Feat/roman/extension fields

.gitignore

@@ -19,3 +19,4 @@
 **/wrappers/rust/icicle-cuda-runtime/src/bindings.rs
 **/build/*
 **tar.gz
+icicle/backend/cuda

examples/c++/best-practice-ntt/example.cpp

@@ -44,7 +44,7 @@ int main(int argc, char* argv[])
   ConfigExtension backend_cfg_ext;
   backend_cfg_ext.set(CudaBackendConfig::CUDA_NTT_FAST_TWIDDLES_MODE, true);
   ntt_init_domain_cfg.ext = &backend_cfg_ext;
-  ICICLE_CHECK(bn254_ntt_init_domain(&basic_root, ntt_init_domain_cfg));
+  ICICLE_CHECK(bn254_ntt_init_domain(&basic_root, &ntt_init_domain_cfg));
 
   std::cout << "Concurrent Download, Upload, and Compute In-place NTT" << std::endl;
   int nof_blocks = 32;
@@ -89,7 +89,7 @@ int main(int argc, char* argv[])
     std::cout << "Compute Vector: " << vec_compute << std::endl;
     std::cout << "Transfer Vector: " << vec_transfer << std::endl;
     START_TIMER(inplace);
-    bn254_ntt(d_vec[vec_compute], ntt_size, NTTDir::kForward, config_compute, d_vec[vec_compute]);
+    bn254_ntt(d_vec[vec_compute], ntt_size, NTTDir::kForward, &config_compute, d_vec[vec_compute]);
     // we have to delay upload to device relative to download from device by one block: preserve write after read
     for (int i = 0; i <= nof_blocks; i++) {
       if (i < nof_blocks) {

examples/c++/msm/example.cpp

@@ -38,7 +38,7 @@ int main(int argc, char* argv[])
   std::cout << "\nRunning MSM kernel with on-host inputs" << std::endl;
   // Execute the MSM kernel
   START_TIMER(MSM_host_mem);
-  ICICLE_CHECK(bn254_msm(scalars.get(), points.get(), msm_size, config, &result));
+  ICICLE_CHECK(bn254_msm(scalars.get(), points.get(), msm_size, &config, &result));
   END_TIMER(MSM_host_mem, "MSM from host-memory took");
   std::cout << projective_t::to_affine(result) << std::endl;
 
@@ -91,7 +91,7 @@ int main(int argc, char* argv[])
   config.are_points_on_device = false;
   g2_projective_t g2_result;
   START_TIMER(MSM_g2);
-  ICICLE_CHECK(bn254_g2_msm(scalars.get(), g2_points.get(), msm_size, config, &g2_result));
+  ICICLE_CHECK(bn254_g2_msm(scalars.get(), g2_points.get(), msm_size, &config, &g2_result));
   END_TIMER(MSM_g2, "MSM G2 from host-memory took");
   std::cout << g2_projective_t::to_affine(g2_result) << std::endl;
 

examples/c++/ntt/example.cpp

@@ -37,7 +37,7 @@ int main(int argc, char* argv[])
   backend_cfg_ext.set(
     CudaBackendConfig::CUDA_NTT_FAST_TWIDDLES_MODE, true); // optionally construct fast_twiddles for CUDA backend
   ntt_init_domain_cfg.ext = &backend_cfg_ext;
-  ICICLE_CHECK(bn254_ntt_init_domain(&basic_root, ntt_init_domain_cfg));
+  ICICLE_CHECK(bn254_ntt_init_domain(&basic_root, &ntt_init_domain_cfg));
 
   // ntt configuration
   NTTConfig<scalar_t> config = default_ntt_config<scalar_t>();
@@ -46,13 +46,13 @@ int main(int argc, char* argv[])
   config.batch_size = batch_size;
 
   // warmup
-  ICICLE_CHECK(bn254_ntt(input.get(), ntt_size, NTTDir::kForward, config, output.get()));
+  ICICLE_CHECK(bn254_ntt(input.get(), ntt_size, NTTDir::kForward, &config, output.get()));
 
   // NTT radix-2 alg
   std::cout << "\nRunning NTT radix-2 alg with on-host data" << std::endl;
   ntt_cfg_ext.set(CudaBackendConfig::CUDA_NTT_ALGORITHM, CudaBackendConfig::NttAlgorithm::Radix2);
   START_TIMER(Radix2);
-  ICICLE_CHECK(bn254_ntt(input.get(), ntt_size, NTTDir::kForward, config, output.get()));
+  ICICLE_CHECK(bn254_ntt(input.get(), ntt_size, NTTDir::kForward, &config, output.get()));
   END_TIMER(Radix2, "Radix2 NTT");
 
   std::cout << "Validating output" << std::endl;
@@ -62,7 +62,7 @@ int main(int argc, char* argv[])
   std::cout << "\nRunning NTT mixed-radix alg with on-host data" << std::endl;
   ntt_cfg_ext.set(CudaBackendConfig::CUDA_NTT_ALGORITHM, CudaBackendConfig::NttAlgorithm::MixedRadix);
   START_TIMER(MixedRadix);
-  ICICLE_CHECK(bn254_ntt(input.get(), ntt_size, NTTDir::kForward, config, output.get()));
+  ICICLE_CHECK(bn254_ntt(input.get(), ntt_size, NTTDir::kForward, &config, output.get()));
   END_TIMER(MixedRadix, "MixedRadix NTT");
 
   std::cout << "Validating output" << std::endl;

examples/c++/pedersen-commitment/example.cpp

@@ -150,7 +150,7 @@ int main(int argc, char** argv)
   std::cout << "Executing MSM" << std::endl;
   auto config = default_msm_config();
   START_TIMER(msm);
-  bn254_msm(scalars, points, N + 1, config, &result);
+  bn254_msm(scalars, points, N + 1, &config, &result);
   END_TIMER(msm, "Time to execute MSM");
 
   std::cout << "Computed commitment: " << result << std::endl;

examples/c++/polynomial_multiplication/example.cpp

@@ -37,7 +37,8 @@ int main(int argc, char** argv)
 
   // init domain
   scalar_t basic_root = scalar_t::omega(NTT_LOG_SIZE);
-  bn254_ntt_init_domain(&basic_root, default_ntt_init_domain_config());
+  auto config = default_ntt_init_domain_config();
+  bn254_ntt_init_domain(&basic_root, &config);
 
   // (1) cpu allocation
   auto polyA = std::make_unique<scalar_t[]>(NTT_SIZE);
@@ -64,8 +65,8 @@ int main(int argc, char** argv)
     ntt_config.are_inputs_on_device = false;
     ntt_config.are_outputs_on_device = true;
     ntt_config.ordering = Ordering::kNM;
-    ICICLE_CHECK(bn254_ntt(polyA.get(), NTT_SIZE, NTTDir::kForward, ntt_config, d_polyA));
-    ICICLE_CHECK(bn254_ntt(polyB.get(), NTT_SIZE, NTTDir::kForward, ntt_config, d_polyB));
+    ICICLE_CHECK(bn254_ntt(polyA.get(), NTT_SIZE, NTTDir::kForward, &ntt_config, d_polyA));
+    ICICLE_CHECK(bn254_ntt(polyB.get(), NTT_SIZE, NTTDir::kForward, &ntt_config, d_polyB));
 
     // (4) multiply A,B
     VecOpsConfig config{
@@ -76,13 +77,13 @@ int main(int argc, char** argv)
       false,  // is_async
       nullptr // ext
     };
-    ICICLE_CHECK(bn254_vector_mul(d_polyA, d_polyB, NTT_SIZE, config, d_polyRes));
+    ICICLE_CHECK(bn254_vector_mul(d_polyA, d_polyB, NTT_SIZE, &config, d_polyRes));
 
     // (5) INTT (in place)
     ntt_config.are_inputs_on_device = true;
     ntt_config.are_outputs_on_device = true;
     ntt_config.ordering = Ordering::kMN;
-    ICICLE_CHECK(bn254_ntt(d_polyRes, NTT_SIZE, NTTDir::kInverse, ntt_config, d_polyRes));
+    ICICLE_CHECK(bn254_ntt(d_polyRes, NTT_SIZE, NTTDir::kInverse, &ntt_config, d_polyRes));
 
     if (print) { END_TIMER(poly_multiply, "polynomial multiplication took"); }
 

icicle/include/icicle/api/babybear.h

@@ -9,96 +9,65 @@
 #include "icicle/ntt.h"
 #include "icicle/vec_ops.h"
 
-extern "C" void babybear_generate_scalars(babybear::scalar_t* scalars, int size);
+extern "C" eIcicleError babybear_ntt_init_domain(
+  babybear::scalar_t* primitive_root, const NTTInitDomainConfig* config);
 
-extern "C" void babybear_scalar_convert_montgomery(
-  const babybear::scalar_t* input, uint64_t size, bool is_into, const VecOpsConfig& config, babybear::scalar_t* output);
+extern "C" eIcicleError babybear_ntt(
+  const babybear::scalar_t* input, int size, NTTDir dir, const NTTConfig<babybear::scalar_t>* config, babybear::scalar_t* output);
 
-extern "C" eIcicleError babybear_ntt_init_domain(babybear::scalar_t* primitive_root, const NTTInitDomainConfig& config);
+extern "C" eIcicleError babybear_ntt_release_domain();
 
-extern "C" eIcicleError babybear_ntt(
+extern "C" eIcicleError babybear_vector_mul(
+  const babybear::scalar_t* vec_a, const babybear::scalar_t* vec_b, uint64_t n, const VecOpsConfig* config, babybear::scalar_t* result);
+
+extern "C" eIcicleError babybear_vector_add(
+  const babybear::scalar_t* vec_a, const babybear::scalar_t* vec_b, uint64_t n, const VecOpsConfig* config, babybear::scalar_t* result);
+
+extern "C" eIcicleError babybear_vector_sub(
+  const babybear::scalar_t* vec_a, const babybear::scalar_t* vec_b, uint64_t n, const VecOpsConfig* config, babybear::scalar_t* result);
+
+extern "C" eIcicleError babybear_matrix_transpose(
   const babybear::scalar_t* input,
-  int size,
-  NTTDir dir,
-  const NTTConfig<babybear::scalar_t>& config,
+  uint32_t nof_rows,
+  uint32_t nof_cols,
+  const VecOpsConfig* config,
   babybear::scalar_t* output);
 
-extern "C" eIcicleError babybear_ntt_release_domain();
+extern "C" eIcicleError babybear_bit_reverse(
+  const babybear::scalar_t* input, uint64_t n, const VecOpsConfig* config, babybear::scalar_t* output);
+
 
 extern "C" eIcicleError babybear_extension_vector_mul(
-  const babybear::extension_t* vec_a,
-  const babybear::extension_t* vec_b,
-  uint64_t n,
-  const VecOpsConfig& config,
-  babybear::extension_t* result);
+  const babybear::extension_t* vec_a, const babybear::extension_t* vec_b, uint64_t n, const VecOpsConfig* config, babybear::extension_t* result);
 
 extern "C" eIcicleError babybear_extension_vector_add(
-  const babybear::extension_t* vec_a,
-  const babybear::extension_t* vec_b,
-  uint64_t n,
-  const VecOpsConfig& config,
-  babybear::extension_t* result);
+  const babybear::extension_t* vec_a, const babybear::extension_t* vec_b, uint64_t n, const VecOpsConfig* config, babybear::extension_t* result);
 
 extern "C" eIcicleError babybear_extension_vector_sub(
-  const babybear::extension_t* vec_a,
-  const babybear::extension_t* vec_b,
-  uint64_t n,
-  const VecOpsConfig& config,
-  babybear::extension_t* result);
+  const babybear::extension_t* vec_a, const babybear::extension_t* vec_b, uint64_t n, const VecOpsConfig* config, babybear::extension_t* result);
 
 extern "C" eIcicleError babybear_extension_matrix_transpose(
   const babybear::extension_t* input,
   uint32_t nof_rows,
   uint32_t nof_cols,
-  const VecOpsConfig& config,
+  const VecOpsConfig* config,
   babybear::extension_t* output);
 
 extern "C" eIcicleError babybear_extension_bit_reverse(
-  const babybear::extension_t* input, uint64_t n, const VecOpsConfig& config, babybear::extension_t* output);
+  const babybear::extension_t* input, uint64_t n, const VecOpsConfig* config, babybear::extension_t* output);
 
-extern "C" void babybear_extension_generate_scalars(babybear::extension_t* scalars, int size);
 
-extern "C" eIcicleError babybear_extension_scalar_convert_montgomery(
-  const babybear::extension_t* input,
-  uint64_t size,
-  bool is_into,
-  const VecOpsConfig& config,
-  babybear::extension_t* output);
+extern "C" void babybear_extension_generate_scalars(babybear::extension_t* scalars, int size);
 
-extern "C" eIcicleError babybear_vector_mul(
-  const babybear::scalar_t* vec_a,
-  const babybear::scalar_t* vec_b,
-  uint64_t n,
-  const VecOpsConfig& config,
-  babybear::scalar_t* result);
+extern "C" eIcicleError babybear_extension_scalar_convert_montgomery(  
+  const babybear::extension_t* input, uint64_t size, bool is_into, const VecOpsConfig* config, babybear::extension_t* output);
 
-extern "C" eIcicleError babybear_vector_add(
-  const babybear::scalar_t* vec_a,
-  const babybear::scalar_t* vec_b,
-  uint64_t n,
-  const VecOpsConfig& config,
-  babybear::scalar_t* result);
+extern "C" eIcicleError babybear_extension_ntt(
+  const babybear::extension_t* input, int size, NTTDir dir, const NTTConfig<babybear::scalar_t>* config, babybear::extension_t* output);
 
-extern "C" eIcicleError babybear_vector_sub(
-  const babybear::scalar_t* vec_a,
-  const babybear::scalar_t* vec_b,
-  uint64_t n,
-  const VecOpsConfig& config,
-  babybear::scalar_t* result);
 
-extern "C" eIcicleError babybear_matrix_transpose(
-  const babybear::scalar_t* input,
-  uint32_t nof_rows,
-  uint32_t nof_cols,
-  const VecOpsConfig& config,
-  babybear::scalar_t* output);
+extern "C" void babybear_generate_scalars(babybear::scalar_t* scalars, int size);
 
-extern "C" eIcicleError babybear_bit_reverse(
-  const babybear::scalar_t* input, uint64_t n, const VecOpsConfig& config, babybear::scalar_t* output);
+extern "C" void babybear_scalar_convert_montgomery(
+  const babybear::scalar_t* input, uint64_t size, bool is_into, const VecOpsConfig* config, babybear::scalar_t* output);
 
-extern "C" eIcicleError babybear_extension_ntt(
-  const babybear::extension_t* input,
-  int size,
-  NTTDir dir,
-  const NTTConfig<babybear::scalar_t>& config,
-  babybear::extension_t* output);