]> git.djapps.eu Git - pkg/ggml/sources/ggml/commit
CUDA/HIP: Share the same unified memory allocation logic. (llama/12934)
authorDavid Huang <redacted>
Tue, 15 Apr 2025 09:20:38 +0000 (17:20 +0800)
committerGeorgi Gerganov <redacted>
Thu, 24 Apr 2025 15:36:25 +0000 (18:36 +0300)
commitff0c43e343be9fde8b0a80be310f9972806ffc15
treee6da2820a5746638247c38fca3ab55b4735942e9
parentfa86452b8f662287638f618709223d8c7a1b64eb
CUDA/HIP: Share the same unified memory allocation logic. (llama/12934)

Replace compile-time `GGML_HIP_UMA` with environment variable `GGML_CUDA_ENABLE_UNIFIED_MEMORY`. This unifies the usage on NVIDIA and AMD GPUs, and allows a single binary to be shared between integrated and dedicated GPUs.
CMakeLists.txt
src/ggml-cuda/ggml-cuda.cu
src/ggml-cuda/vendors/hip.h
src/ggml-hip/CMakeLists.txt