ggml-backend : add device and backend reg interfaces (llama/9707)

Also:

- metal : fix compute pass descriptor autorelease crash
- ggml-backend : add device description to CPU backend
- ggml: unify backend logging mechanism

ggml/include/ggml-backend.h

@@ -164,7 +164,7 @@ extern "C" {
     GGML_API size_t             ggml_backend_reg_dev_count(ggml_backend_reg_t reg);
     GGML_API ggml_backend_dev_t ggml_backend_reg_dev_get(ggml_backend_reg_t reg, size_t index);
     GGML_API void *             ggml_backend_reg_get_proc_address(ggml_backend_reg_t reg, const char * name);
-    GGML_API void               ggml_backend_reg_set_log_callback(ggml_backend_reg_t reg, ggml_log_callback log_callback, void * user_data);
+
 
     // Functions that may be obtained using ggml_backend_reg_get_proc_address
     typedef ggml_backend_buffer_type_t (*ggml_backend_split_buffer_type_t)(const float *);
@@ -184,9 +184,6 @@ extern "C" {
     GGML_API ggml_backend_dev_t ggml_backend_dev_by_name(const char * name);
     GGML_API ggml_backend_dev_t ggml_backend_dev_by_type(enum ggml_backend_dev_type type);
 
-    // Set the log callback for all registered backends
-    GGML_API void ggml_backend_set_log_callback(ggml_log_callback log_callback, void * user_data);
-
     // Direct backend (stream) initialization
     // = ggml_backend_dev_init(ggml_backend_dev_by_name(name), params)
     GGML_API ggml_backend_t ggml_backend_init_by_name(const char * name, const char * params);

ggml/include/ggml-cann.h

@@ -116,17 +116,6 @@ GGML_API void ggml_backend_cann_get_device_memory(int32_t device,
                                                   size_t* free,
                                                   size_t* total);
 
-/**
- * @brief Set the logging callback for GGML.
- *
- * This function sets the logging callback and user data for logging.
- *
- * @param log_callback The logging callback to set.
- * @param user_data User data to pass to the logging callback.
- */
-GGML_API void ggml_backend_cann_log_set_callback(ggml_log_callback log_callback,
-                                                 void* user_data);
-
 #ifdef __cplusplus
 }
 #endif

ggml/include/ggml-cuda.h

@@ -40,8 +40,6 @@ GGML_API void ggml_backend_cuda_get_device_memory(int device, size_t * free, siz
 GGML_API bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size);
 GGML_API void ggml_backend_cuda_unregister_host_buffer(void * buffer);
 
-GGML_API void ggml_backend_cuda_log_set_callback(ggml_log_callback log_callback, void * user_data);
-
 GGML_API ggml_backend_reg_t ggml_backend_cuda_reg(void);
 
 #ifdef  __cplusplus

ggml/include/ggml-metal.h

@@ -39,8 +39,6 @@ extern "C" {
 // user-code should use only these functions
 //
 
-GGML_API void ggml_backend_metal_log_set_callback(ggml_log_callback log_callback, void * user_data);
-
 GGML_API ggml_backend_t ggml_backend_metal_init(void);
 
 GGML_API bool ggml_backend_is_metal(ggml_backend_t backend);

ggml/include/ggml.h

@@ -2174,6 +2174,10 @@ extern "C" {
     typedef void (*ggml_opt_callback)(void * data, int accum_step, float * sched, bool * cancel);
     typedef void (*ggml_log_callback)(enum ggml_log_level level, const char * text, void * user_data);
 
+    // Set callback for all future logging events.
+    // If this is not called, or NULL is supplied, everything is output on stderr.
+    GGML_API void ggml_log_set(ggml_log_callback log_callback, void * user_data);
+
     // optimization parameters
     //
     //   see ggml.c (ggml_opt_default_params) for default values