lidavidm commented on a change in pull request #10520:
URL: https://github.com/apache/arrow/pull/10520#discussion_r651990126
##########
File path: cpp/src/arrow/compute/kernels/scalar_string_benchmark.cc
##########
@@ -169,6 +169,48 @@ static void BinaryJoinArrayArray(benchmark::State& state) {
});
}
+static void BinaryJoinElementWise(benchmark::State& state,
+ SeparatorFactory make_separator) {
+ // Unfortunately benchmark is not 1:1 with BinaryJoin since BinaryJoin can
join a
+ // varying number of inputs per output
+ const int64_t n_strings = 1000;
Review comment:
Ah that's fair. I've bumped it up to 65536 rows.
Old impl:
```
-----------------------------------------------------------------------------------------------
Benchmark Time CPU
Iterations UserCounters...
-----------------------------------------------------------------------------------------------
BinaryJoinArrayScalar 104542 ns 104540 ns
6671 bytes_per_second=1113.52M/s
BinaryJoinArrayArray 114749 ns 114750 ns
6062 bytes_per_second=1014.45M/s
BinaryJoinElementWiseArrayScalar/2 3017902 ns 3017894 ns
229 bytes_per_second=506.693M/s
BinaryJoinElementWiseArrayScalar/8 10470391 ns 10470204 ns
67 bytes_per_second=585.289M/s
BinaryJoinElementWiseArrayScalar/64 77328574 ns 77273626 ns
9 bytes_per_second=634.122M/s
BinaryJoinElementWiseArrayScalar/128 112622860 ns 112606581 ns
6 bytes_per_second=870.334M/s
BinaryJoinElementWiseArrayArray/2 3508926 ns 3508898 ns
200 bytes_per_second=435.791M/s
BinaryJoinElementWiseArrayArray/8 11349990 ns 11349827 ns
61 bytes_per_second=539.928M/s
BinaryJoinElementWiseArrayArray/64 76564083 ns 76563761 ns
9 bytes_per_second=640.001M/s
BinaryJoinElementWiseArrayArray/128 111560076 ns 111556809 ns
6 bytes_per_second=878.524M/s
```
Current impl:
```
-----------------------------------------------------------------------------------------------
Benchmark Time CPU
Iterations UserCounters...
-----------------------------------------------------------------------------------------------
BinaryJoinArrayScalar 107309 ns 107307 ns
6537 bytes_per_second=1084.81M/s
BinaryJoinArrayArray 117132 ns 117131 ns
5966 bytes_per_second=993.829M/s
BinaryJoinElementWiseArrayScalar/2 2624376 ns 2624345 ns
267 bytes_per_second=582.677M/s
BinaryJoinElementWiseArrayScalar/8 9394312 ns 9394273 ns
73 bytes_per_second=652.322M/s
BinaryJoinElementWiseArrayScalar/64 77934790 ns 77934514 ns
9 bytes_per_second=628.744M/s
BinaryJoinElementWiseArrayScalar/128 128550012 ns 128549432 ns
5 bytes_per_second=762.394M/s
BinaryJoinElementWiseArrayArray/2 3258818 ns 3258767 ns
214 bytes_per_second=469.24M/s
BinaryJoinElementWiseArrayArray/8 10201948 ns 10201752 ns
66 bytes_per_second=600.69M/s
BinaryJoinElementWiseArrayArray/64 79305591 ns 79303789 ns
8 bytes_per_second=617.888M/s
BinaryJoinElementWiseArrayArray/128 129524057 ns 129524481 ns
5 bytes_per_second=756.655M/s
```
Still about the same relative to each other.
--
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:
[email protected]
