CUDA - From Correctness to Performance

This repo includes codes & examples for "CUDA - From Correctness to Performance".

How to Build

Make sure you have installed the CUDA toolkit, and a CUDA-compatible GPU is available.

Run make all to build this repo.

Usage:

./gemm_test <n> <m> <k> [implementation]

If implementation is not specified, all implementations will be benchmarked.

Name		Name	Last commit message	Last commit date
Latest commit History 15 Commits
.gitignore		.gitignore
LICENSE		LICENSE
Makefile		Makefile
README.md		README.md
gemm_cpu_naive.cc		gemm_cpu_naive.cc
gemm_cpu_naive.h		gemm_cpu_naive.h
gemm_cpu_simd.cc		gemm_cpu_simd.cc
gemm_cpu_simd.h		gemm_cpu_simd.h
gemm_gpu_1thread.cu		gemm_gpu_1thread.cu
gemm_gpu_1thread.h		gemm_gpu_1thread.h
gemm_gpu_mult_block.cu		gemm_gpu_mult_block.cu
gemm_gpu_mult_block.h		gemm_gpu_mult_block.h
gemm_gpu_mult_block_no_restrict.cu		gemm_gpu_mult_block_no_restrict.cu
gemm_gpu_mult_block_no_restrict.h		gemm_gpu_mult_block_no_restrict.h
gemm_gpu_mult_block_no_restrict_reg.cu		gemm_gpu_mult_block_no_restrict_reg.cu
gemm_gpu_mult_block_no_restrict_reg.h		gemm_gpu_mult_block_no_restrict_reg.h
gemm_gpu_mult_thread.cu		gemm_gpu_mult_thread.cu
gemm_gpu_mult_thread.h		gemm_gpu_mult_thread.h
gemm_gpu_tiling.cu		gemm_gpu_tiling.cu
gemm_gpu_tiling.h		gemm_gpu_tiling.h
gemm_test.cc		gemm_test.cc