FEAT(stark252 field): Adds Stark252 curve

.github/workflows/rust.yml

@@ -60,10 +60,10 @@ jobs:
       if: needs.check-changed-files.outputs.rust == 'true' || needs.check-changed-files.outputs.cpp_cuda == 'true'
       # Running tests from the root workspace will run all workspace members' tests by default
       # We need to limit the number of threads to avoid running out of memory on weaker machines
-      # ignored tests are polynomial tests. Since they conflict with NTT tests, they are executed sperately
+      # ignored tests are polynomial tests. Since they conflict with NTT tests, they are executed separately
       run: |
-        cargo test --workspace --exclude icicle-babybear --release --verbose --features=g2 -- --test-threads=2 --ignored
-        cargo test --workspace --exclude icicle-babybear --release --verbose --features=g2 -- --test-threads=2
+        cargo test --workspace --exclude icicle-babybear icicle-stark252 --release --verbose --features=g2 -- --test-threads=2 --ignored
+        cargo test --workspace --exclude icicle-babybear icicle-stark252 --release --verbose --features=g2 -- --test-threads=2
 
     - name: Run baby bear tests
       working-directory: ./wrappers/rust/icicle-fields/icicle-babybear
@@ -72,6 +72,13 @@ jobs:
         cargo test --release --verbose -- --ignored
         cargo test --release --verbose
 
+    - name: Run stark252 tests
+      working-directory: ./wrappers/rust/icicle-fields/icicle-stark252
+      if: needs.check-changed-files.outputs.rust == 'true' || needs.check-changed-files.outputs.cpp_cuda == 'true'
+      run: |
+        cargo test --release --verbose -- --ignored
+        cargo test --release --verbose
+
   build-windows:
     name: Build on Windows
     runs-on: windows-2022

icicle/cmake/FieldsCommon.cmake

@@ -1,5 +1,5 @@
 function(check_field)
-  set(SUPPORTED_FIELDS babybear)
+  set(SUPPORTED_FIELDS babybear;stark252)
 
   set(IS_FIELD_SUPPORTED FALSE)
   set(I 1000)
@@ -14,4 +14,4 @@ function(check_field)
   if (NOT IS_FIELD_SUPPORTED)
     message( FATAL_ERROR "The value of FIELD variable: ${FIELD} is not one of the supported fields: ${SUPPORTED_FIELDS}" )
   endif ()
-endfunction()
+endfunction()

icicle/include/api/stark252.h

@@ -0,0 +1,47 @@
+// WARNING: This file is auto-generated by a script.
+// Any changes made to this file may be overwritten.
+// Please modify the code generation script instead.
+// Path to the code generation script: scripts/gen_c_api.py
+
+#pragma once
+#ifndef STARK252_API_H
+#define STARK252_API_H
+
+#include <cuda_runtime.h>
+#include "gpu-utils/device_context.cuh"
+#include "fields/stark_fields/stark252.cuh"
+#include "ntt/ntt.cuh"
+#include "vec_ops/vec_ops.cuh"
+
+extern "C" cudaError_t stark252_mul_cuda(
+  stark252::scalar_t* vec_a, stark252::scalar_t* vec_b, int n, vec_ops::VecOpsConfig& config, stark252::scalar_t* result);
+
+extern "C" cudaError_t stark252_add_cuda(
+  stark252::scalar_t* vec_a, stark252::scalar_t* vec_b, int n, vec_ops::VecOpsConfig& config, stark252::scalar_t* result);
+
+extern "C" cudaError_t stark252_sub_cuda(
+  stark252::scalar_t* vec_a, stark252::scalar_t* vec_b, int n, vec_ops::VecOpsConfig& config, stark252::scalar_t* result);
+
+extern "C" cudaError_t stark252_transpose_matrix_cuda(
+  const stark252::scalar_t* input,
+  uint32_t row_size,
+  uint32_t column_size,
+  stark252::scalar_t* output,
+  device_context::DeviceContext& ctx,
+  bool on_device,
+  bool is_async);
+
+extern "C" void stark252_generate_scalars(stark252::scalar_t* scalars, int size);
+
+extern "C" cudaError_t stark252_scalar_convert_montgomery(
+  stark252::scalar_t* d_inout, size_t n, bool is_into, device_context::DeviceContext& ctx);
+
+extern "C" cudaError_t stark252_initialize_domain(
+  stark252::scalar_t* primitive_root, device_context::DeviceContext& ctx, bool fast_twiddles_mode);
+
+extern "C" cudaError_t stark252_ntt_cuda(
+  const stark252::scalar_t* input, int size, ntt::NTTDir dir, ntt::NTTConfig<stark252::scalar_t>& config, stark252::scalar_t* output);
+
+extern "C" cudaError_t stark252_release_domain(device_context::DeviceContext& ctx);
+
+#endif

icicle/include/fields/field_config.cuh

@@ -30,6 +30,9 @@ namespace field_config = grumpkin;
 #elif FIELD_ID == BABY_BEAR
 #include "fields/stark_fields/babybear.cuh"
 namespace field_config = babybear;
+#elif FIELD_ID == STARK_252
+#include "fields/stark_fields/stark252.cuh"
+namespace field_config = stark252;
 #endif
 
 #endif

icicle/include/fields/id.h

@@ -9,5 +9,6 @@
 #define GRUMPKIN  5
 
 #define BABY_BEAR 1001
+#define STARK_252 1002
 
 #endif

scripts/gen_c_api.py

@@ -53,6 +53,12 @@
 FIELDS_CONFIG = {
     "babybear": {
         "poseidon.h",
+    },
+    "stark252": {
+        "poseidon.h",
+        "field_ext.h",
+        "vec_ops_ext.h",
+        "ntt_ext.h",
     }
 }
 

wrappers/rust/Cargo.toml

@@ -9,6 +9,7 @@ members = [
   "icicle-curves/icicle-bn254",
   "icicle-curves/icicle-grumpkin",
   "icicle-fields/icicle-babybear",
+  "icicle-fields/icicle-stark252",
   "icicle-hash",
 ]
 exclude = [

wrappers/rust/icicle-fields/icicle-stark252/Cargo.toml

@@ -0,0 +1,22 @@
+[package]
+name = "icicle-stark252"
+version.workspace = true
+edition.workspace = true
+authors.workspace = true
+description = "Rust wrapper for the CUDA implementation of stark252 prime field by Ingonyama"
+homepage.workspace = true
+repository.workspace = true
+
+[dependencies]
+icicle-core = { workspace = true }
+icicle-cuda-runtime = { workspace = true }
+
+[build-dependencies]
+cmake = "0.1.50"
+
+[dev-dependencies]
+lambdaworks-math = "0.6.0"
+
+[features]
+default = []
+devmode = ["icicle-core/devmode"]

wrappers/rust/icicle-fields/icicle-stark252/build.rs

@@ -0,0 +1,24 @@
+use cmake::Config;
+
+fn main() {
+    println!("cargo:rerun-if-env-changed=CXXFLAGS");
+    println!("cargo:rerun-if-changed=../../../../icicle");
+
+    // Base config
+    let mut config = Config::new("../../../../icicle/");
+    config
+        .define("FIELD", "stark252")
+        .define("CMAKE_BUILD_TYPE", "Release")
+        .define("EXT_FIELD", "OFF");
+
+    // Build
+    let out_dir = config
+        .build_target("icicle_field")
+        .build();
+
+    println!("cargo:rustc-link-search={}/build/lib", out_dir.display());
+
+    println!("cargo:rustc-link-lib=ingo_field_stark252");
+    println!("cargo:rustc-link-lib=stdc++");
+    println!("cargo:rustc-link-lib=cudart");
+}

wrappers/rust/icicle-fields/icicle-stark252/src/field.rs

@@ -0,0 +1,19 @@
+use icicle_core::field::{Field, MontgomeryConvertibleField};
+use icicle_core::traits::{FieldConfig, FieldImpl, GenerateRandom};
+use icicle_core::{impl_field, impl_scalar_field};
+use icicle_cuda_runtime::device::check_device;
+use icicle_cuda_runtime::device_context::DeviceContext;
+use icicle_cuda_runtime::error::CudaError;
+use icicle_cuda_runtime::memory::{DeviceSlice, HostOrDeviceSlice};
+
+pub(crate) const SCALAR_LIMBS: usize = 8;
+
+impl_scalar_field!("stark252", stark252, SCALAR_LIMBS, ScalarField, ScalarCfg, Fr);
+#[cfg(test)]
+mod tests {
+    use super::ScalarField;
+    use icicle_core::impl_field_tests;
+    use icicle_core::tests::*;
+
+    impl_field_tests!(ScalarField);
+}

wrappers/rust/icicle-fields/icicle-stark252/src/lib.rs

@@ -0,0 +1,4 @@
+pub mod field;
+pub mod ntt;
+pub mod polynomials;
+pub mod vec_ops;

wrappers/rust/icicle-fields/icicle-stark252/src/ntt/mod.rs

@@ -0,0 +1,60 @@
+use crate::field::{ScalarCfg, ScalarField};
+
+use icicle_core::error::IcicleResult;
+use icicle_core::ntt::{NTTConfig, NTTDir, NTTDomain, NTT};
+use icicle_core::traits::IcicleResultWrap;
+use icicle_core::{impl_ntt, impl_ntt_without_domain};
+use icicle_cuda_runtime::device_context::DeviceContext;
+use icicle_cuda_runtime::error::CudaError;
+use icicle_cuda_runtime::memory::HostOrDeviceSlice;
+
+impl_ntt!("stark252", stark252, ScalarField, ScalarCfg);
+
+#[cfg(test)]
+pub(crate) mod tests {
+    use super::ScalarField;
+    use icicle_core::{
+        ntt::{initialize_domain, ntt_inplace, NTTConfig, NTTDir},
+        traits::{FieldImpl, GenerateRandom},
+    };
+    use icicle_cuda_runtime::{device_context::DeviceContext, memory::HostSlice};
+    use lambdaworks_math::{
+        field::{
+            element::FieldElement, fields::fft_friendly::stark_252_prime_field::Stark252PrimeField, traits::IsFFTField,
+        },
+        polynomial::Polynomial,
+        traits::ByteConversion,
+    };
+
+    pub type FE = FieldElement<Stark252PrimeField>;
+
+    #[test]
+    fn test_against_lambdaworks() {
+        let log_sizes = [15, 20];
+        let ctx = DeviceContext::default();
+        let lw_root_of_unity = Stark252PrimeField::get_primitive_root_of_unity(log_sizes[log_sizes.len() - 1]).unwrap();
+        initialize_domain(ScalarField::from_bytes_le(&lw_root_of_unity.to_bytes_le()), &ctx, false).unwrap();
+        for log_size in log_sizes {
+            let ntt_size = 1 << log_size;
+
+            let mut scalars: Vec<ScalarField> = <ScalarField as FieldImpl>::Config::generate_random(ntt_size);
+            let scalars_lw: Vec<FE> = scalars
+                .iter()
+                .map(|x| FieldElement::from_bytes_le(&x.to_bytes_le()).unwrap())
+                .collect();
+
+            let ntt_cfg: NTTConfig<'_, ScalarField> = NTTConfig::default();
+            ntt_inplace(HostSlice::from_mut_slice(&mut scalars[..]), NTTDir::kForward, &ntt_cfg).unwrap();
+
+            let poly = Polynomial::new(&scalars_lw[..]);
+            let evaluations = Polynomial::evaluate_fft::<Stark252PrimeField>(&poly, 1, None).unwrap();
+
+            for (s1, s2) in scalars
+                .iter()
+                .zip(evaluations.iter())
+            {
+                assert_eq!(s1.to_bytes_le(), s2.to_bytes_le());
+            }
+        }
+    }
+}

wrappers/rust/icicle-fields/icicle-stark252/src/polynomials/mod.rs

@@ -0,0 +1,10 @@
+use crate::field::{ScalarCfg, ScalarField};
+use icicle_core::impl_univariate_polynomial_api;
+
+impl_univariate_polynomial_api!("stark252", stark252, ScalarField, ScalarCfg);
+
+#[cfg(test)]
+mod tests {
+    use icicle_core::impl_polynomial_tests;
+    impl_polynomial_tests!(stark252, ScalarField);
+}

wrappers/rust/icicle-fields/icicle-stark252/src/vec_ops/mod.rs

@@ -0,0 +1,20 @@
+use crate::field::{ScalarCfg, ScalarField};
+
+use icicle_core::error::IcicleResult;
+use icicle_core::impl_vec_ops_field;
+use icicle_core::traits::IcicleResultWrap;
+use icicle_core::vec_ops::{VecOps, VecOpsConfig};
+use icicle_cuda_runtime::device_context::DeviceContext;
+use icicle_cuda_runtime::error::CudaError;
+use icicle_cuda_runtime::memory::HostOrDeviceSlice;
+
+impl_vec_ops_field!("stark252", stark252, ScalarField, ScalarCfg);
+
+#[cfg(test)]
+pub(crate) mod tests {
+    use crate::field::ScalarField;
+    use icicle_core::impl_vec_add_tests;
+    use icicle_core::vec_ops::tests::*;
+
+    impl_vec_add_tests!(ScalarField);
+}