2 Commits

Author	SHA1	Message	Date
Adam Siemieniuk	a16211c032	[mlir][amx] Direct AMX data transfers (#154114) ... Extends Vector to AMX conversion to attempt populating AMX tiles directly from memory. When possible, contraction producers and consumers are replaced by AMX tile data transfer operations. This shortens data path by skipping intermediate register loads and stores.	2025-08-27 08:54:54 +02:00
Adam Siemieniuk	7d1b9cad87	[mlir][amx] Vector to AMX conversion pass (#151121) ... Adds a pass for Vector to AMX operation conversion. Initially, a direct rewrite for vector contraction in packed VNNI layout is supported. Operations are expected to already be in shapes which are AMX-compatible for the rewriting to occur.	2025-08-13 11:08:52 +02:00