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

ggml : add metal backend registry / device (#9713)

Browse source 
* ggml : add metal backend registry / device

ggml-ci

* metal : fix names [no ci]

* metal : global registry and device instances

ggml-ci

* cont : alternative initialization of global objects

ggml-ci

* llama : adapt to backend changes

ggml-ci

* fixes

* metal : fix indent

* metal : fix build when MTLGPUFamilyApple3 is not available

ggml-ci

* fix merge

* metal : avoid unnecessary singleton accesses

ggml-ci

* metal : minor fix [no ci]

* metal : g_state -> g_ggml_ctx_dev_main [no ci]

* metal : avoid reference of device context in the backend context

ggml-ci

* metal : minor [no ci]

* metal : fix maxTransferRate check

* metal : remove transfer rate stuff

---------

Co-authored-by: slaren <slarengh@gmail.com>
This commit is contained in:
Georgi Gerganov 2024-10-07 18:27:51 +03:00 • committed by GitHub
parent 96b6912103
commit d5ac8cf2f2
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
6 changed files with 535 additions and 284 deletions
Download patch file Download diff file Expand all files Collapse all files

21
ggml/src/ggml-backend.cpp
View file

@ -463,6 +463,7 @@ enum ggml_backend_dev_type ggml_backend_dev_type(ggml_backend_dev_t device) {
}
void ggml_backend_dev_get_props(ggml_backend_dev_t device, struct ggml_backend_dev_props * props) {
memset(props, 0, sizeof(*props));
device->iface.get_props(device, props);
}
@ -479,6 +480,10 @@ ggml_backend_buffer_type_t ggml_backend_dev_buffer_type(ggml_backend_dev_t devic
}
ggml_backend_buffer_type_t ggml_backend_dev_host_buffer_type(ggml_backend_dev_t device) {
if (device->iface.get_host_buffer_type == NULL) {
return NULL;
}
return device->iface.get_host_buffer_type(device);
}
@ -525,6 +530,10 @@ void * ggml_backend_reg_get_proc_address(ggml_backend_reg_t reg, const char * na
#include "ggml-cuda.h"
#endif
#ifdef GGML_USE_METAL
#include "ggml-metal.h"
#endif
struct ggml_backend_registry {
std::vector<ggml_backend_reg_t> backends;
std::vector<ggml_backend_dev_t> devices;
@ -533,10 +542,13 @@ struct ggml_backend_registry {
#ifdef GGML_USE_CUDA
register_backend(ggml_backend_cuda_reg());
#endif
#ifdef GGML_USE_METAL
register_backend(ggml_backend_metal_reg());
#endif
register_backend(ggml_backend_cpu_reg());
// TODO: sycl, metal, vulkan, kompute, cann
// TODO: sycl, vulkan, kompute, cann
}
void register_backend(ggml_backend_reg_t reg) {
@ -1118,9 +1130,10 @@ static void ggml_backend_cpu_device_get_props(ggml_backend_dev_t dev, struct ggm
props->type = ggml_backend_cpu_device_get_type(dev);
ggml_backend_cpu_device_get_memory(dev, &props->memory_free, &props->memory_total);
props->caps = {
/* async */ false,
/* host_buffer */ false,
/* events */ false,
/* .async = */ false,
/* .host_buffer = */ false,
/* .buffer_from_host_ptr = */ true,
/* .events = */ false,
};
}