pull upstream main, fix conflicts

Cargo.toml

@@ -25,6 +25,7 @@ members = [
     "monty-31",
     "poseidon",
     "poseidon2",
+    "poseidon2-air",
     "rescue",
     "sha256",
     "symmetric",

baby-bear/Cargo.toml

@@ -13,12 +13,12 @@ p3-mds = { path = "../mds" }
 p3-monty-31 = { path = "../monty-31" }
 p3-poseidon2 = { path = "../poseidon2" }
 p3-symmetric = { path = "../symmetric" }
-num-bigint = { version = "0.4.3", default-features = false }
 rand = "0.8.5"
 serde = { version = "1.0", default-features = false, features = ["derive"] }
 
 [dev-dependencies]
 p3-field-testing = { path = "../field-testing" }
+p3-dft = { path = "../dft" }
 rand = { version = "0.8.5", features = ["min_const_gen"] }
 criterion = "0.5.1"
 rand_chacha = "0.3.1"

baby-bear/benches/bench_field.rs

@@ -1,9 +1,12 @@
+use std::any::type_name;
+
 use criterion::{criterion_group, criterion_main, BatchSize, Criterion};
 use p3_baby_bear::BabyBear;
-use p3_field::AbstractField;
+use p3_field::{AbstractField, Field};
 use p3_field_testing::bench_func::{
     benchmark_add_latency, benchmark_add_throughput, benchmark_inv, benchmark_iter_sum,
-    benchmark_sub_latency, benchmark_sub_throughput,
+    benchmark_mul_latency, benchmark_mul_throughput, benchmark_sub_latency,
+    benchmark_sub_throughput,
 };
 
 type F = BabyBear;
@@ -33,5 +36,20 @@ fn bench_field(c: &mut Criterion) {
     });
 }
 
-criterion_group!(baby_bear_arithmetic, bench_field);
+fn bench_packedfield(c: &mut Criterion) {
+    let name = type_name::<<F as Field>::Packing>().to_string();
+    // Note that each round of throughput has 10 operations
+    // So we should have 10 * more repetitions for latency tests.
+    const REPS: usize = 100;
+    const L_REPS: usize = 10 * REPS;
+
+    benchmark_add_latency::<<F as Field>::Packing, L_REPS>(c, &name);
+    benchmark_add_throughput::<<F as Field>::Packing, REPS>(c, &name);
+    benchmark_sub_latency::<<F as Field>::Packing, L_REPS>(c, &name);
+    benchmark_sub_throughput::<<F as Field>::Packing, REPS>(c, &name);
+    benchmark_mul_latency::<<F as Field>::Packing, L_REPS>(c, &name);
+    benchmark_mul_throughput::<<F as Field>::Packing, REPS>(c, &name);
+}
+
+criterion_group!(baby_bear_arithmetic, bench_field, bench_packedfield);
 criterion_main!(baby_bear_arithmetic);

baby-bear/benches/extension.rs

@@ -1,23 +1,32 @@
 use criterion::{criterion_group, criterion_main, Criterion};
 use p3_baby_bear::BabyBear;
 use p3_field::extension::BinomialExtensionField;
-use p3_field_testing::bench_func::{benchmark_inv, benchmark_mul, benchmark_square};
+use p3_field_testing::bench_func::{
+    benchmark_inv, benchmark_mul_latency, benchmark_mul_throughput, benchmark_square,
+};
 
 type EF4 = BinomialExtensionField<BabyBear, 4>;
 type EF5 = BinomialExtensionField<BabyBear, 5>;
 
+// Note that each round of throughput has 10 operations
+// So we should have 10 * more repetitions for latency tests.
+const REPS: usize = 100;
+const L_REPS: usize = 10 * REPS;
+
 fn bench_quartic_extension(c: &mut Criterion) {
     let name = "BinomialExtensionField<BabyBear, 4>";
     benchmark_square::<EF4>(c, name);
     benchmark_inv::<EF4>(c, name);
-    benchmark_mul::<EF4>(c, name);
+    benchmark_mul_throughput::<EF4, REPS>(c, name);
+    benchmark_mul_latency::<EF4, L_REPS>(c, name);
 }
 
 fn bench_qunitic_extension(c: &mut Criterion) {
     let name = "BinomialExtensionField<BabyBear, 5>";
     benchmark_square::<EF5>(c, name);
     benchmark_inv::<EF5>(c, name);
-    benchmark_mul::<EF5>(c, name);
+    benchmark_mul_throughput::<EF5, REPS>(c, name);
+    benchmark_mul_latency::<EF5, L_REPS>(c, name);
 }
 
 criterion_group!(

baby-bear/src/aarch64_neon/packing.rs

@@ -17,12 +17,10 @@ pub type PackedBabyBearNeon = PackedMontyField31Neon<BabyBearParameters>;
 
 #[cfg(test)]
 mod tests {
-    use p3_field::AbstractField;
     use p3_field_testing::test_packed_field;
-    use p3_monty_31::PackedMontyField31Neon;
 
     use super::WIDTH;
-    use crate::{BabyBear, BabyBearParameters};
+    use crate::BabyBear;
 
     const SPECIAL_VALS: [BabyBear; WIDTH] =
         BabyBear::new_array([0x00000000, 0x00000001, 0x00000002, 0x78000000]);
@@ -32,38 +30,4 @@ mod tests {
         crate::PackedBabyBearNeon::zero(),
         p3_monty_31::PackedMontyField31Neon::<crate::BabyBearParameters>(super::SPECIAL_VALS)
     );
-
-    #[test]
-    fn test_cube_vs_mul() {
-        let vec = PackedMontyField31Neon::<BabyBearParameters>(BabyBear::new_array([
-            0x4efd5eaf, 0x311b8e0c, 0x74dd27c1, 0x449613f0,
-        ]));
-        let res0 = vec * vec.square();
-        let res1 = vec.cube();
-        assert_eq!(res0, res1);
-    }
-
-    #[test]
-    fn test_cube_vs_scalar() {
-        let arr = BabyBear::new_array([0x57155037, 0x71bdcc8e, 0x301f94d, 0x435938a6]);
-
-        let vec = PackedMontyField31Neon::<BabyBearParameters>(arr);
-        let vec_res = vec.cube();
-
-        #[allow(clippy::needless_range_loop)]
-        for i in 0..WIDTH {
-            assert_eq!(vec_res.0[i], arr[i].cube());
-        }
-    }
-
-    #[test]
-    fn test_cube_vs_scalar_special_vals() {
-        let vec = PackedMontyField31Neon::<BabyBearParameters>(SPECIAL_VALS);
-        let vec_res = vec.cube();
-
-        #[allow(clippy::needless_range_loop)]
-        for i in 0..WIDTH {
-            assert_eq!(vec_res.0[i], SPECIAL_VALS[i].cube());
-        }
-    }
 }

baby-bear/src/baby_bear.rs

@@ -66,14 +66,24 @@ impl FieldParameters for BabyBearParameters {
 impl TwoAdicData for BabyBearParameters {
     const TWO_ADICITY: usize = 27;
 
-    type ArrayLike = [BabyBear; Self::TWO_ADICITY + 1];
+    type ArrayLike = &'static [BabyBear];
 
-    const TWO_ADIC_GENERATORS: Self::ArrayLike = BabyBear::new_array([
+    const TWO_ADIC_GENERATORS: Self::ArrayLike = &BabyBear::new_array([
         0x1, 0x78000000, 0x67055c21, 0x5ee99486, 0xbb4c4e4, 0x2d4cc4da, 0x669d6090, 0x17b56c64,
         0x67456167, 0x688442f9, 0x145e952d, 0x4fe61226, 0x4c734715, 0x11c33e2a, 0x62c3d2b1,
         0x77cad399, 0x54c131f4, 0x4cabd6a6, 0x5cf5713f, 0x3e9430e8, 0xba067a3, 0x18adc27d,
         0x21fd55bc, 0x4b859b3d, 0x3bd57996, 0x4483d85a, 0x3a26eef8, 0x1a427a41,
     ]);
+
+    const ROOTS_8: Self::ArrayLike = &BabyBear::new_array([0x5ee99486, 0x67055c21, 0xc9ea3ba]);
+    const INV_ROOTS_8: Self::ArrayLike = &BabyBear::new_array([0x6b615c47, 0x10faa3e0, 0x19166b7b]);
+
+    const ROOTS_16: Self::ArrayLike = &BabyBear::new_array([
+        0xbb4c4e4, 0x5ee99486, 0x4b49e08, 0x67055c21, 0x5376917a, 0xc9ea3ba, 0x563112a7,
+    ]);
+    const INV_ROOTS_16: Self::ArrayLike = &BabyBear::new_array([
+        0x21ceed5a, 0x6b615c47, 0x24896e87, 0x10faa3e0, 0x734b61f9, 0x19166b7b, 0x6c4b3b1d,
+    ]);
 }
 
 impl BinomialExtensionData<4> for BabyBearParameters {
@@ -101,8 +111,8 @@ impl BinomialExtensionData<5> for BabyBearParameters {
 mod tests {
     use core::array;
 
-    use p3_field::{AbstractField, Field, PrimeField32, PrimeField64, TwoAdicField};
-    use p3_field_testing::{test_field, test_two_adic_field};
+    use p3_field::{PrimeField32, PrimeField64, TwoAdicField};
+    use p3_field_testing::{test_field, test_field_dft, test_two_adic_field};
 
     use super::*;
 
@@ -215,4 +225,13 @@ mod tests {
 
     test_field!(crate::BabyBear);
     test_two_adic_field!(crate::BabyBear);
+
+    test_field_dft!(radix2dit, crate::BabyBear, p3_dft::Radix2Dit<_>);
+    test_field_dft!(bowers, crate::BabyBear, p3_dft::Radix2Bowers);
+    test_field_dft!(parallel, crate::BabyBear, p3_dft::Radix2DitParallel);
+    test_field_dft!(
+        recur_dft,
+        crate::BabyBear,
+        p3_monty_31::dft::RecursiveDft<_>
+    );
 }

bn254-fr/Cargo.toml

@@ -13,7 +13,7 @@ ff = { version = "0.13", features = ["derive", "derive_bits"] }
 num-bigint = { version = "0.4.3", default-features = false }
 rand = "0.8.5"
 serde = { version = "1.0", default-features = false, features = ["derive"] }
-halo2curves = { version = "0.6.1", features = ["bits", "derive_serde"] }
+halo2curves = { version = "0.7.0", features = ["bits", "derive_serde"] }
 
 [dev-dependencies]
 p3-field-testing = { path = "../field-testing" }

circle/Cargo.toml

@@ -24,7 +24,6 @@ p3-keccak = { path = "../keccak" }
 p3-mds = { path = "../mds" }
 p3-mersenne-31 = { path = "../mersenne-31" }
 p3-merkle-tree = { path = "../merkle-tree" }
-p3-poseidon = { path = "../poseidon" }
 p3-symmetric = { path = "../symmetric" }
 
 hashbrown = "0.14.3"

circle/benches/cfft.rs

@@ -1,5 +1,3 @@
-use std::any::type_name;
-
 use criterion::measurement::Measurement;
 use criterion::{criterion_group, criterion_main, BenchmarkGroup, BenchmarkId, Criterion};
 use p3_baby_bear::BabyBear;
@@ -8,18 +6,10 @@ use p3_dft::{Radix2Bowers, Radix2Dit, Radix2DitParallel, TwoAdicSubgroupDft};
 use p3_field::TwoAdicField;
 use p3_matrix::dense::RowMajorMatrix;
 use p3_mersenne_31::Mersenne31;
+use p3_util::pretty_name;
 use rand::distributions::{Distribution, Standard};
 use rand::thread_rng;
 
-fn pretty_name<T>() -> String {
-    let name = type_name::<T>();
-    let mut result = String::new();
-    for qual in name.split_inclusive(&['<', '>', ',']) {
-        result.push_str(qual.split("::").last().unwrap());
-    }
-    result
-}
-
 fn bench_lde(c: &mut Criterion) {
     let log_n = 18;
     let log_w = 8;

circle/src/cfft.rs

@@ -273,7 +273,6 @@ pub fn circle_basis<F: Field>(p: Point<F>, log_n: usize) -> Vec<F> {
 mod tests {
     use itertools::iproduct;
     use p3_field::extension::BinomialExtensionField;
-    use p3_matrix::dense::RowMajorMatrix;
     use p3_mersenne_31::Mersenne31;
     use rand::{random, thread_rng};
 

circle/src/folding.rs

@@ -132,7 +132,6 @@ mod tests {
     use itertools::iproduct;
     use p3_field::extension::BinomialExtensionField;
     use p3_matrix::dense::RowMajorMatrix;
-    use p3_matrix::Matrix;
     use p3_mersenne_31::Mersenne31;
     use rand::{random, thread_rng};
 

circle/src/pcs.rs

@@ -15,7 +15,7 @@ use p3_matrix::{Dimensions, Matrix};
 use p3_maybe_rayon::prelude::*;
 use p3_util::log2_strict_usize;
 use serde::{Deserialize, Serialize};
-use tracing::{info_span, instrument};
+use tracing::info_span;
 
 use crate::deep_quotient::{deep_quotient_reduce_row, extract_lambda};
 use crate::domain::CircleDomain;
@@ -131,7 +131,6 @@ where
         }
     }
 
-    #[instrument(skip_all)]
     fn open(
         &self,
         // For each round,
@@ -468,7 +467,7 @@ where
 #[cfg(test)]
 mod tests {
     use p3_challenger::{HashChallenger, SerializingChallenger32};
-    use p3_commit::{ExtensionMmcs, Pcs};
+    use p3_commit::ExtensionMmcs;
     use p3_field::extension::BinomialExtensionField;
     use p3_keccak::Keccak256Hash;
     use p3_merkle_tree::FieldMerkleTreeMmcs;

commit/Cargo.toml

@@ -20,6 +20,5 @@ serde = { version = "1.0", default-features = false }
 p3-dft = { path = "../dft", optional = true }
 
 [dev-dependencies]
-p3-baby-bear = { path = "../baby-bear" }
 p3-dft = { path = "../dft" }
 rand = "0.8.5"

dft/Cargo.toml

@@ -10,8 +10,10 @@ p3-matrix = { path = "../matrix" }
 p3-maybe-rayon = { path = "../maybe-rayon" }
 p3-util = { path = "../util" }
 tracing = "0.1.37"
+itertools = "0.13.0"
 
 [dev-dependencies]
+p3-monty-31 = { path = "../monty-31" }
 p3-baby-bear = { path = "../baby-bear" }
 p3-goldilocks = { path = "../goldilocks" }
 p3-mersenne-31 = { path = "../mersenne-31" }