ver4a/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

CUDA/HIP: Share the same unified memory allocation logic. (#12934)

Browse source 
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.
This commit is contained in:
David Huang 2025-04-15 17:20:38 +08:00 • committed by GitHub
parent 510676475f
commit 84778e9770
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
6 changed files with 22 additions and 26 deletions
Download patch file Download diff file Expand all files Collapse all files

2
ggml/src/ggml-cuda/vendors/hip.h vendored
View file

@ -71,6 +71,8 @@
#define cudaLaunchHostFunc hipLaunchHostFunc
#define cudaMalloc hipMalloc
#define cudaMallocHost(ptr, size) hipHostMalloc(ptr, size, hipHostMallocDefault)
#define cudaMallocManaged hipMallocManaged
#define cudaMemAdvise hipMemAdvise
#define cudaMemcpy hipMemcpy
#define cudaMemcpyAsync hipMemcpyAsync
#define cudaMemcpyPeerAsync hipMemcpyPeerAsync