24.02.1
|
Go to the documentation of this file.
47 const ITensorInfo *vector_sum_col,
48 const ITensorInfo *vector_sum_row,
49 const ITensorInfo *
bias,
75 const bool reinterpret_as_3d =
76 mm_result->num_dimensions() > 1 && mm_result->tensor_shape().y() != vector_sum_row->tensor_shape().x();
80 (mm_result->dimension(1) * mm_result->dimension(2)));
86 const unsigned int output_batch_idx = reinterpret_as_3d ? 3 : 2;
88 TensorShape vector_sum_row_shape = vector_sum_row->tensor_shape();
93 "mm_result tensor must have the same number of batches of output tensor");
97 TensorShape vector_sum_col_shape = vector_sum_col->tensor_shape();
98 vector_sum_col_shape.collapse_from(1);
101 vector_sum_col_shape[1] != vector_sum_row_shape[1],
102 "vector_sum_col tensor must have the same number of batches of "
103 "vector_sum_row_shape or the number of batches must be set to 1");
133 const bool reinterpret_as_3d = vector_sum_row !=
nullptr && mm_result->
num_dimensions() > 1 &&
141 build_opts.
add_option(
"-DVEC_SIZE_LEFTOVER=" +
159 std::string
kernel_name(
"gemmlowp_offset_contribution");
169 IClKernel::configure_internal(win);
199 const auto vector_sum_col =
201 const auto vector_sum_row =
227 unsigned int idx = 0;
Class to describe a number of elements in each dimension.
std::string to_string(T &&value)
Convert integer and float values to string.
virtual const TensorShape & tensor_shape() const =0
Size for each dimension of the tensor.
const StringSet & options() const
Gets the current options list set.
Window calculate_max_window(const ValidRegion &valid_region, const Steps &steps, bool skip_border, BorderSize border_size)
Status validate_arguments(const ITensorInfo *src, const ITensorInfo *weights, const ITensorInfo *dst, const PadStrideInfo &conv_info)
#define ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(k)
static constexpr size_t DimX
Alias for dimension 0 also known as X dimension.
Window collapse_if_possible(const Window &full_window, size_t first, size_t last, bool *has_collapsed=nullptr) const
Collapse the dimensions between first and last if possible.
ClGemmLowpOffsetContributionKernel()
ITensor * get_tensor(int id)
Get tensor of a given id from the pac.
std::string upper_string(const std::string &val)
Raise a given string to upper case.
#define ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(t, c,...)
#define ARM_COMPUTE_RETURN_ON_ERROR(status)
Checks if a status contains an error and returns it.
virtual size_t dimension(size_t index) const =0
Return the size of the requested dimension.
#define ARM_COMPUTE_ERROR_ON_NULLPTR(...)
#define ARM_COMPUTE_ERROR_ON(cond)
If the condition is true then an error message is printed and an exception thrown.
void run_op(ITensorPack &tensors, const Window &window, cl::CommandQueue &queue) override
Enqueue the OpenCL kernel to process the given window on the passed OpenCL command queue.
const ITensor * get_const_tensor(int id) const
Get constant tensor of a given id.
void add_2D_tensor_argument_if(bool cond, unsigned int &idx, const ICLTensor *tensor, const Window &window)
Add the passed 2D tensor's parameters to the object's kernel's arguments starting from the index idx ...
#define ARM_COMPUTE_ERROR_THROW_ON(status)
void add_option(std::string option)
Adds option to the existing build option list.
#define ARM_COMPUTE_RETURN_ERROR_ON(cond)
If the condition is true, an error is returned.
cl::Kernel create_kernel(const CLCompileContext &ctx, const std::string &kernel_name, const std::set< std::string > &build_opts=std::set< std::string >())
Creates an opencl kernel using a compile context.
void add_option_if(bool cond, std::string option)
Adds option if a given condition is true;.
#define ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(f, s)
T x() const
Alias to access the size of the first dimension.
bool slide_window_slice_3D(Window &slice) const
Slide the passed 3D window slice.
void collapse_from(size_t start)
Collapse dimensions starting from a given point.
Describe one of the image's dimensions with a start, end and step.
void set(size_t dimension, const Dimension &dim)
Set the values of a given dimension.
static constexpr size_t DimY
Alias for dimension 1 also known as Y dimension.
Window first_slice_window_3D() const
First 3D slice of the window.
void configure(const CLCompileContext &compile_context, const ITensorInfo *mm_result, const ITensorInfo *vector_sum_col, const ITensorInfo *vector_sum_row, const ITensorInfo *bias, int32_t k, int32_t a_offset, int32_t b_offset)
Initialise the kernel's input and output.
const Window & window() const
The maximum window the kernel can be executed on.
void add_3D_tensor_argument(unsigned int &idx, const ICLTensor *tensor, const Window &window)
Add the passed 3D tensor's parameters to the object's kernel's arguments starting from the index idx.
void add_1D_tensor_argument_if(bool cond, unsigned int &idx, const ICLTensor *tensor, const Window &window)
Add the passed 1D tensor's parameters to the object's kernel's arguments starting from the index idx ...
Describe a multidimensional execution window.
@ ELEMENTWISE
Elementwise CL kernel type.
#define ARM_COMPUTE_RETURN_ERROR_ON_MSG(cond, msg)
If the condition is true, an error is returned.
Copyright (c) 2017-2024 Arm Limited.
unsigned int adjust_vec_size(unsigned int vec_size, size_t dim0)
Returns the adjusted vector size in case it is less than the input's first dimension,...
bool has_padding_changed(const std::unordered_map< const ITensorInfo *, PaddingSize > &padding_map)
Check if the previously stored padding info has changed after configuring a kernel.
static constexpr size_t DimZ
Alias for dimension 2 also known as Z dimension.
@ S32
signed 32-bit number
cl::NDRange lws_hint() const
Return the Local-Workgroup-Size hint.
static Status validate(const ITensorInfo *mm_result, const ITensorInfo *vector_sum_col, const ITensorInfo *vector_sum_row, const ITensorInfo *bias, int32_t a_offset, int32_t b_offset)
Static function to check if given info will lead to a valid configuration.
Store the tensor's metadata.
T y() const
Alias to access the size of the second dimension.
unsigned int num_elems_processed_per_iteration
unsigned int num_dimensions() const
Returns the effective dimensionality of the tensor.
std::unordered_map< const ITensorInfo *, PaddingSize > get_padding_info(std::initializer_list< const ITensorInfo * > infos)
Stores padding information before configuring a kernel.
SimpleTensor< T > slice(const SimpleTensor< T > &src, Coordinates starts, Coordinates ends)
virtual size_t num_dimensions() const =0
The number of dimensions of the tensor (rank)
void enqueue(cl::CommandQueue &queue, ICLKernel &kernel, const Window &window, const cl::NDRange &lws_hint=CLKernelLibrary::get().default_ndrange(), bool use_dummy_work_items=false)
Add the kernel to the command queue with the given window.