Dayuxiaoshui 4ae8c7e8bd [TE] align USE_MACA with MUSA GPU paths and docs (#1814) ... * [TRANSFER_ENGINE] align USE_MACA with MUSA GPU paths and docs Extend CUDA/MUSA/HIP preprocessor guards to include USE_MACA across TE core, tests, and examples; add maca:0 to NIC priority matrix JSON; exclude USE_MACA from USE_MNNVL auto CUDA enablement. Document MACA build (MACA_HOME, USE_MACA, optional MACA_RUNTIME_LIBS) in English and Chinese build guides. Fix maca.h cudaPointerAttributes mapping to mcPointerAttribute_t for MACA SDK compatibility.		2026-04-09 11:54:18 +08:00
..
hip.h	[TE/HIP] Switch to IPC mode by default and enable P2P access (#1344)	2026-01-09 15:54:49 +08:00
intra_nvlink.h	[TE] Nvlink intraNode Transport isolation (#1341)	2026-01-22 10:17:12 +08:00
maca.h	[TE] align USE_MACA with MUSA GPU paths and docs (#1814)	2026-04-09 11:54:18 +08:00
mlu.h	[TE] feat: setup the RDMA for mlu device. (#1799)	2026-04-08 12:08:53 +08:00
mnnvl.h	ubshmem transport with fabric mem without conflicts (#1399)	2026-01-26 15:18:20 +08:00
musa.h	[TE] feat: Support H20 intraNode nvlink by introducing fallback mechanism to leverage cudaIPC (#1234)	2025-12-20 10:57:07 +08:00
ubshmem.h	ubshmem transport with fabric mem without conflicts (#1399)	2026-01-26 15:18:20 +08:00