Top 1K Features Creators Events Podcasts Books Extensions Interviews Blog Explorer CSV

Scale

< >

Scale is a programming language created in 2024.

#3078on PLDB 0Years Old
HomepageBlog

SCALE is a GPGPU programming toolkit that allows CUDA applications to be natively compiled for AMD GPUs.


Example from the web:
#include <vector> #include <iostream> // The kernel we are going to launch __global__ void basicSum(const int * a, const int * b, size_t n, int * out) { int idx = threadIdx.x + blockIdx.x * blockDim.x; if(idx < n) { out[idx] = a[idx] + b[idx]; } } // A generic helper function to simplify error handling. void check(cudaError_t error, const char * file, size_t line) { if (error != cudaSuccess) { std::cout << "cuda error: " << cudaGetErrorString(error) << " at " << file << ":" << line << std::endl; exit(1); } } // A wrapper for the helper function above to include the filename and line number // where the error occurs into the output. #define CHECK(error) check(error, __FILE__, __LINE__) int main(int argc, char ** argv) { const size_t N = 4096; const size_t BYTES = N * sizeof(int); std::vector<int> a(N); std::vector<int> b(N); std::vector<int> out(N); // Generate input data for (size_t i = 0; i < N; i++) { a[i] = i * 2; b[i] = N - i; } int * devA; int * devB; int * devOut; // Allocate memory for the inputs and the output CHECK(cudaMalloc(&devA, BYTES)); CHECK(cudaMalloc(&devB, BYTES)); CHECK(cudaMalloc(&devOut, BYTES)); // Copy the input data to the device CHECK(cudaMemcpy(devA, a.data(), BYTES, cudaMemcpyHostToDevice)); CHECK(cudaMemcpy(devB, b.data(), BYTES, cudaMemcpyHostToDevice)); // Launch the kernel basicSum<<<N / 256 + 1, 256>>>(devA, devB, N, devOut); CHECK(cudaDeviceSynchronize()); CHECK(cudaGetLastError()); // Copy the output data back to host CHECK(cudaMemcpy(out.data(), devOut, BYTES, cudaMemcpyDeviceToHost)); // Free up the memory we allocated for the inputs and the output CHECK(cudaFree(devA)); CHECK(cudaFree(devB)); CHECK(cudaFree(devOut)); // Test that the output matches our expectations for (size_t i = 0; i < N; i++) { if (a[i] + b[i] != out[i]) { std::cout << "Incorrect sum: " << a[i] << " + " << b[i] << " = " << out[i] << " ?\n"; } } std::cout << "Example finished" << std::endl; return 0; }

- Build the next great programming language Add Issues About Search Keywords Livestreams Labs Resources Acknowledgements

Built with Scroll v154.3.0