Loop blocking for linear algebra codes often have three levels: register blocking, L2 cache blocking, and L3 cache (or TLB) blocking.
Some notes on blocking for different purposes in GEMM operations.
2 Matching Annotations
- Dec 2024
-
community.intel.com community.intel.com
-
- Oct 2019
-
jackwish.net jackwish.net
-
通用矩阵乘（GEMM）优化算法
软件优化策略: 1)改进访存局部性; 2)利用向量指令.
-