chore: increase row count and batch size for more deterministic tests

Author: Weijun-H

arrow-json/benches/serde.rs

@@ -22,12 +22,15 @@ use rand::{Rng, rng};
 use serde::Serialize;
 use std::sync::Arc;
 
+const ROWS: usize = 1 << 18;
+
 #[allow(deprecated)]
 fn do_bench<R: Serialize>(c: &mut Criterion, name: &str, rows: &[R], schema: &Schema) {
     let schema = Arc::new(schema.clone());
+    let batch_size = rows.len();
     c.bench_function(name, |b| {
         b.iter(|| {
-            let builder = ReaderBuilder::new(schema.clone()).with_batch_size(64);
+            let builder = ReaderBuilder::new(schema.clone()).with_batch_size(batch_size);
             let mut decoder = builder.build_decoder().unwrap();
             decoder.serialize(rows)
         })
@@ -37,26 +40,26 @@ fn do_bench<R: Serialize>(c: &mut Criterion, name: &str, rows: &[R], schema: &Sc
 fn criterion_benchmark(c: &mut Criterion) {
     let mut rng = rng();
     let schema = Schema::new(vec![Field::new("i32", DataType::Int32, false)]);
-    let v: Vec<i32> = (0..2048).map(|_| rng.random_range(0..10000)).collect();
+    let v: Vec<i32> = (0..ROWS).map(|_| rng.random_range(0..10000)).collect();
 
     do_bench(c, "small_i32", &v, &schema);
-    let v: Vec<i32> = (0..2048).map(|_| rng.random()).collect();
+    let v: Vec<i32> = (0..ROWS).map(|_| rng.random()).collect();
     do_bench(c, "large_i32", &v, &schema);
 
     let schema = Schema::new(vec![Field::new("i64", DataType::Int64, false)]);
-    let v: Vec<i64> = (0..2048).map(|_| rng.random_range(0..10000)).collect();
+    let v: Vec<i64> = (0..ROWS).map(|_| rng.random_range(0..10000)).collect();
     do_bench(c, "small_i64", &v, &schema);
-    let v: Vec<i64> = (0..2048)
+    let v: Vec<i64> = (0..ROWS)
         .map(|_| rng.random_range(0..i32::MAX as _))
         .collect();
     do_bench(c, "medium_i64", &v, &schema);
-    let v: Vec<i64> = (0..2048).map(|_| rng.random()).collect();
+    let v: Vec<i64> = (0..ROWS).map(|_| rng.random()).collect();
     do_bench(c, "large_i64", &v, &schema);
 
     let schema = Schema::new(vec![Field::new("f32", DataType::Float32, false)]);
-    let v: Vec<f32> = (0..2048).map(|_| rng.random_range(0.0..10000.)).collect();
+    let v: Vec<f32> = (0..ROWS).map(|_| rng.random_range(0.0..10000.)).collect();
     do_bench(c, "small_f32", &v, &schema);
-    let v: Vec<f32> = (0..2048).map(|_| rng.random_range(0.0..f32::MAX)).collect();
+    let v: Vec<f32> = (0..ROWS).map(|_| rng.random_range(0.0..f32::MAX)).collect();
     do_bench(c, "large_f32", &v, &schema);
 }