[GitHub] [arrow] zeroshade commented on a diff in pull request #34666: GH-32950: [Go] REE Benchmarks

["zeroshade (via GitHub)" <gi...@apache.org>]

zeroshade commented on code in PR #34666:
URL: https://github.com/apache/arrow/pull/34666#discussion_r1145116487


##########
go/arrow/compute/vector_run_end_test.go:
##########
@@ -295,3 +299,125 @@ func TestRunEndFunctions(t *testing.T) {
 		})
 	}
 }
+
+func benchRunEndEncode(b *testing.B, sz int, nullProb float64, runEndType, valueType arrow.DataType) {
+	b.Run("encode", func(b *testing.B) {
+		var (
+			mem = memory.NewCheckedAllocator(memory.DefaultAllocator)
+			rng = gen.NewRandomArrayGenerator(seed, mem)
+		)
+
+		values := rng.ArrayOf(valueType.ID(), int64(sz), nullProb)
+		b.Cleanup(func() {
+			values.Release()
+		})
+
+		var (
+			res   compute.Datum
+			err   error
+			ctx   = compute.WithAllocator(context.Background(), mem)
+			input = &compute.ArrayDatum{Value: values.Data()}
+			opts  = compute.RunEndEncodeOptions{RunEndType: runEndType}
+
+			byts int64
+		)
+
+		for _, buf := range values.Data().Buffers() {
+			if buf != nil {
+				byts += int64(buf.Len())
+			}
+		}
+
+		b.SetBytes(byts)
+		b.ResetTimer()
+		for n := 0; n < b.N; n++ {
+			res, err = compute.RunEndEncode(ctx, opts, input)

Review Comment:
   so right now it's using a specific seed to ensure consistency, but that's a good point that this benchmark is essentially still just benchmarking a random normal distribution of *values* with no guarantees to any particular distribution of runs and run-lengths. It might make more sense to  restrict the min/max values to better guarantee some runs.



-- 
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.

To unsubscribe, e-mail: github-unsubscribe@arrow.apache.org

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