24.04
|
Internal implementation of workload context. More...
#include <GpuWorkloadContextImpl.h>
Public Member Functions | |
Impl (GpuLanguage gpu_language, CLCompileContext *cl_compile_ctx) | |
Constructor. More... | |
Impl (Impl &)=default | |
Copy constructor. More... | |
Impl & | operator= (Impl &)=default |
Assignment. More... | |
GpuLanguage | gpu_language () const |
Get target GPU language. More... | |
const CLCompileContext * | cl_compile_context () const |
Get CL compile context. More... | |
const MemoryDescriptorMap & | mem_map () const |
Get memory descriptor registry. More... | |
void | register_user_tensor (std::unique_ptr< TensorInfo > &&tensor_info) |
Set a new ID and register the user tensor info. More... | |
ITensorInfo * | create_virtual_tensor () |
Create a virtual (see MemoryType) tensor info and save it. More... | |
ITensorInfo * | create_auxiliary_tensor (const ITensorInfo &tensor_info) |
Create an auxiliary (see MemoryType) tensor info and save it. More... | |
ITensorInfo * | get_tensor_info (ITensorInfo::Id id) |
Get tensor info created by this context, from id. More... | |
const ITensorInfo * | get_tensor_info (ITensorInfo::Id id) const |
Get tensor info created by this context, from id. More... | |
Internal implementation of workload context.
Definition at line 40 of file GpuWorkloadContextImpl.h.
Impl | ( | GpuLanguage | gpu_language, |
CLCompileContext * | cl_compile_ctx | ||
) |
Constructor.
[in] | gpu_language | Target GPU language. |
[in] | cl_compile_ctx | CL compile context. |
Definition at line 78 of file GpuWorkloadContext.cpp.
const CLCompileContext * cl_compile_context | ( | ) | const |
Get CL compile context.
Definition at line 92 of file GpuWorkloadContext.cpp.
ITensorInfo * create_auxiliary_tensor | ( | const ITensorInfo & | tensor_info | ) |
Create an auxiliary (see MemoryType) tensor info and save it.
[in] | tensor_info | ITensorInfo to copy from |
Definition at line 125 of file GpuWorkloadContext.cpp.
References arm_compute::experimental::dynamic_fusion::Auxiliary, and tensor_info.
Referenced by GpuWorkloadSketch::Implementation::create_auxiliary_tensor().
ITensorInfo * create_virtual_tensor | ( | ) |
Create a virtual (see MemoryType) tensor info and save it.
Definition at line 115 of file GpuWorkloadContext.cpp.
References tensor_info, and arm_compute::experimental::dynamic_fusion::Virtual.
Referenced by GpuWorkloadSketch::Implementation::create_virtual_tensor().
ITensorInfo * get_tensor_info | ( | ITensorInfo::Id | id | ) |
Get tensor info created by this context, from id.
Definition at line 135 of file GpuWorkloadContext.cpp.
Referenced by GpuWorkloadSketch::Implementation::get_tensor_info().
const ITensorInfo * get_tensor_info | ( | ITensorInfo::Id | id | ) | const |
Get tensor info created by this context, from id.
Definition at line 140 of file GpuWorkloadContext.cpp.
GpuLanguage gpu_language | ( | ) | const |
Get target GPU language.
Definition at line 87 of file GpuWorkloadContext.cpp.
const MemoryDescriptorMap & mem_map | ( | ) | const |
Get memory descriptor registry.
Definition at line 97 of file GpuWorkloadContext.cpp.
Referenced by GpuWorkloadSketch::Implementation::generate_source_code().
void register_user_tensor | ( | std::unique_ptr< TensorInfo > && | tensor_info | ) |
Set a new ID and register the user tensor info.
The ownership of the tensor info object will be transfered to this context object.
[in] | tensor_info | The tensor info to be registered. |
Definition at line 102 of file GpuWorkloadContext.cpp.
References ARM_COMPUTE_ERROR_ON, tensor_info, and arm_compute::experimental::dynamic_fusion::User.