Compute Library
 21.02
GCGEMMMatrixAccumulateBiasesKernel Class Reference

Interface to add a bias to each row of the input tensor. More...

#include <GCGEMMMatrixAccumulateBiasesKernel.h>

Collaboration diagram for GCGEMMMatrixAccumulateBiasesKernel:
[legend]

Public Member Functions

 GCGEMMMatrixAccumulateBiasesKernel ()
 Default constructor. More...
 
 GCGEMMMatrixAccumulateBiasesKernel (const GCGEMMMatrixAccumulateBiasesKernel &)=delete
 Prevent instances of this class from being copied (As this class contains pointers) More...
 
GCGEMMMatrixAccumulateBiasesKerneloperator= (const GCGEMMMatrixAccumulateBiasesKernel &)=delete
 Prevent instances of this class from being copied (As this class contains pointers) More...
 
 GCGEMMMatrixAccumulateBiasesKernel (GCGEMMMatrixAccumulateBiasesKernel &&)=default
 Allow instances of this class to be moved. More...
 
GCGEMMMatrixAccumulateBiasesKerneloperator= (GCGEMMMatrixAccumulateBiasesKernel &&)=default
 Allow instances of this class to be moved. More...
 
void configure (IGCTensor *accum, const IGCTensor *biases)
 Set the accumulate buffer and the biases of the kernel. More...
 
void run (const Window &window) override
 Enqueue the OpenGL ES shader to process the given window. More...
 
- Public Member Functions inherited from IGCKernel
 IGCKernel ()
 Constructor. More...
 
GCKernelkernel ()
 Returns a reference to the GLES kernel of this object. More...
 
void add_1D_tensor_argument (unsigned int &idx, const IGCTensor *tensor, const unsigned int binding_point, const Window &window)
 Add the passed 1D tensor's parameters to the object's kernel's arguments starting from the index idx. More...
 
void add_2D_tensor_argument (unsigned int &idx, const IGCTensor *tensor, const unsigned int binding_point, const Window &window)
 Add the passed 2D tensor's parameters to the object's kernel's arguments starting from the index idx. More...
 
void add_3D_tensor_argument (unsigned int &idx, const IGCTensor *tensor, const unsigned int binding_point, const Window &window)
 Add the passed 3D tensor's parameters to the object's kernel's arguments starting from the index idx. More...
 
unsigned int num_arguments_per_1D_tensor () const
 Returns the number of arguments enqueued per 1D tensor object. More...
 
unsigned int num_arguments_per_2D_tensor () const
 Returns the number of arguments enqueued per 2D tensor object. More...
 
unsigned int num_arguments_per_3D_tensor () const
 Returns the number of arguments enqueued per 3D tensor object. More...
 
void set_lws_hint (gles::NDRange &lws_hint)
 Set the Local-Workgroup-Size hint. More...
 
void set_target (GPUTarget target)
 Set the targeted GPU architecture. More...
 
GPUTarget get_target () const
 Get the targeted GPU architecture. More...
 
- Public Member Functions inherited from IKernel
 IKernel ()
 Constructor. More...
 
virtual ~IKernel ()=default
 Destructor. More...
 
virtual bool is_parallelisable () const
 Indicates whether or not the kernel is parallelisable. More...
 
virtual BorderSize border_size () const
 The size of the border for that kernel. More...
 
const Windowwindow () const
 The maximum window the kernel can be executed on. More...
 

Detailed Description

Interface to add a bias to each row of the input tensor.

Definition at line 34 of file GCGEMMMatrixAccumulateBiasesKernel.h.

Constructor & Destructor Documentation

◆ GCGEMMMatrixAccumulateBiasesKernel() [1/3]

Default constructor.

Definition at line 42 of file GCGEMMMatrixAccumulateBiasesKernel.cpp.

References arm_compute::utils::cast::U.

43  : _accum(nullptr), _biases(nullptr), _lws(gles::NDRange(1U, 1U, 1U))
44 {
45 }
Class interface for specifying NDRange values.
Definition: OpenGLES.h:53

◆ GCGEMMMatrixAccumulateBiasesKernel() [2/3]

Prevent instances of this class from being copied (As this class contains pointers)

◆ GCGEMMMatrixAccumulateBiasesKernel() [3/3]

Allow instances of this class to be moved.

Member Function Documentation

◆ configure()

void configure ( IGCTensor accum,
const IGCTensor biases 
)

Set the accumulate buffer and the biases of the kernel.

Parameters
[in,out]accumThe accumulate tensor to convert. Data types supported: F16/F32
[in]biasesThe shared biases tensor to append. It must be 1D tensor. Data types supported: Same as input

Definition at line 47 of file GCGEMMMatrixAccumulateBiasesKernel.cpp.

References ARM_COMPUTE_ERROR_ON, ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN, ARM_COMPUTE_ERROR_ON_MISMATCHING_DATA_TYPES, arm_compute::calculate_max_enlarged_window(), arm_compute::ceil_to_multiple(), GCKernelLibrary::create_kernel(), ITensorInfo::data_type(), ITensorInfo::dimension(), arm_compute::F16, arm_compute::F32, GCKernelLibrary::get(), ITensor::info(), ITensorInfo::num_dimensions(), num_elems_processed_per_iteration, arm_compute::support::cpp11::to_string(), and arm_compute::update_window_and_padding().

Referenced by GCFullyConnectedLayer::configure().

48 {
51  ARM_COMPUTE_ERROR_ON(biases->info()->num_dimensions() != 1);
52 
53  _biases = biases;
54  _accum = accum;
55 
56  std::set<std::string> build_opts;
57  build_opts.emplace("#define LOCAL_SIZE_X " + support::cpp11::to_string(_lws[0]));
58  build_opts.emplace("#define LOCAL_SIZE_Y " + support::cpp11::to_string(_lws[1]));
59  build_opts.emplace("#define LOCAL_SIZE_Z " + support::cpp11::to_string(_lws[2]));
60 
61  // Create kernel
62  build_opts.emplace("#define GEMM_ACCUMULATE_BIASES");
63 
64 #define ACCUM_PROCESS_4X
65 
66 #if defined(ACCUM_PROCESS_4X)
67  build_opts.emplace("#define ACCUM_PROCESS_4X");
68 #elif defined(ACCUM_PROCESS_8X) /* ACCUM_PROCESS_4X */
69  build_opts.emplace("#define ACCUM_PROCESS_8X");
70 #endif /* ACCUM_PROCESS_4X */
71  std::string dt_name = (accum->info()->data_type() == DataType::F32) ? "DATA_TYPE_FP32" : "DATA_TYPE_FP16";
72  build_opts.emplace(("#define " + dt_name));
73 
74  _kernel = GCKernelLibrary::get().create_kernel("gemm_accumulate_biases", build_opts);
75 
76  // Configure kernel window
77  unsigned int num_elems_processed_per_iteration = 1;
78 
79  if(_accum->info()->data_type() == DataType::F32)
80  {
81  num_elems_processed_per_iteration = 16;
82  }
83  else if(_accum->info()->data_type() == DataType::F16)
84  {
85 #if defined(ACCUM_PROCESS_4X)
86  num_elems_processed_per_iteration = 4;
87 #elif defined(ACCUM_PROCESS_8X) /* ACCUM_PROCESS_4X */
88  num_elems_processed_per_iteration = 8;
89 #endif /* ACCUM_PROCESS_4X */
90  }
91 
92  const int accum_width = accum->info()->dimension(0);
93  const int accum_padding_right = ceil_to_multiple(accum_width, num_elems_processed_per_iteration * _lws[0]) - accum_width;
94  BorderSize border = BorderSize(0, accum_padding_right, 0, 0);
95 
96  Window win = calculate_max_enlarged_window(*_accum->info(), Steps(num_elems_processed_per_iteration), border);
97 
98  AccessWindowStatic biases_access(biases->info(), 0, 0, ceil_to_multiple(biases->info()->dimension(0), num_elems_processed_per_iteration * _lws[0]), biases->info()->dimension(1));
99  AccessWindowStatic accum_access(_accum->info(), 0, 0, accum_width + accum_padding_right, _accum->info()->dimension(1));
100 
101  update_window_and_padding(win, biases_access, accum_access);
102 
103  IGCKernel::configure(win);
104 }
virtual size_t num_dimensions() const =0
The number of dimensions of the tensor (rank)
virtual size_t dimension(size_t index) const =0
Return the size of the requested dimension.
Container for 2D border size.
Definition: Types.h:273
std::string to_string(T &&value)
Convert integer and float values to string.
virtual DataType data_type() const =0
Data type used for each element of the tensor.
1 channel, 1 F32 per channel
#define ARM_COMPUTE_ERROR_ON(cond)
If the condition is true then an error message is printed and an exception thrown.
Definition: Error.h:466
1 channel, 1 F16 per channel
Implementation of a static rectangular access pattern.
bool update_window_and_padding(Window &win, Ts &&... patterns)
Update window and padding size for each of the access patterns.
Definition: WindowHelpers.h:46
#define ARM_COMPUTE_ERROR_ON_MISMATCHING_DATA_TYPES(...)
Definition: Validate.h:543
auto ceil_to_multiple(S value, T divisor) -> decltype(((value+divisor - 1)/divisor) *divisor)
Computes the smallest number larger or equal to value that is a multiple of divisor.
Definition: Utils.h:71
Class to describe a number of elements in each dimension.
Definition: Steps.h:40
virtual ITensorInfo * info() const =0
Interface to be implemented by the child class to return the tensor&#39;s metadata.
#define ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(t, c,...)
Definition: Validate.h:790
static GCKernelLibrary & get()
Get the static instance of GCKernelLibrary.
Window calculate_max_enlarged_window(const ValidRegion &valid_region, const Steps &steps, BorderSize border_size)
GCKernel create_kernel(const std::string &shader_name, const StringSet &build_options_set={}) const
Creates a kernel from the kernel library.
unsigned int num_elems_processed_per_iteration
Describe a multidimensional execution window.
Definition: Window.h:39

◆ operator=() [1/2]

Prevent instances of this class from being copied (As this class contains pointers)

◆ operator=() [2/2]

Allow instances of this class to be moved.

◆ run()

void run ( const Window window)
overridevirtual

Enqueue the OpenGL ES shader to process the given window.

Parameters
[in]windowRegion on which to execute the kernel. (Must be a valid region of the window returned by window()).

Implements IGCKernel.

Definition at line 106 of file GCGEMMMatrixAccumulateBiasesKernel.cpp.

References IGCKernel::add_1D_tensor_argument(), IGCKernel::add_2D_tensor_argument(), ARM_COMPUTE_ERROR_ON_MISMATCHING_WINDOWS, ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL, Window::DimY, arm_compute::enqueue(), Window::first_slice_window_2D(), Window::set(), Window::slide_window_slice_2D(), and IKernel::window().

107 {
110 
111  _kernel.use();
112 
113  Window accum_slice = window.first_slice_window_2D();
114 
115  Window biases_slice(accum_slice);
116  biases_slice.set(Window::DimY, Window::Dimension(0, 1, 1));
117 
118  // Run kernel
119  do
120  {
121  // Set arguments
122  unsigned int idx = 0;
123 
124  add_2D_tensor_argument(idx, _accum, 1, accum_slice);
125  add_1D_tensor_argument(idx, _biases, 2, biases_slice);
126  _kernel.update_shader_params();
127 
128  enqueue(*this, accum_slice, _lws);
129  }
130  while(window.slide_window_slice_2D(accum_slice));
131 }
Window first_slice_window_2D() const
First 2D slice of the window.
Definition: Window.h:283
const Window & window() const
The maximum window the kernel can be executed on.
Definition: IKernel.cpp:28
void enqueue(IGCKernel &kernel, const Window &window, const gles::NDRange &lws=gles::NDRange(1U, 1U, 1U))
Add the kernel to the command queue with the given window.
Definition: IGCKernel.cpp:41
Describe one of the image&#39;s dimensions with a start, end and step.
Definition: Window.h:77
bool slide_window_slice_2D(Window &slice) const
Slide the passed 2D window slice.
Definition: Window.h:323
#define ARM_COMPUTE_ERROR_ON_MISMATCHING_WINDOWS(f, w)
Definition: Validate.h:183
void add_2D_tensor_argument(unsigned int &idx, const IGCTensor *tensor, const unsigned int binding_point, const Window &window)
Add the passed 2D tensor&#39;s parameters to the object&#39;s kernel&#39;s arguments starting from the index idx...
Definition: IGCKernel.cpp:127
#define ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(k)
Definition: Validate.h:941
static constexpr size_t DimY
Alias for dimension 1 also known as Y dimension.
Definition: Window.h:45
Describe a multidimensional execution window.
Definition: Window.h:39
void add_1D_tensor_argument(unsigned int &idx, const IGCTensor *tensor, const unsigned int binding_point, const Window &window)
Add the passed 1D tensor&#39;s parameters to the object&#39;s kernel&#39;s arguments starting from the index idx...
Definition: IGCKernel.cpp:122

The documentation for this class was generated from the following files: