chore: add math benches

native/spark-expr/Cargo.toml

@@ -225,4 +225,16 @@ harness = false
 
 [[bench]]
 name = "cast_int_to_decimal"
+harness = false
+
+[[bench]]
+name = "log"
+harness = false
+
+[[bench]]
+name = "pow"
+harness = false
+
+[[bench]]
+name = "abs"
 harness = false

native/spark-expr/benches/abs.rs

@@ -0,0 +1,51 @@
+// 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.
+
+use criterion::{criterion_group, criterion_main, BenchmarkId, Criterion};
+use datafusion::physical_plan::ColumnarValue;
+use datafusion_comet_spark_expr::abs;
+use std::hint::black_box;
+
+#[path = "common/mod.rs"]
+mod common;
+use common::{i64_array, NULL_RATIOS, ROW_COUNTS};
+
+fn criterion_benchmark(c: &mut Criterion) {
+    let mut group = c.benchmark_group("abs");
+    for rows in ROW_COUNTS {
+        for (null_ratio, tag) in NULL_RATIOS {
+            let arr = i64_array(rows, null_ratio, |i| {
+                let v = (i as i64) % 1000;
+                if i % 2 == 0 {
+                    -v
+                } else {
+                    v
+                }
+            });
+            let args = vec![ColumnarValue::Array(arr)];
+            group.bench_with_input(
+                BenchmarkId::from_parameter(format!("{rows}/{tag}")),
+                &args,
+                |b, args| b.iter(|| black_box(abs(black_box(args)).unwrap())),
+            );
+        }
+    }
+    group.finish();
+}
+
+criterion_group!(benches, criterion_benchmark);
+criterion_main!(benches);

native/spark-expr/benches/common/mod.rs

@@ -18,11 +18,53 @@
 //! Helpers shared by the cast-from-string benchmarks, pulled in with
 //! `#[path = "common/mod.rs"] mod common;`. This lives in a subdirectory so that Cargo's bench
 //! auto-discovery, which only looks at `benches/*.rs`, does not treat it as a bench target.
+#![allow(dead_code)]
 
-use arrow::array::{builder::StringBuilder, RecordBatch};
+use arrow::array::{builder::StringBuilder, ArrayRef, Float64Array, Int64Array, RecordBatch};
 use arrow::datatypes::{DataType, Field, Schema};
 use std::sync::Arc;
 
+pub const ROW_COUNTS: [usize; 3] = [8_192, 65_536, 524_288];
+
+pub const NULL_RATIOS: [(f64, &str); 3] = [(0.0, "no_nulls"), (0.1, "sparse"), (1.0, "all_null")];
+
+fn is_null(i: usize, null_ratio: f64) -> bool {
+    if null_ratio <= 0.0 {
+        false
+    } else if null_ratio >= 1.0 {
+        true
+    } else {
+        let stride = (1.0 / null_ratio).round() as usize;
+        stride != 0 && i.is_multiple_of(stride)
+    }
+}
+
+pub fn f64_array(rows: usize, null_ratio: f64, value: impl Fn(usize) -> f64) -> ArrayRef {
+    let arr: Float64Array = (0..rows)
+        .map(|i| {
+            if is_null(i, null_ratio) {
+                None
+            } else {
+                Some(value(i))
+            }
+        })
+        .collect();
+    Arc::new(arr)
+}
+
+pub fn i64_array(rows: usize, null_ratio: f64, value: impl Fn(usize) -> i64) -> ArrayRef {
+    let arr: Int64Array = (0..rows)
+        .map(|i| {
+            if is_null(i, null_ratio) {
+                None
+            } else {
+                Some(value(i))
+            }
+        })
+        .collect();
+    Arc::new(arr)
+}
+
 /// A single-column `Utf8` batch of `rows` rows, where row `i` holds `value(i)` unless
 /// `i % null_modulus == 0`, in which case it is null.
 pub fn string_batch(

native/spark-expr/benches/log.rs

@@ -0,0 +1,49 @@
+// 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.
+
+use criterion::{criterion_group, criterion_main, BenchmarkId, Criterion};
+use datafusion::physical_plan::ColumnarValue;
+use datafusion_comet_spark_expr::spark_log;
+use std::hint::black_box;
+use std::sync::Arc;
+
+#[path = "common/mod.rs"]
+mod common;
+use common::{f64_array, NULL_RATIOS, ROW_COUNTS};
+
+fn criterion_benchmark(c: &mut Criterion) {
+    let mut group = c.benchmark_group("spark_log");
+    for rows in ROW_COUNTS {
+        let base = f64_array(rows, 0.0, |_| 10.0);
+        for (null_ratio, tag) in NULL_RATIOS {
+            let value = f64_array(rows, null_ratio, |i| (i % 1000) as f64 + 1.0);
+            let args = vec![
+                ColumnarValue::Array(Arc::clone(&base)),
+                ColumnarValue::Array(value),
+            ];
+            group.bench_with_input(
+                BenchmarkId::from_parameter(format!("{rows}/{tag}")),
+                &args,
+                |b, args| b.iter(|| black_box(spark_log(black_box(args)).unwrap())),
+            );
+        }
+    }
+    group.finish();
+}
+
+criterion_group!(benches, criterion_benchmark);
+criterion_main!(benches);

native/spark-expr/benches/pow.rs

@@ -0,0 +1,49 @@
+// 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.
+
+use criterion::{criterion_group, criterion_main, BenchmarkId, Criterion};
+use datafusion::physical_plan::ColumnarValue;
+use datafusion_comet_spark_expr::spark_pow;
+use std::hint::black_box;
+use std::sync::Arc;
+
+#[path = "common/mod.rs"]
+mod common;
+use common::{f64_array, NULL_RATIOS, ROW_COUNTS};
+
+fn criterion_benchmark(c: &mut Criterion) {
+    let mut group = c.benchmark_group("spark_pow");
+    for rows in ROW_COUNTS {
+        let base = f64_array(rows, 0.0, |_| 2.0);
+        for (null_ratio, tag) in NULL_RATIOS {
+            let exp = f64_array(rows, null_ratio, |i| (i % 10) as f64 + 1.0);
+            let args = vec![
+                ColumnarValue::Array(Arc::clone(&base)),
+                ColumnarValue::Array(exp),
+            ];
+            group.bench_with_input(
+                BenchmarkId::from_parameter(format!("{rows}/{tag}")),
+                &args,
+                |b, args| b.iter(|| black_box(spark_pow(black_box(args)).unwrap())),
+            );
+        }
+    }
+    group.finish();
+}
+
+criterion_group!(benches, criterion_benchmark);
+criterion_main!(benches);

native/spark-expr/src/lib.rs

@@ -84,11 +84,11 @@ pub use error::{decimal_overflow_error, SparkError, SparkErrorWithContext, Spark
 pub use hash_funcs::*;
 pub use json_funcs::{FromJson, ToJson};
 pub use math_funcs::{
-    checked_add, checked_div, checked_mul, checked_sub, create_modulo_expr, create_negate_expr,
-    spark_ceil, spark_decimal_div, spark_decimal_integral_div, spark_floor, spark_log,
-    spark_make_decimal, spark_pow, spark_round, spark_unhex, spark_unscaled_value, CheckOverflow,
-    DecimalRescaleCheckOverflow, NegativeExpr, NormalizeNaNAndZero, WideDecimalBinaryExpr,
-    WideDecimalOp,
+    abs, checked_add, checked_div, checked_mul, checked_sub, create_modulo_expr,
+    create_negate_expr, spark_ceil, spark_decimal_div, spark_decimal_integral_div, spark_floor,
+    spark_log, spark_make_decimal, spark_pow, spark_round, spark_unhex, spark_unscaled_value,
+    CheckOverflow, DecimalRescaleCheckOverflow, NegativeExpr, NormalizeNaNAndZero,
+    WideDecimalBinaryExpr, WideDecimalOp,
 };
 pub use query_context::{create_query_context_map, QueryContext, QueryContextMap};
 pub use string_funcs::*;

native/spark-expr/src/math_funcs/mod.rs

@@ -30,6 +30,7 @@ pub(crate) mod unhex;
 mod utils;
 mod wide_decimal_binary_expr;
 
+pub use abs::abs;
 pub use ceil::spark_ceil;
 pub use checked_arithmetic::{checked_add, checked_div, checked_mul, checked_sub};
 pub use div::spark_decimal_div;