24.02.1
|
Go to the documentation of this file.
48 void batch_concat(
const ITensor *
src, ITensor *
dst,
unsigned int batch_offset,
const Window &window)
51 uint8_t *src_ptr =
src->buffer() +
src->info()->offset_first_element_in_bytes();
54 uint8_t *dst_ptr =
dst->buffer() +
dst->info()->offset_first_element_in_bytes() +
55 batch_offset *
dst->info()->strides_in_bytes()[3];
57 const auto window_start_x =
static_cast<int>(window.x().start());
58 const auto window_end_x =
static_cast<int>(window.x().end());
59 const int window_step_x = 16 /
dst->info()->element_size();
63 win.set(3, Window::Dimension(0,
src->info()->tensor_shape()[3], 1));
65 Iterator src_it(
src, win);
66 Iterator dst_it(
dst, win);
69 const UniformQuantizationInfo src_qinfo =
src->info()->quantization_info().uniform();
70 const UniformQuantizationInfo dst_qinfo =
dst->info()->quantization_info().uniform();
75 [&](
const Coordinates &)
77 const auto in_ptr =
reinterpret_cast<const uint8_t *
>(src_ptr + src_it.offset());
78 const auto out_ptr =
reinterpret_cast<uint8_t *
>(dst_ptr + dst_it.offset());
80 int x = window_start_x;
81 for (; x <= (window_end_x - window_step_x); x += window_step_x)
87 for (; x < window_end_x; ++x)
98 [&](
const Coordinates &)
100 const auto in_ptr =
reinterpret_cast<const int8_t *
>(src_ptr + src_it.offset());
101 const auto out_ptr =
reinterpret_cast<int8_t *
>(dst_ptr + dst_it.offset());
102 int x = window_start_x;
103 for (; x <= (window_end_x - window_step_x); x += window_step_x)
109 for (; x < window_end_x; ++x)
121 [&](
const Coordinates &)
123 const auto in_ptr =
reinterpret_cast<const T *
>(src_ptr + src_it.offset());
124 const auto out_ptr =
reinterpret_cast<T *
>(dst_ptr + dst_it.offset());
126 int x = window_start_x;
127 for (; x <= (window_end_x - window_step_x); x += window_step_x)
133 for (; x < window_end_x; ++x)
135 *(out_ptr + x) = *(in_ptr + x);
165 _batch_offset = batch_offset;
167 switch (
src->data_type())
173 _func = &batch_concat<uint8_t>;
178 _func = &batch_concat<uint16_t>;
183 _func = &batch_concat<uint32_t>;
191 ICpuKernel::configure(win);
195 unsigned int batch_offset,
215 return "CpuConcatenateBatchKernel";
Class to describe a number of elements in each dimension.
float32x4x2_t vdequantize(const uint8x8_t &qv, const UniformQuantizationInfo &qi)
Dequantize a neon vector holding 8 quantized values.
float dequantize_qasymm8(uint8_t value, const INFO_TYPE &qinfo)
Dequantize a value given an unsigned 8-bit asymmetric quantization scheme.
SimpleTensor< float > src
Window calculate_max_window(const ValidRegion &valid_region, const Steps &steps, bool skip_border, BorderSize border_size)
@ QASYMM8
quantized, asymmetric fixed-point 8-bit number unsigned
@ U16
unsigned 16-bit number
Status validate_arguments(const ITensorInfo *src, const ITensorInfo *weights, const ITensorInfo *dst, const PadStrideInfo &conv_info)
void configure(const ITensorInfo *src, unsigned int batch_offset, ITensorInfo *dst)
Configure kernel for a given list of arguments.
#define ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(k)
#define ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_SHAPES(...)
static constexpr size_t DimX
Alias for dimension 0 also known as X dimension.
uint8_t quantize_qasymm8(float value, const INFO_TYPE &qinfo, RoundingPolicy rounding_policy=RoundingPolicy::TO_NEAREST_UP)
Quantize a value given an unsigned 8-bit asymmetric quantization scheme.
#define ARM_COMPUTE_ERROR(msg)
Print the given message then throw an std::runtime_error.
ITensor * get_tensor(int id)
Get tensor of a given id from the pac.
uint8x16_t vloadq(const uint8_t *ptr)
Includes all wrapper headers at once.
#define ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(...)
#define ARM_COMPUTE_RETURN_ON_ERROR(status)
Checks if a status contains an error and returns it.
#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.
const ITensor * get_const_tensor(int id) const
Get constant tensor of a given id.
const char * name() const override
Name of the kernel.
#define ARM_COMPUTE_ERROR_THROW_ON(status)
@ U32
unsigned 32-bit number
void execute_window_loop(const Window &w, L &&lambda_function, Ts &&...iterators)
Iterate through the passed window, automatically adjusting the iterators and calling the lambda_funct...
int8_t quantize_qasymm8_signed(float value, const INFO_TYPE &qinfo, RoundingPolicy rounding_policy=RoundingPolicy::TO_NEAREST_UP)
Quantize a value given a signed 8-bit asymmetric quantization scheme.
#define ARM_COMPUTE_RETURN_ERROR_ON(cond)
If the condition is true, an error is returned.
@ U8
unsigned 8-bit number
@ S16
signed 16-bit number
@ QASYMM8_SIGNED
quantized, asymmetric fixed-point 8-bit number signed
#define ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(f, s)
void run_op(ITensorPack &tensors, const Window &window, const ThreadInfo &info) override
Execute the kernel on the passed window.
#define ARM_COMPUTE_UNUSED(...)
To avoid unused variables warnings.
static constexpr size_t DimY
Alias for dimension 1 also known as Y dimension.
const Window & window() const
The maximum window the kernel can be executed on.
Information about executing thread and CPU.
void vstore(uint8_t *ptr, uint8x8_t val)
Describe a multidimensional execution window.
float dequantize_qasymm8_signed(int8_t value, const INFO_TYPE &qinfo)
Dequantize a value given a signed 8-bit asymmetric quantization scheme.
uint8x8_t vquantize(const float32x4x2_t &qv, const UniformQuantizationInfo &qi)
Quantize a neon vector holding 8 floating point values.
Copyright (c) 2017-2024 Arm Limited.
@ F16
16-bit floating-point number
static constexpr size_t DimZ
Alias for dimension 2 also known as Z dimension.
@ S32
signed 32-bit number
#define ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(...)
Store the tensor's metadata.
@ F32
32-bit floating-point number
ScaleKernelInfo info(interpolation_policy, default_border_mode, PixelValue(), sampling_policy, false)
static Status validate(const ITensorInfo *src, unsigned int batch_offset, const ITensorInfo *dst)
Static function to check if given info will lead to a valid configuration.
@ UNKNOWN
Unknown data type.
int8x8_t vquantize_signed(const float32x4x2_t &qv, const UniformQuantizationInfo &qi)
Quantize a neon vector holding 8 floating point values.
DataType
Available data types.