Mirrors/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-10-30 06:30:15 +01:00
Code Issues Packages Projects Releases Wiki Activity

mtgpu: enable VMM (#9597)

Browse Source 
Signed-off-by: Xiaodong Ye <xiaodong.ye@mthreads.com>
This commit is contained in:
R0CKSTAR 2024-09-26 09:27:40 +08:00 committed by GitHub
parent ea9c32be71
commit 7691654c68
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 6 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
ggml/src/ggml-cuda.cu
View File

@ -187,7 +187,7 @@ static ggml_cuda_device_info ggml_cuda_init() {
for (int id = 0; id < info.device_count; ++id) { for (int id = 0; id < info.device_count; ++id) {
int device_vmm = 0; int device_vmm = 0;
#if !defined(GGML_USE_HIPBLAS) && !defined(GGML_CUDA_NO_VMM) && !defined(GGML_USE_MUSA) #if !defined(GGML_USE_HIPBLAS) && !defined(GGML_CUDA_NO_VMM)
CUdevice device; CUdevice device;
CU_CHECK(cuDeviceGet(&device, id)); CU_CHECK(cuDeviceGet(&device, id));
CU_CHECK(cuDeviceGetAttribute(&device_vmm, CU_DEVICE_ATTRIBUTE_VIRTUAL_MEMORY_MANAGEMENT_SUPPORTED, device)); CU_CHECK(cuDeviceGetAttribute(&device_vmm, CU_DEVICE_ATTRIBUTE_VIRTUAL_MEMORY_MANAGEMENT_SUPPORTED, device));
@ -199,7 +199,7 @@ static ggml_cuda_device_info ggml_cuda_init() {
alloc_prop.location.id = id; alloc_prop.location.id = id;
CU_CHECK(cuMemGetAllocationGranularity(&info.devices[id].vmm_granularity, &alloc_prop, CU_MEM_ALLOC_GRANULARITY_RECOMMENDED)); CU_CHECK(cuMemGetAllocationGranularity(&info.devices[id].vmm_granularity, &alloc_prop, CU_MEM_ALLOC_GRANULARITY_RECOMMENDED));
} }
#endif // !defined(GGML_USE_HIPBLAS) && !defined(GGML_CUDA_NO_VMM) && !defined(GGML_USE_MUSA) #endif // !defined(GGML_USE_HIPBLAS) && !defined(GGML_CUDA_NO_VMM)
info.devices[id].vmm = !!device_vmm; info.devices[id].vmm = !!device_vmm;
cudaDeviceProp prop; cudaDeviceProp prop;
@ -335,7 +335,7 @@ struct ggml_cuda_pool_leg : public ggml_cuda_pool {
}; };
// pool with virtual memory // pool with virtual memory
#if !defined(GGML_USE_HIPBLAS) && !defined(GGML_CUDA_NO_VMM) && !defined(GGML_USE_MUSA) #if !defined(GGML_USE_HIPBLAS) && !defined(GGML_CUDA_NO_VMM)
struct ggml_cuda_pool_vmm : public ggml_cuda_pool { struct ggml_cuda_pool_vmm : public ggml_cuda_pool {
static const size_t CUDA_POOL_VMM_MAX_SIZE = 1ull << 35; // 32 GB static const size_t CUDA_POOL_VMM_MAX_SIZE = 1ull << 35; // 32 GB
@ -429,14 +429,14 @@ struct ggml_cuda_pool_vmm : public ggml_cuda_pool {
GGML_ASSERT(ptr == (void *) (pool_addr + pool_used)); GGML_ASSERT(ptr == (void *) (pool_addr + pool_used));
} }
}; };
#endif // !defined(GGML_USE_HIPBLAS) && !defined(GGML_CUDA_NO_VMM) && !defined(GGML_USE_MUSA) #endif // !defined(GGML_USE_HIPBLAS) && !defined(GGML_CUDA_NO_VMM)
std::unique_ptr<ggml_cuda_pool> ggml_backend_cuda_context::new_pool_for_device(int device) { std::unique_ptr<ggml_cuda_pool> ggml_backend_cuda_context::new_pool_for_device(int device) {
#if !defined(GGML_USE_HIPBLAS) && !defined(GGML_CUDA_NO_VMM) && !defined(GGML_USE_MUSA) #if !defined(GGML_USE_HIPBLAS) && !defined(GGML_CUDA_NO_VMM)
if (ggml_cuda_info().devices[device].vmm) { if (ggml_cuda_info().devices[device].vmm) {
return std::unique_ptr<ggml_cuda_pool>(new ggml_cuda_pool_vmm(device)); return std::unique_ptr<ggml_cuda_pool>(new ggml_cuda_pool_vmm(device));
} }
#endif // !defined(GGML_USE_HIPBLAS) && !defined(GGML_CUDA_NO_VMM) && !defined(GGML_USE_MUSA) #endif // !defined(GGML_USE_HIPBLAS) && !defined(GGML_CUDA_NO_VMM)
return std::unique_ptr<ggml_cuda_pool>(new ggml_cuda_pool_leg(device)); return std::unique_ptr<ggml_cuda_pool>(new ggml_cuda_pool_leg(device));
} }