Double-buffer GEMM

[CNugteren]

This issue is mainly for information purposes, indicating that https://github.com/lsl036/CL-DB-GEMM is a repository with a double-buffer GEMM implemented using CLBlast. Might be interesting to have a closer look.

Found by @tangjinchuan.

[octaveoclx]

https://github.com/yzhaiustc/Optimizing-SGEMM-on-NVIDIA-Turing-GPUs?tab=readme-ov-file
also claims the near cuBLAS performance with double buffer in the last kernel.
From
Jinchuan