Getting Started with CUDA

More documents

Recommendations

Info

CUBLAS Helper Functions cublasInit() Initializes CUBLAS library cublasShutdown() Releases resources used by CUBLAS library cublasGetError() Returns last error from CUBLAS core function (+ resets) cublasAlloc() Wrapper around cudaMalloc() to allocate space for array cublasFree() destroys object in GPU memory cublas[Set|Get][Vector|Matrix]() Copies array elements between CPU and GPU memory Accommodates non-unit strides © 2008 NVIDIA Corporation.
sgemmExample.c #include #include #include "cublas.h" int main(void) { float *a_h, *b_h, *c_h; float *a_d, *b_d, *c_d; float alpha = 1.0f, beta = 0.0f; int N = 2048, n2 = N*N; int nBytes, i; nBytes = n2*sizeof(float); a_h = (float *)malloc(nBytes); b_h = (float *)malloc(nBytes); c_h = (float *)malloc(nBytes); for (i=0; i < n2; i++) { } a_h[i] = rand() / (float) RAND_MAX; b_h[i] = rand() / (float) RAND_MAX; } cublasInit(); cublasAlloc(n2, sizeof(float), (void **)&a_d); cublasAlloc(n2, sizeof(float), (void **)&b_d); cublasAlloc(n2, sizeof(float), (void **)&c_d); cublasSetVector(n2, sizeof(float), a_h, 1, a_d, 1); cublasSetVector(n2, sizeof(float), b_h, 1, b_d, 1); cublasSgemm('n', 'n', N, N, N, alpha, a_d, N, b_d, N, beta, c_d, N); cublasGetVector(n2, sizeof(float), c_d, 1, c_h, 1); free(a_h); free(b_h); free(c_h); cublasFree(a_d); cublasFree(b_d); cublasFree(c_d); cublasShutdown(); return 0; © 2008 NVIDIA Corporation.
Page 1 and 2:
Getting Started with CUDA Greg Ruet
Page 3 and 4:
Outline CUDA programming model Basi
Page 5 and 6: CUDA Kernels and Threads Parallel p
Page 7 and 8: Thread Cooperation The Missing Piec
Page 9 and 10: Transparent Scalability Hardware is
Page 11 and 12: 10-Series Architecture 240 thread p
Page 13 and 14: Physical Memory Layout Host “Loca
Page 15 and 16: Key Parallel Abstractions in CUDA T
Page 17 and 18: CUDA Installation CUDA installation
Page 19 and 20: Compiling CUDA Code C/C++ CUDA Appl
Page 21 and 22: Outline CUDA programming model Basi
Page 23 and 24: GPU Memory Allocation / Release cud
Page 25 and 26: Data Movement Example int main(void
Page 35 and 36: Executing Code on the GPU Kernels a
Page 37 and 38: Launching Kernels Modified C functi
Page 39 and 40: CUDA Built-in Device Variables All
Page 41 and 42: Minimal Kernels __global__ void min
Page 43 and 44: Host Synchronization All kernel lau
Page 45 and 46: Variable Qualifiers (GPU code) __de
Page 47 and 48: GPU Thread Synchronization void __s
Page 49 and 50: Built-in Vector Types Can be used i
Page 51 and 52: CUDA Programming Resources We only
Page 53 and 54: CUBLAS Implementation of BLAS (Basi
Page 55: Using CUBLAS Interface to CUBLAS li
Page 59 and 60: Outline CUDA programming model Basi
Page 61 and 62: More on CUFFT For 2D and 3D transfo
Page 63 and 64: CUFFT Example #include #include #
Page 65: Getting Started with CUDA Greg Ruet
show all

Getting Started with CUDA

Create successful ePaper yourself

Delete template?

Save as template?