Intra-warp
locality is often associated with strided accesses [19, 35], which
lead to divergent memory accesses when stride size is large. The
execution model, intra-warp locality, and potential memory divergence
together pose a great challenge for GPU cache management,
i.e., data blocks fetched by a divergent load instruction should be
cached as a wholistic group.
Intra-warplocality is often associated with strided accesses [19, 35], whichlead to divergent memory accesses when stride size is large. Theexecution model, intra-warp locality, and potential memory divergencetogether pose a great challenge for GPU cache management,i.e., data blocks fetched by a divergent load instruction should becached as a wholistic group.
การแปล กรุณารอสักครู่..