fix: fixed seed for micro-benchmarks (#5574)

Signed-off-by: Alexander Droste <[email protected]>

Author: 0ax1

bench-vortex/src/datasets/struct_list_of_ints.rs

@@ -5,6 +5,7 @@ use anyhow::Result;
 use async_trait::async_trait;
 use rand::Rng;
 use rand::SeedableRng;
+use rand::rngs::StdRng;
 use vortex::ArrayRef;
 use vortex::IntoArray;
 use vortex::arrays::ChunkedArray;
@@ -46,7 +47,7 @@ impl Dataset for StructListOfInts {
         let names: FieldNames = (0..self.num_columns)
             .map(|col_idx| col_idx.to_string())
             .collect();
-        let mut rng = rand::rngs::StdRng::seed_from_u64(0);
+        let mut rng = StdRng::seed_from_u64(0);
 
         let rows_per_chunk = (self.row_count / self.chunk_count).max(1usize);
         let chunks: Result<Vec<_>> = (0..self.row_count)

encodings/fastlanes/benches/bitpacking_take.rs

@@ -9,7 +9,6 @@ use rand::Rng;
 use rand::SeedableRng;
 use rand::distr::Uniform;
 use rand::prelude::StdRng;
-use rand::rng;
 use vortex_array::IntoArray as _;
 use vortex_array::arrays::PrimitiveArray;
 use vortex_array::compute::take;
@@ -100,7 +99,7 @@ fn take_200k_first_chunk_only(bencher: Bencher) {
 }
 
 fn fixture(len: usize, bits: usize) -> Buffer<u32> {
-    let rng = rng();
+    let rng = StdRng::seed_from_u64(0);
     let range = Uniform::new(0_u32, 2_u32.pow(bits as u32)).unwrap();
     rng.sample_iter(range).take(len).collect()
 }

vortex/benches/common_encoding_tree_throughput.rs

@@ -63,10 +63,12 @@ macro_rules! with_counter {
 // Encoding tree setup functions
 
 mod setup {
+    use rand::rngs::StdRng;
+
     use super::*;
 
     fn setup_primitive_arrays() -> (PrimitiveArray, PrimitiveArray, PrimitiveArray) {
-        let mut rng = rand::rngs::StdRng::seed_from_u64(0);
+        let mut rng = StdRng::seed_from_u64(0);
         let uint_array =
             PrimitiveArray::from_iter((0..NUM_VALUES).map(|_| rng.random_range(42u32..256)));
         let int_array = cast(uint_array.as_ref(), PType::I32.into())
@@ -113,7 +115,7 @@ mod setup {
     /// Create Dict <- VarBinView encoding tree for strings with BitPacked codes
     #[allow(clippy::cast_possible_truncation)]
     pub fn dict_varbinview_string() -> ArrayRef {
-        let mut rng = rand::rngs::StdRng::seed_from_u64(42);
+        let mut rng = StdRng::seed_from_u64(42);
 
         // Create unique values (0.005% uniqueness = 50 unique strings)
         let num_unique = ((NUM_VALUES as f64) * 0.00005) as usize;
@@ -147,7 +149,7 @@ mod setup {
     /// Create RunEnd <- FoR <- BitPacked encoding tree for u32
     #[allow(clippy::cast_possible_truncation)]
     pub fn runend_for_bp_u32() -> ArrayRef {
-        let mut rng = rand::rngs::StdRng::seed_from_u64(42);
+        let mut rng = StdRng::seed_from_u64(42);
         // Create data with runs of repeated values
         let mut values = Vec::with_capacity(NUM_VALUES as usize);
         let mut current_value = rng.random_range(0u32..100);
@@ -189,7 +191,7 @@ mod setup {
     /// Create Dict <- FSST <- VarBin encoding tree for strings
     #[allow(clippy::cast_possible_truncation)]
     pub fn dict_fsst_varbin_string() -> ArrayRef {
-        let mut rng = rand::rngs::StdRng::seed_from_u64(43);
+        let mut rng = StdRng::seed_from_u64(43);
 
         // Create unique values (1% uniqueness = 10,000 unique strings)
         let num_unique = ((NUM_VALUES as f64) * 0.01) as usize;
@@ -221,7 +223,7 @@ mod setup {
     /// Compress the VarBin offsets inside FSST with BitPacked
     #[allow(clippy::cast_possible_truncation)]
     pub fn dict_fsst_varbin_bp_string() -> ArrayRef {
-        let mut rng = rand::rngs::StdRng::seed_from_u64(45);
+        let mut rng = StdRng::seed_from_u64(45);
 
         // Create unique values (1% uniqueness = 10,000 unique strings)
         let num_unique = ((NUM_VALUES as f64) * 0.01) as usize;
@@ -276,7 +278,7 @@ mod setup {
     /// Create DateTimeParts <- FoR <- BitPacked encoding tree
     pub fn datetime_for_bp() -> ArrayRef {
         // Create timestamp data (microseconds since epoch)
-        let mut rng = rand::rngs::StdRng::seed_from_u64(123);
+        let mut rng = StdRng::seed_from_u64(123);
         let base_timestamp = 1_600_000_000_000_000i64; // Sept 2020 in microseconds
         let timestamps: Vec<i64> = (0..NUM_VALUES)
             .map(|_| base_timestamp + rng.random_range(0..86_400_000_000)) // Random times within a day

vortex/benches/pipeline.rs

@@ -117,6 +117,8 @@
 use divan::Bencher;
 use fastlanes::BitPacking;
 use rand::Rng;
+use rand::SeedableRng;
+use rand::rngs::StdRng;
 use vortex_alp::ALPFloat;
 use vortex_alp::Exponents;
 use vortex_error::vortex_panic;
@@ -256,7 +258,7 @@ fn setup(size: usize) -> (InputData, BenchmarkBuffers) {
 fn create_random_values(len: usize) -> Vec<f32> {
     assert!(len.is_multiple_of(N));
 
-    let mut rng = rand::rng();
+    let mut rng = StdRng::seed_from_u64(0);
     (0..len)
         .map(|_| rng.random_range(0..1024))
         .map(|x| x as f32 / 100.0)

vortex/benches/single_encoding_throughput.rs

@@ -11,6 +11,7 @@ use mimalloc::MiMalloc;
 use rand::Rng;
 use rand::SeedableRng;
 use rand::prelude::IndexedRandom;
+use rand::rngs::StdRng;
 use vortex::IntoArray;
 use vortex::ToCanonical;
 use vortex::arrays::PrimitiveArray;
@@ -55,7 +56,7 @@ macro_rules! with_counter {
 
 // Setup functions
 fn setup_primitive_arrays() -> (PrimitiveArray, PrimitiveArray, PrimitiveArray) {
-    let mut rng = rand::rngs::StdRng::seed_from_u64(0);
+    let mut rng = StdRng::seed_from_u64(0);
     let uint_array =
         PrimitiveArray::from_iter((0..NUM_VALUES).map(|_| rng.random_range(42u32..256)));
     let int_array = cast(uint_array.as_ref(), PType::I32.into())
@@ -69,7 +70,7 @@ fn setup_primitive_arrays() -> (PrimitiveArray, PrimitiveArray, PrimitiveArray)
 
 #[allow(clippy::cast_possible_truncation)]
 fn gen_varbin_words(len: usize, uniqueness: f64) -> Vec<String> {
-    let mut rng = rand::rng();
+    let mut rng = StdRng::seed_from_u64(0);
     let uniq_cnt = (len as f64 * uniqueness) as usize;
     let dict: Vec<String> = (0..uniq_cnt)
         .map(|_| {