24.04
|
Go to the documentation of this file.
36 namespace experimental
38 namespace dynamic_fusion
79 const auto op = group.
new_operator(operator_type, tensors);
80 const auto success = group.try_add_operator(op,
true);
Operator new_operator(const GpuOperatorType &operator_type, const ArgumentPack< ITensorInfo > &tensors) const
Create a new operator.
SimpleTensor< float > src
@ Run
Run every time after the first time.
@ Simple
Simple operators are operators that:
#define ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_SHAPES(...)
const GpuOperatorGroup & operator_group() const
Get operator group.
#define ARM_COMPUTE_ERROR(msg)
Print the given message then throw an std::runtime_error.
This is a generic class that packs the arguments of an operator.
const GpuWorkloadContext * gpu_context() const
Get the gpu workload context of this sketch.
#define ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(...)
Implementation & implementation()
Get the implementation.
#define ARM_COMPUTE_ERROR_ON(cond)
If the condition is true then an error message is printed and an exception thrown.
#define ARM_COMPUTE_ERROR_THROW_ON(status)
const GpuKernelComponentGraph & component_graph() const
Get component graph.
Properties common to all kernel component types.
A descriptor of a workload of operators.
#define ARM_COMPUTE_RETURN_ERROR_ON(cond)
If the condition is true, an error is returned.
bool auto_init_if_empty(ITensorInfo &info, const TensorShape &shape, int num_channels, DataType data_type, QuantizationInfo quantization_info=QuantizationInfo())
Auto initialize the tensor info (shape, number of channels and data type) if the current assignment i...
bool is_alloc_tensor(const ITensorInfo *tensor_info)
Tensor should have backing memory.
Provide context necessary for the creation and configuration of a workload e.g.
void add_const_tensor(Id id, const T *tensor)
Add const tensor to the pack.
const Context * context() const
Get workload context.
static Status is_supported_op(const GpuWorkloadContext &context, const ITensorInfo *src, const ITensorInfo *dst)
Check if the operator configuration is supported, irrespective of fusion.
#define ARM_COMPUTE_UNUSED(...)
To avoid unused variables warnings.
KernelProperties & stage(const UnitWorkloadStage &stage)
Describes when a unit workload is run.
GpuOperatorType
Contain properties common to all operator types.
Store the tensor's metadata.
#define ARM_COMPUTE_RETURN_ERROR_ON_MSG(cond, msg)
If the condition is true, an error is returned.
static void create_op(GpuWorkloadSketch &sketch, ITensorInfo *src, ITensorInfo *dst)
Create an operator and fuse it into the workload sketch.
An operator for the sole purpose of validating fusion.
Copyright (c) 2017-2024 Arm Limited.
#define ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(...)
Store the tensor's metadata.
void add_operator(const Operator &op, bool is_output=false)
Add an operator to the group.
#define ARM_COMPUTE_LOG_PARAMS(...)
static Status validate_op(const GpuWorkloadSketch &sketch, const ITensorInfo *src, const ITensorInfo *dst)
Validate the operator and check if the its configuration is supported and if it can be fused into the...