1,92 @@
+// Licensed to the Apache Software Foundation (ASF) under one
+// or more contributor license agreements.  See the NOTICE file
+// distributed with this work for additional information
+// regarding copyright ownership.  The ASF licenses this file
+// to you under the Apache License, Version 2.0 (the
+// "License"); you may not use this file except in compliance
+// with the License.  You may obtain a copy of the License at
+//
+//   http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing,
+// software distributed under the License is distributed on an
+// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+// KIND, either express or implied.  See the License for the
+// specific language governing permissions and limitations
+// under the License.
+
+#include "benchmark/benchmark.h"
+
+#include <vector>
+
+#include "arrow/compute/api_scalar.h"
+#include "arrow/compute/benchmark_util.h"
+#include "arrow/compute/kernels/test_util.h"
+#include "arrow/testing/gtest_util.h"
+#include "arrow/testing/random.h"
+
+namespace arrow {
+namespace compute {
+
+constexpr auto kSeed = 0x94378165;
+
+template <typename ArrowType, typename CType = typename ArrowType::c_type>
+static void AddArrayScalarKernel(benchmark::State& state) {
+  RegressionArgs args(state);
+
+  const int64_t array_size = args.size / sizeof(CType);
+  auto min = std::numeric_limits<CType>::lowest();
+  auto max = std::numeric_limits<CType>::max();
+
+  auto rand = random::RandomArrayGenerator(kSeed);
+  auto lhs = std::static_pointer_cast<NumericArray<ArrowType>>(
+      rand.Numeric<ArrowType>(array_size, min, max, args.null_proportion));
+
+  for (auto _ : state) {
+    ABORT_NOT_OK(Add(lhs, Datum(CType(15))).status());
+  }
+}
+
+template <typename ArrowType, typename CType = typename ArrowType::c_type>
+static void AddArrayArrayKernel(benchmark::State& state) {
+  RegressionArgs args(state);
+
+  const int64_t array_size = args.size / sizeof(CType);
+  auto min = std::numeric_limits<CType>::lowest();
+  auto max = std::numeric_limits<CType>::max();
+
+  auto rand = random::RandomArrayGenerator(kSeed);
+  auto lhs = std::static_pointer_cast<NumericArray<ArrowType>>(
+      rand.Numeric<ArrowType>(array_size, min, max, args.null_proportion));
+  auto rhs = std::static_pointer_cast<NumericArray<ArrowType>>(
+      rand.Numeric<ArrowType>(array_size, min, max, args.null_proportion));
+
+  for (auto _ : state) {
+    ABORT_NOT_OK(Add(lhs, rhs).status());
+  }
+}
+
+BENCHMARK_TEMPLATE(AddArrayArrayKernel, Int64Type)->Apply(RegressionSetArgs);
+BENCHMARK_TEMPLATE(AddArrayArrayKernel, Int32Type)->Apply(RegressionSetArgs);
+BENCHMARK_TEMPLATE(AddArrayArrayKernel, Int16Type)->Apply(RegressionSetArgs);
+BENCHMARK_TEMPLATE(AddArrayArrayKernel, Int8Type)->Apply(RegressionSetArgs);
+BENCHMARK_TEMPLATE(AddArrayArrayKernel, UInt64Type)->Apply(RegressionSetArgs);
+BENCHMARK_TEMPLATE(AddArrayArrayKernel, UInt32Type)->Apply(RegressionSetArgs);
+BENCHMARK_TEMPLATE(AddArrayArrayKernel, UInt16Type)->Apply(RegressionSetArgs);
+BENCHMARK_TEMPLATE(AddArrayArrayKernel, UInt8Type)->Apply(RegressionSetArgs);
+BENCHMARK_TEMPLATE(AddArrayArrayKernel, FloatType)->Apply(RegressionSetArgs);
+BENCHMARK_TEMPLATE(AddArrayArrayKernel, DoubleType)->Apply(RegressionSetArgs);
+
+BENCHMARK_TEMPLATE(AddArrayScalarKernel, Int64Type)->Apply(RegressionSetArgs);
+BENCHMARK_TEMPLATE(AddArrayScalarKernel, Int32Type)->Apply(RegressionSetArgs);
+BENCHMARK_TEMPLATE(AddArrayScalarKernel, Int16Type)->Apply(RegressionSetArgs);
+BENCHMARK_TEMPLATE(AddArrayScalarKernel, Int8Type)->Apply(RegressionSetArgs);
+BENCHMARK_TEMPLATE(AddArrayScalarKernel, UInt64Type)->Apply(RegressionSetArgs);
+BENCHMARK_TEMPLATE(AddArrayScalarKernel, UInt32Type)->Apply(RegressionSetArgs);
+BENCHMARK_TEMPLATE(AddArrayScalarKernel, UInt16Type)->Apply(RegressionSetArgs);
+BENCHMARK_TEMPLATE(AddArrayScalarKernel, UInt8Type)->Apply(RegressionSetArgs);
+BENCHMARK_TEMPLATE(AddArrayScalarKernel, FloatType)->Apply(RegressionSetArgs);
+BENCHMARK_TEMPLATE(AddArrayScalarKernel, DoubleType)->Apply(RegressionSetArgs);

Review comment:
       I don't think that having benchmarks across so many different null percentages for these kernels is useful because of how the validity bitmaps are handled (batch intersection). I'd suggest just having a "with nulls" and a "no nulls" benchmark (currently there isn't a no-nulls benchmark, so we don't know the overhead involved with propagating the nulls relative to computing the arithmetic output)




----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org