varbinview zip bench

vortex-array/Cargo.toml

@@ -163,3 +163,7 @@ harness = false
 [[bench]]
 name = "dict_unreferenced_mask"
 harness = false
+
+[[bench]]
+name = "varbinview_zip"
+harness = false

vortex-array/benches/varbinview_zip.rs

@@ -0,0 +1,82 @@
+// SPDX-License-Identifier: Apache-2.0
+// SPDX-FileCopyrightText: Copyright the Vortex contributors
+
+#![allow(clippy::unwrap_used)]
+
+use divan::Bencher;
+use vortex_array::IntoArray;
+use vortex_array::arrays::VarBinViewArray;
+use vortex_array::compute::zip;
+use vortex_dtype::DType;
+use vortex_dtype::Nullability;
+use vortex_mask::Mask;
+
+fn main() {
+    divan::main();
+}
+
+/// Benchmarks zip on VarBinView arrays with a highly fragmented mask (worst case for per-slice lookup paths).
+#[divan::bench]
+fn varbinview_zip_fragmented_mask(bencher: Bencher) {
+    let len = 65_536;
+    let if_true = fixture(len).into_array();
+    let if_false = other_fixture(len).into_array();
+    let mask = alternating_mask(len);
+
+    bencher
+        .with_inputs(|| (&if_true, &if_false, &mask))
+        .bench_refs(|(t, f, m)| {
+            zip(t.as_ref(), f.as_ref(), m).unwrap();
+        });
+}
+
+/// Benchmarks zip on VarBinView arrays with blocky mask segments to contrast with the fragmented case.
+#[divan::bench]
+fn varbinview_zip_block_mask(bencher: Bencher) {
+    let len = 65_536;
+    let if_true = fixture(len).into_array();
+    let if_false = other_fixture(len).into_array();
+    let mask = block_mask(len, 128);
+
+    bencher
+        .with_inputs(|| (&if_true, &if_false, &mask))
+        .bench_refs(|(t, f, m)| {
+            zip(t.as_ref(), f.as_ref(), m).unwrap();
+        });
+}
+
+fn fixture(len: usize) -> VarBinViewArray {
+    VarBinViewArray::from_iter(
+        [
+            Some("short"),
+            None,
+            Some("this is a much longer string to force outlining"),
+        ]
+        .into_iter()
+        .cycle()
+        .take(len),
+        DType::Utf8(Nullability::Nullable),
+    )
+}
+
+fn other_fixture(len: usize) -> VarBinViewArray {
+    VarBinViewArray::from_iter(
+        [
+            Some("different"),
+            Some("another longer string that will be outlined as well"),
+            None,
+        ]
+        .into_iter()
+        .cycle()
+        .take(len),
+        DType::Utf8(Nullability::Nullable),
+    )
+}
+
+fn alternating_mask(len: usize) -> Mask {
+    Mask::from_iter((0..len).map(|i| i % 2 == 0))
+}
+
+fn block_mask(len: usize, block: usize) -> Mask {
+    Mask::from_iter((0..len).map(|i| (i / block) % 2 == 0))
+}