Fix: Run with cuda12 on A100 and H100 GPUs.

weifengliu-ssslab / Benchmark_SpMV_using_CSR5

CSR5-based SpMV on CPUs, GPUs and Xeon Phi

MIT License

93 stars 31 forks source link

Open pmpakos opened 3 months ago

pmpakos commented 3 months ago

Fixed deprecated shfl functions, replaced with shfl_sync equivalent ones. Now able to run on Ampere and Hopper GPUs.