24.02.1
|
Go to the documentation of this file.
30 namespace experimental
32 namespace dynamic_fusion
51 //------------------ START KERNEL {{meta_kernel_id}} STORE ---------------------
53 bool x_cond = PARTIAL_N0 != 0 && get_global_id(0) == 0;
55 T_STORE_INDIRECT_WIDTH_SELECT({{DST_DATA_TYPE}}, M0, N0, PARTIAL_N0, {{DST_TENSOR_TYPE}}, {{dst}}, g_ind_0, {{dst}}_stride_y, x_cond, {{src}}, g_dst_indirect_y);
56 //------------------ END KERNEL {{meta_kernel_id}} STORE ---------------------
79 lut[
"meta_kernel_id"] =
id();
80 lut[
"DST_TENSOR_TYPE"] =
"BUFFER";
TagLUT get_tag_lut(const GpuKernelVariableTable &vtable, const ComponentGroup &comp_group) const override
Generate the tag look-up table used to instantiate the component code.
std::string get_name() const override
Generate kernel component name.
std::unordered_map< Tag, TagVal > TagLUT
Tag lookup table.
ClTemplateStore(ComponentId id, const ArgumentPack< ITensorInfo > &tensors)
Constructor.
An interface used by ClTemplateWriter to write source code for a kernel component.
This is a generic class that packs the arguments of an operator.
ArgumentPack< ITensorInfo > tensors() const
Get tensor arguments.
void declare_variables(GpuKernelVariableTable &vtable, const ComponentGroup &comp_group) const override
Declare all variables used by the component in the vtable.
virtual DataType data_type() const =0
Data type used for each element of the tensor.
ComponentId id() const
Get component id.
Contain information required to set up a kernel argument at run time.
#define ARM_COMPUTE_UNUSED(...)
To avoid unused variables warnings.
TensorVariable get_variable(const ITensorInfo *tensor) const
Get the TensorVariable associated with tensor.
int32_t ComponentId
Uniquely identifies a kernel component within a workload.
Copyright (c) 2017-2024 Arm Limited.
A group of gpu kernel components to be fused together PRECONDITIONS:
std::string get_component_code(const ComponentGroup &comp_group) const override
Generate kernel component code template.
A table of all the variables used in the kernel.
void declare_variable(const GpuKernelComponentGroup &comp_group, const ITensorInfo *tensor, GpuKernelArgumentInfo argument_info, const std::string &alias="unnamed")
Declare a TensorVariable for a corresponding tensor info.