GPU Kernel Optimization

Prefetching Strategy

Anticipatory technique of loading data into shared memory before their use, thus hiding the latency of global memory accesses.

← Terug