Originally published at: Focus on Your Algorithm—NVIDIA CUDA Tile Handles the Hardware | NVIDIA Technical Blog
With its largest advancement since the NVIDIA CUDA platform was invented in 2006, CUDA 13.1 is launching NVIDIA CUDA Tile. This exciting innovation introduces a virtual instruction set for tile-based parallel programming, focusing on the ability to write algorithms at a higher level and abstract away the details of specialized hardware, such as tensor cores.…
Hey, great blog! I find that the link if CUDA Tile IR is same with link of NVIDIA cuTile Python. Is this intended or unexpected?
The only real progress would be to eliminate Python, PyTorch, and all those libraries, and go back to pure C++, letting those who truly understand hardware do the work. Maximum efficiency.
Maybe LLM can do the mapping in the future
Can anyone tell me about the relationship between `PTX` and `TileIR`? Do we need to translate TileIR to PTX or just TileIR to SASS?
The relations and differences Tile IR and PTX are addressed in the documentation: