Abstract: A computer-implemented method for performing a code generation for calculating a Softmax function of a neural network includes (i) providing a displacement s and a multiplier for the quantized representation of the input tensors of the Softmax function of the neural network, (ii) creating a second lookup table to replace a nested function to calculate the EXP_ON_NEG (MUL_SAT ( )) function from a CMSIS NN library depending on the displacement and the multiplier, wherein element values of an input tensor normalized to a negative value range between 0 and a minimum value are used as arguments, wherein 0 is assigned to a maximum possible element value of the input tensor and the minimum value is assigned to the smallest possible element value of the input tensor, and (iii) implementing an access to the second lookup table in a code generated to calculate the Softmax function, so that it replaces the function call of the EXP_ON_NEG (MUL_SAT ( )) function.

Abstract: A computer-implemented method for performing a code generation for determining a recalibratable code for computing a neural network includes (i) providing a source code for implementing the neural network that does not contain the network parameters defining the neural network, i.e. their configuration and their parameterization, as constants, wherein the source code is configured to access a parameter dataset with predetermined network parameters, in which all parameters necessary for recalibration of the neural network are contained, (ii) compiling the source code so that a program code for the desired hardware environment is generated, (iii) providing the network parameters to be accessible by the program code, and (iv) implementing the generated program code in the hardware environment.

Abstract: A computer-implemented method for performing memory planning for code generation to generate code for computing a neural network in a hardware environment is disclosed.

Abstract: A computer-implemented method for performing memory scheduling for code generation to determine a code for computing a neural network in a hardware environment includes (i) providing successive computation steps of the neural network, wherein a plurality of the computation steps provide for the computation of a weight prefetching layer for which weight prefetching is applicable, (ii) assigning possible weight prefetching to at least a portion of the plurality of weight-prefetching layers in order to obtain multiple different combinations of weight-prefetching layers, wherein the weight prefetching provides for preloading of network parameters into a working memory of the hardware environment for the respective weight-prefetching layer, (iii) performing memory scheduling for the code to be generated for the plurality of different combinations, wherein the respective memory scheduling is carried out for the successive computation steps of the neural network while taking into account the respective combination o

Abstract: A method is for performing memory planning for code generation to determine a code for calculating a neural network for use in a hardware environment. The method includes providing successive calculation steps of the neural network. A size of at least one input data block and at least one output data block is determined for each calculation step. The method further includes determining a maximum overlap area between an input data block and an output data block for each calculation step. The method includes performing memory planning in which the memory area of the respective input data block and output data block is determined in the working memory, depending on the maximum overlap area for each calculation step.

Abstract: A computer-implemented method for performing memory planning for code generation to determine a code for calculating a neural network in a hardware environment includes (i) providing successive calculation steps of layers of the neural network, wherein for each calculation step the size of an input data block, an output data block and, depending on the type of calculation step, one or more model parameter blocks is determined, wherein the one or more model parameter blocks have model parameters for a respective calculation step, (ii) determining a memory planning rule for each specific calculation step that requires the use of model parameters, wherein the rule specifies that the model parameters are loaded into a contiguous memory space in a working memory of the hardware environment, and (iii) performing memory planning, in which the memory space of the respective input data block, output data block, and model parameter block is determined in the working memory for each calculation step, taking into account

Abstract: A computer-implemented method for performing memory planning for a code generation for determining a code for computing a neural network includes (i) providing successive calculation steps of the neural network, wherein for each calculation step the size of an input data block and an output data block is determined, (ii) determining a condition for the memory planning for each specific computing step, in which an input data block in a memory area is at least partially contained in the output data block, wherein the condition indicates that the memory area associated with the output data block of the calculation step preceding the specific calculation step is contained in the memory area of the output data block of the specific calculation step, and (iii) performing memory planning in which the memory area of the respective input data block and output data block in the working memory is determined for each calculation step, taking into account the determined conditions.

Abstract: A method is for parallelized calculation of two convolutions of a filter having first and second receptive fields of the filter on input data. The first and second receptive fields correspond to a filter shifted by one step on the input data. The method includes initializing first and second output variables each having an initial value, and executing a loop for each line of the filter. The loop performs loading a first kernel element of the filter of the line and the corresponding data values to the first kernel element of the first and second receptive fields of the input data.