50 void depth_concat(
const ITensor *
src, ITensor *
dst,
unsigned int depth_offset,
const Window &window)
53 uint8_t *src_ptr = src->buffer() + src->info()->offset_first_element_in_bytes();
56 uint8_t *dst_ptr = dst->buffer() + dst->info()->offset_first_element_in_bytes() + depth_offset * dst->info()->strides_in_bytes()[2];
58 const auto window_start_x =
static_cast<int>(window.x().start());
59 const auto window_end_x =
static_cast<int>(window.x().end());
60 const int window_step_x = 16 / dst->info()->element_size();
64 win.set(
Window::DimZ, Window::Dimension(0, src->info()->tensor_shape().z(), 1));
66 Iterator src_it(src, win);
67 Iterator dst_it(dst, win);
70 const UniformQuantizationInfo src_qinfo = src->info()->quantization_info().uniform();
71 const UniformQuantizationInfo dst_qinfo = dst->info()->quantization_info().uniform();
76 const auto in_ptr =
reinterpret_cast<const uint8_t *
>(src_ptr + src_it.offset());
77 const auto out_ptr =
reinterpret_cast<uint8_t *
>(dst_ptr + dst_it.offset());
78 int x = window_start_x;
79 for(; x <= (window_end_x - window_step_x); x += window_step_x)
85 for(; x < window_end_x; ++x)
96 const auto in_ptr =
reinterpret_cast<const int8_t *
>(src_ptr + src_it.offset());
97 const auto out_ptr =
reinterpret_cast<int8_t *
>(dst_ptr + dst_it.offset());
98 int x = window_start_x;
99 for(; x <= (window_end_x - window_step_x); x += window_step_x)
105 for(; x < window_end_x; ++x)
116 const auto in_ptr =
reinterpret_cast<const T *
>(src_ptr + src_it.offset());
117 const auto out_ptr =
reinterpret_cast<T *
>(dst_ptr + dst_it.offset());
118 int x = window_start_x;
119 for(; x <= (window_end_x - window_step_x); x += window_step_x)
124 for(; x < window_end_x; ++x)
126 *(out_ptr + x) = *(in_ptr + x);
133 Status validate_arguments(
const ITensorInfo *
input,
unsigned int depth_offset,
const ITensorInfo *output)
155 _depth_offset = depth_offset;
160 _func = &depth_concat<uint8_t>;
163 _func = &depth_concat<int8_t>;
166 _func = &depth_concat<uint16_t>;
169 _func = &depth_concat<uint32_t>;
177 ICpuKernel::configure(win);
181 unsigned int depth_offset,
203 return "CpuConcatenateDepthKernel";
Window calculate_max_window(const ValidRegion &valid_region, const Steps &steps, bool skip_border, BorderSize border_size)
const Window & window() const
The maximum window the kernel can be executed on.
void run_op(ITensorPack &tensors, const Window &window, const ThreadInfo &info) override
Execute the kernel on the passed window.
float dequantize_qasymm8(uint8_t value, const INFO_TYPE &qinfo)
Dequantize a value given an unsigned 8-bit asymmetric quantization scheme.
const char * name() const override
Name of the kernel.
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.
float32x4x2_t vdequantize(const uint8x8_t &qv, const UniformQuantizationInfo &qi)
Dequantize a neon vector holding 8 quantized values.
uint8x16_t vloadq(const uint8_t *ptr)
#define ARM_COMPUTE_RETURN_ON_ERROR(status)
Checks if a status contains an error and returns it.
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.
Store the tensor's metadata.
#define ARM_COMPUTE_ERROR_THROW_ON(status)
#define ARM_COMPUTE_RETURN_ERROR_ON(cond)
If the condition is true, an error is returned.
SimpleTensor< float > src
Copyright (c) 2017-2021 Arm Limited.
1 channel, 1 F16 per channel
#define ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(...)
void configure(const ITensorInfo *src, unsigned int depth_offset, ITensorInfo *dst)
Configure kernel for a given list of arguments.
const ITensor * get_const_tensor(int id) const
Get constant tensor of a given id.
static constexpr size_t DimX
Alias for dimension 0 also known as X dimension.
#define ARM_COMPUTE_UNUSED(...)
To avoid unused variables warnings.
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.
quantized, asymmetric fixed-point 8-bit number unsigned
Class to describe a number of elements in each dimension.
#define ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(k)
static constexpr size_t DimY
Alias for dimension 1 also known as Y dimension.
ScaleKernelInfo info(interpolation_policy, default_border_mode, PixelValue(), sampling_policy, false)
ITensor * get_tensor(int id)
Get tensor of a given id from the pac.
Information about executing thread and CPU.
#define ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_SHAPES(...)
static constexpr size_t DimZ
Alias for dimension 2 also known as Z dimension.
#define ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(...)
#define ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(t, c,...)
uint8x8_t vquantize(const float32x4x2_t &qv, const UniformQuantizationInfo &qi)
Quantize a neon vector holding 8 floating point values.
void vstore(uint8_t *ptr, uint8x8_t val)
float dequantize_qasymm8_signed(int8_t value, const INFO_TYPE &qinfo)
Dequantize a value given a signed 8-bit asymmetric quantization scheme.
#define ARM_COMPUTE_ERROR_ON_NULLPTR(...)
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...
quantized, asymmetric fixed-point 8-bit number signed
Includes all wrapper headers at once.
int8x8_t vquantize_signed(const float32x4x2_t &qv, const UniformQuantizationInfo &qi)
Quantize a neon vector holding 8 floating point values.
static Status validate(const ITensorInfo *src, unsigned int depth_offset, const ITensorInfo *dst)
Static function to check if given info will lead to a valid configuration.
DataType
Available data types.
Describe a multidimensional execution window.
#define ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(f, s)