bkietz commented on a change in pull request #9621:
URL: https://github.com/apache/arrow/pull/9621#discussion_r592445217
##########
File path: cpp/src/arrow/compute/kernels/aggregate_benchmark.cc
##########
@@ -300,6 +300,70 @@ BENCHMARK_TEMPLATE(ReferenceSum,
SumBitmapVectorizeUnroll<int64_t>)
->Apply(BenchmarkSetArgs);
#endif // ARROW_WITH_BENCHMARKS_REFERENCE
+//
+// GroupBy
+//
+
+static void BenchmarkGroupBy(benchmark::State& state,
+ std::vector<GroupByOptions::Aggregate> aggregates,
+ std::vector<Datum> aggregands, std::vector<Datum>
keys) {
+ std::vector<Datum> arguments;
+ for (const Datum& aggregand : aggregands) {
+ arguments.push_back(aggregand);
+ }
+
+ for (const Datum& key : keys) {
+ arguments.push_back(key);
+ }
+
+ GroupByOptions options;
+ options.aggregates = aggregates;
+ options.key_names.resize(keys.size(), "ignored");
+
+ for (auto _ : state) {
+ ABORT_NOT_OK(CallFunction("group_by", arguments, &options).status());
+ }
+}
+
+#define GROUP_BY_BENCHMARK(Name, Impl) \
+ static void Name(benchmark::State& state) { \
+ RegressionArgs args(state, false); \
+ auto rng = random::RandomArrayGenerator(1923); \
+ (Impl)(); \
+ } \
+ BENCHMARK(Name)->Apply([](benchmark::internal::Benchmark* bench) { \
+ BenchmarkSetArgsWithSizes(bench, {1 * 1024 * 1024}); \
+ })
+
+GROUP_BY_BENCHMARK(SumDoublesGroupedBySmallStringSet, [&] {
+ auto summand = rng.Float64(args.size,
+ /*min=*/0.0,
+ /*max=*/1.0e14,
+ /*null_probability=*/args.null_proportion,
+ /*nan_probability=*/args.null_proportion / 10);
+
+ auto key = rng.StringWithRepeats(args.size,
+ /*unique=*/16,
+ /*min_length=*/3,
+ /*max_length=*/32);
+
+ BenchmarkGroupBy(state, {{"sum", NULLPTR, "summed f64"}}, {summand}, {key});
Review comment:
Well it doesn't work. I'll remove the second string
----------------------------------------------------------------
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:
us...@infra.apache.org
