llama.cpp/ggml/src/ggml-metal
Jeremy Demeule 165caaf5fb
metal: use shared buffers on eGPU (#17866)
* metal: use shared buffers on eGPU

With #15906, I noticed on important regression when using metal backend on eGPU.
This commit restore the previous behavior and add an option to force its activation.

* metal: use shared buffers on eGPU

* metal: use shared buffers on eGPU
2025-12-15 16:14:49 +02:00
..
CMakeLists.txt metal : refactor + optimize v2 (#15995) 2025-09-17 20:38:12 +03:00
ggml-metal-common.cpp metal : fix loop bound in ggml_mem_ranges (#16412) 2025-10-03 19:18:56 +03:00
ggml-metal-common.h metal : refactor + optimize v2 (#15995) 2025-09-17 20:38:12 +03:00
ggml-metal-context.h metal : refactor + optimize v2 (#15995) 2025-09-17 20:38:12 +03:00
ggml-metal-context.m metal : add residency sets keep-alive heartbeat (#17766) 2025-12-05 19:38:54 +02:00
ggml-metal-device.cpp metal: SSM kernel improvements (#17876) 2025-12-09 21:30:02 +02:00
ggml-metal-device.h metal: SSM kernel improvements (#17876) 2025-12-09 21:30:02 +02:00
ggml-metal-device.m metal: use shared buffers on eGPU (#17866) 2025-12-15 16:14:49 +02:00
ggml-metal-impl.h metal: SSM kernel improvements (#17876) 2025-12-09 21:30:02 +02:00
ggml-metal-ops.cpp metal: SSM kernel improvements (#17876) 2025-12-09 21:30:02 +02:00
ggml-metal-ops.h metal: TRI, FILL, EXPM1, SOFTPLUS (#16623) 2025-12-04 19:12:19 +02:00
ggml-metal.cpp ggml : add ggml_top_k (#17365) 2025-11-25 15:31:43 +02:00
ggml-metal.metal metal: SSM kernel improvements (#17876) 2025-12-09 21:30:02 +02:00