search

TiledTensor / TiledCUDA

TiledCUDA is a highly efficient kernel template library designed to elevate CUDA C’s level of abstraction for processing tiles.
MIT License
158 stars 10 forks source link

Rename `SharedTileIterator` to `TileIterator` to reduce redundancy #85

Closed haruhi55 closed 4 months ago

haruhi55 commented 4 months ago

TileIterator works for both global memory tiles and shared memory tiles. There is no significant difference between these two, as both are contiguous memory compared to distributed thread-local register files.

Therefore, shall we rename SharedTileIterator to TileIterator to simplify the concept?

KuangjuX commented 4 months ago

Agreed, we can rename SharedTileIterator to TileIterator so that it can be used for both Global and Shared memory.

haruhi55 commented 4 months ago

Let me make this change.