fix cuda test - remove boost

Dockerfile

@@ -0,0 +1,28 @@
+# Use the specified base image
+FROM nvidia/cuda:12.0.0-devel-ubuntu22.04
+
+# Update and install dependencies
+RUN apt-get update && apt-get install -y \
+    cmake \
+    protobuf-compiler \
+    curl \
+    build-essential \
+    && rm -rf /var/lib/apt/lists/*
+
+# Install Rust
+RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y
+ENV PATH="/root/.cargo/bin:${PATH}"
+
+# Install Golang
+ENV GOLANG_VERSION 1.21.1
+RUN curl -L https://golang.org/dl/go${GOLANG_VERSION}.linux-amd64.tar.gz | tar -xz -C /usr/local
+ENV PATH="/usr/local/go/bin:${PATH}"
+
+# Set the working directory in the container
+WORKDIR /app
+
+# Copy the content of the local directory to the working directory
+COPY . .
+
+# Specify the default command for the container
+CMD ["/bin/bash"]

README.md

@@ -167,6 +167,16 @@ The script does the following:
 - Creates a test file with the curve name in ``src``. 
 
 Testing the new curve could be done by running the tests in ``tests_curve_name`` (e.g. ``tests_bls12_381``).
+
+## Docker
+
+We offer a simple Docker container so you can simply run ICICLE without settig everything up locally.
+
+```
+docker build -t <name_of_your_choice> .
+docker run --gpus all -it <name_of_your_choice> /bin/bash
+```
+
 ## Contributions
 
 Join our [Discord Server][DISCORD] and find us on the icicle channel. We will be happy to work together to support your use case and talk features, bugs and design.

icicle/README.md

@@ -1,4 +1,6 @@
-# Tests
+# ICICLE CUDA
+
+## Running tests
 
 ```sh
 mkdir -p build; cmake -S . -B build; cmake --build build; cd build && ctest; cd ..

icicle/primitives/test.cu

@@ -1,9 +1,7 @@
 #include "test_kernels.cuh"
-#include <boost/multiprecision/cpp_int.hpp>
 #include <cuda_runtime.h>
 #include <gtest/gtest.h>
 #include <iostream>
-namespace mp = boost::multiprecision;
 
 template <class T>
 int device_populate_random(T* d_elements, unsigned n)
@@ -23,15 +21,6 @@ int device_set(T* d_elements, T el, unsigned n)
   return cudaMemcpy(d_elements, h_elements, sizeof(T) * n, cudaMemcpyHostToDevice);
 }
 
-mp::int1024_t convert_to_boost_mp(uint32_t* a, uint32_t length)
-{
-  mp::int1024_t res = 0;
-  for (uint32_t i = 0; i < length; i++) {
-    res += (mp::int1024_t)(a[i]) << 32 * i;
-  }
-  return res;
-}
-
 class PrimitivesTest : public ::testing::Test
 {
 protected:
@@ -406,92 +395,6 @@ TEST_F(PrimitivesTest, INGO_MP_MULT)
   }
 }
 
-TEST_F(PrimitivesTest, INGO_MP_MSB_MULT)
-{
-  // MSB multiply, take n msb bits of multiplication, assert that the error is up to 1.
-  ASSERT_EQ(ingo_mp_msb_mult(scalars1, scalars2, res_scalars_wide, n), cudaSuccess);
-  std::cout << "INGO MSB   = 0x";
-  for (int i = 2 * scalar_field_t::TLC - 1; i >= 0; i--) {
-    std::cout << std::hex << res_scalars_wide[0].limbs_storage.limbs[i] << " ";
-  }
-  std::cout << std::endl;
-
-  ASSERT_EQ(mp_mult(scalars1, scalars2, res_scalars_wide_full), cudaSuccess);
-  std::cout << "ZKSYNC = 0x";
-  for (int i = 2 * scalar_field_t::TLC - 1; i >= 0; i--) {
-    std::cout << std::hex << res_scalars_wide_full[0].limbs_storage.limbs[i] << " ";
-  }
-
-  std::cout << std::endl;
-
-  // for (int i=scalar_field::TLC; i < 2*scalar_field::TLC - 1; i++)
-  // {
-  //   ASSERT_EQ(in_bound, true);
-  // }
-  // for (int j=0; j<n; j++)
-  // {
-  //   for (int i=0; i < 2*scalar_field::TLC - 1; i++)
-  //   {
-  //     ASSERT_EQ(res_scalars_wide_full[j].limbs_storage.limbs[i], res_scalars_wide[j].limbs_storage.limbs[i]);
-  //   }
-  // }
-  // mp testing
-  mp::int1024_t scalar_1_mp = 0;
-  mp::int1024_t scalar_2_mp = 0;
-  mp::int1024_t res_mp = 0;
-  mp::int1024_t res_gpu = 0;
-  uint32_t num_limbs = scalar_field_t::TLC;
-
-  for (int j = 0; j < n; j++) {
-    uint32_t* scalar1_limbs = scalars1[j].limbs_storage.limbs;
-    uint32_t* scalar2_limbs = scalars2[j].limbs_storage.limbs;
-    scalar_1_mp = convert_to_boost_mp(scalar1_limbs, num_limbs);
-    scalar_2_mp = convert_to_boost_mp(scalar2_limbs, num_limbs);
-    res_mp = scalar_1_mp * scalar_2_mp;
-    res_mp = res_mp >> (num_limbs * 32);
-    res_gpu = convert_to_boost_mp(&(res_scalars_wide[j]).limbs_storage.limbs[num_limbs], num_limbs);
-    std::cout << "res  mp = " << res_mp << std::endl;
-    std::cout << "res gpu = " << res_gpu << std::endl;
-    std::cout << "error = " << res_mp - res_gpu << std::endl;
-    bool upper_bound = res_gpu <= res_mp;
-    bool lower_bound = res_gpu > (res_mp - num_limbs);
-    bool in_bound = upper_bound && lower_bound;
-
-    ASSERT_EQ(in_bound, true);
-  }
-}
-
-TEST_F(PrimitivesTest, INGO_MP_MOD_MULT)
-{
-  std::cout << " taking num limbs " << std::endl;
-  uint32_t num_limbs = scalar_field_t::TLC;
-  std::cout << " calling gpu... = " << std::endl;
-  ASSERT_EQ(ingo_mp_mod_mult(scalars1, scalars2, res_scalars1, n), cudaSuccess);
-  std::cout << " gpu call done " << std::endl;
-  // mp testing
-  mp::int1024_t scalar_1_mp = 0;
-  mp::int1024_t scalar_2_mp = 0;
-  mp::int1024_t res_mp = 0;
-  mp::int1024_t res_gpu = 0;
-  mp::int1024_t p = convert_to_boost_mp(scalar_field_t::get_modulus().limbs, num_limbs);
-  std::cout << " p = " << p << std::endl;
-
-  for (int j = 0; j < n; j++) {
-    uint32_t* scalar1_limbs = scalars1[j].limbs_storage.limbs;
-    uint32_t* scalar2_limbs = scalars2[j].limbs_storage.limbs;
-    scalar_1_mp = convert_to_boost_mp(scalar1_limbs, num_limbs);
-    scalar_2_mp = convert_to_boost_mp(scalar2_limbs, num_limbs);
-    // std::cout << " s1 = " << scalar_1_mp << std::endl;
-    // std::cout << " s2 = " << scalar_2_mp << std::endl;
-    res_mp = (scalar_1_mp * scalar_2_mp) % p;
-    res_gpu = convert_to_boost_mp((res_scalars1[j]).limbs_storage.limbs, num_limbs);
-    std::cout << "res  mp = " << res_mp << std::endl;
-    std::cout << "res gpu = " << res_gpu << std::endl;
-    std::cout << "error = " << res_mp - res_gpu << std::endl;
-    ASSERT_EQ(res_gpu, res_mp);
-  }
-}
-
 TEST_F(PrimitivesTest, G2ECRandomPointsAreOnCurve)
 {
   for (unsigned i = 0; i < n; i++)