Concatenation Functions
-
void riscv_concatenation_s8_w(const int8_t *input, const uint16_t input_x, const uint16_t input_y, const uint16_t input_z, const uint16_t input_w, int8_t *output, const uint32_t offset_w)
-
void riscv_concatenation_s8_x(const int8_t *input, const uint16_t input_x, const uint16_t input_y, const uint16_t input_z, const uint16_t input_w, int8_t *output, const uint16_t output_x, const uint32_t offset_x)
-
void riscv_concatenation_s8_y(const int8_t *input, const uint16_t input_x, const uint16_t input_y, const uint16_t input_z, const uint16_t input_w, int8_t *output, const uint16_t output_y, const uint32_t offset_y)
-
void riscv_concatenation_s8_z(const int8_t *input, const uint16_t input_x, const uint16_t input_y, const uint16_t input_z, const uint16_t input_w, int8_t *output, const uint16_t output_z, const uint32_t offset_z)
- group Concatenation
Functions
-
void riscv_concatenation_s8_w(const int8_t *input, const uint16_t input_x, const uint16_t input_y, const uint16_t input_z, const uint16_t input_w, int8_t *output, const uint32_t offset_w)
int8/uint8 concatenation function to be used for concatenating N-tensors along the W axis (Batch size) This function should be called for each input tensor to concatenate. The argument offset_w will be used to store the input tensor in the correct position in the output tensor
i.e. offset_w = 0 for(i = 0 i < num_input_tensors; ++i) { riscv_concatenation_s8_w(&input[i], …, &output, …, …, offset_w) offset_w += input_w[i] }
This function assumes that the output tensor has:
The same width of the input tensor
The same height of the input tensor
The same number o channels of the input tensor
Unless specified otherwise, arguments are mandatory.
Note
This function, data layout independent, can be used to concatenate either int8 or uint8 tensors because it does not involve any arithmetic operation
- Parameters
input – [in] Pointer to input tensor
input_x – [in] Width of input tensor
input_y – [in] Height of input tensor
input_z – [in] Channels in input tensor
input_w – [in] Batch size in input tensor
output – [out] Pointer to output tensor. Expected to be at least input_x * input_y * input_z * input_w bytes.
offset_w – [in] The offset on the W axis to start concatenating the input tensor It is user responsibility to provide the correct value
-
void riscv_concatenation_s8_x(const int8_t *input, const uint16_t input_x, const uint16_t input_y, const uint16_t input_z, const uint16_t input_w, int8_t *output, const uint16_t output_x, const uint32_t offset_x)
int8/uint8 concatenation function to be used for concatenating N-tensors along the X axis This function should be called for each input tensor to concatenate. The argument offset_x will be used to store the input tensor in the correct position in the output tensor
i.e. offset_x = 0 for(i = 0 i < num_input_tensors; ++i) { riscv_concatenation_s8_x(&input[i], …, &output, …, …, offset_x) offset_x += input_x[i] }
This function assumes that the output tensor has:
The same height of the input tensor
The same number of channels of the input tensor
The same batch size of the input tensor
Unless specified otherwise, arguments are mandatory.
Input constraints offset_x is less than output_x
Note
This function, data layout independent, can be used to concatenate either int8 or uint8 tensors because it does not involve any arithmetic operation
- Parameters
input – [in] Pointer to input tensor. Input tensor must not overlap with the output tensor.
input_x – [in] Width of input tensor
input_y – [in] Height of input tensor
input_z – [in] Channels in input tensor
input_w – [in] Batch size in input tensor
output – [out] Pointer to output tensor. Expected to be at least (input_x * input_y * input_z * input_w) + offset_x bytes.
output_x – [in] Width of output tensor
offset_x – [in] The offset (in number of elements) on the X axis to start concatenating the input tensor It is user responsibility to provide the correct value
-
void riscv_concatenation_s8_y(const int8_t *input, const uint16_t input_x, const uint16_t input_y, const uint16_t input_z, const uint16_t input_w, int8_t *output, const uint16_t output_y, const uint32_t offset_y)
int8/uint8 concatenation function to be used for concatenating N-tensors along the Y axis This function should be called for each input tensor to concatenate. The argument offset_y will be used to store the input tensor in the correct position in the output tensor
i.e. offset_y = 0 for(i = 0 i < num_input_tensors; ++i) { riscv_concatenation_s8_y(&input[i], …, &output, …, …, offset_y) offset_y += input_y[i] }
This function assumes that the output tensor has:
The same width of the input tensor
The same number of channels of the input tensor
The same batch size of the input tensor
Unless specified otherwise, arguments are mandatory.
Input constraints offset_y is less than output_y
Note
This function, data layout independent, can be used to concatenate either int8 or uint8 tensors because it does not involve any arithmetic operation
- Parameters
input – [in] Pointer to input tensor. Input tensor must not overlap with the output tensor.
input_x – [in] Width of input tensor
input_y – [in] Height of input tensor
input_z – [in] Channels in input tensor
input_w – [in] Batch size in input tensor
output – [out] Pointer to output tensor. Expected to be at least (input_z * input_w * input_x * input_y) + offset_y bytes.
output_y – [in] Height of output tensor
offset_y – [in] The offset on the Y axis to start concatenating the input tensor It is user responsibility to provide the correct value
-
void riscv_concatenation_s8_z(const int8_t *input, const uint16_t input_x, const uint16_t input_y, const uint16_t input_z, const uint16_t input_w, int8_t *output, const uint16_t output_z, const uint32_t offset_z)
int8/uint8 concatenation function to be used for concatenating N-tensors along the Z axis This function should be called for each input tensor to concatenate. The argument offset_z will be used to store the input tensor in the correct position in the output tensor
i.e. offset_z = 0 for(i = 0 i < num_input_tensors; ++i) { riscv_concatenation_s8_z(&input[i], …, &output, …, …, offset_z) offset_z += input_z[i] }
This function assumes that the output tensor has:
The same width of the input tensor
The same height of the input tensor
The same batch size of the input tensor
Unless specified otherwise, arguments are mandatory.
Input constraints offset_z is less than output_z
Note
This function, data layout independent, can be used to concatenate either int8 or uint8 tensors because it does not involve any arithmetic operation
- Parameters
input – [in] Pointer to input tensor. Input tensor must not overlap with output tensor.
input_x – [in] Width of input tensor
input_y – [in] Height of input tensor
input_z – [in] Channels in input tensor
input_w – [in] Batch size in input tensor
output – [out] Pointer to output tensor. Expected to be at least (input_x * input_y * input_z * input_w) + offset_z bytes.
output_z – [in] Channels in output tensor
offset_z – [in] The offset on the Z axis to start concatenating the input tensor It is user responsibility to provide the correct value
-
void riscv_concatenation_s8_w(const int8_t *input, const uint16_t input_x, const uint16_t input_y, const uint16_t input_z, const uint16_t input_w, int8_t *output, const uint32_t offset_w)