Abstract: This disclosure provides an instruction generation method, a data processing method, and an electronic device. In the instruction generation method, a hardware parameter of a neural network processor performing data processing is first determined; a data parameter of to-be-processed data is determined, where the data parameter includes a data type and a quantity of data bits of the to-be-processed data, and the data type includes a signed integer and an unsigned integer; and an instruction for data processing that is executable by the neural network processor is generated based on the hardware parameter and the data parameter. According to this disclosure, the hardware parameter of the neural network processor and the data parameter of the to-be-processed data are comprehensively considered when generating the instruction for data processing that is executable by the neural network processor.

Abstract: Disclosed is a method for convolution calculation in a neural network, comprising: reading an input feature map, depthwise convolution kernels and pointwise convolution kernels from a dynamic random access memory (DRAM); performing depthwise convolution calculations and pointwise convolution calculations by depthwise convolution calculation units and pointwise convolution calculation units, according to the input feature map, the depthwise convolution kernels and the pointwise convolution kernels to obtain output feature values of a first predetermined number p of points on all pointwise convolution output channels; storing the output feature values of a first predetermined number p of points on all pointwise convolution output channels into an on-chip memory; and repeating above operation to obtain output feature values of all points on all point wise convolution output channels. Therefore, the storage space for storing intermediate results may be reduced.

Abstract: Systems and methods of data processing are provided. The method comprises receiving an input data to be processed by a series of operations, identifying a first operation from the series of operations, selecting at least one second operation from the series of operations to be grouped with the first operation based at least in part on an amount of an input data and an output data of the grouped operations and the capacity of the memory unit, and processing a portion of the input data of the grouped operations. An efficiency of the series of data operations can be improved by ensuring the input data and output data of any data operations are both stored in the memory unit.

Abstract: An instruction generation method and apparatus for a neural network accelerator, and an electronic device are disclosed.

Abstract: An epoxy conductive paste is disclosed, based on 100 parts by total mass, comprising the following raw material components: 30˜81 parts of conductive particles, 16˜30 parts of epoxy, 0.2˜3 parts of acrylic, 1˜15 parts of reactive diluent, 1˜15 parts of toughening agent, 0.4˜5 parts of silane coupling agent, and 0.4˜5 parts of cationic curing agent; wherein, the conductive particles include conductive particles with a three-dimensional dendritic microstructure. The conductive paste of the disclosure has the characteristics of good conductivity, short curing time, strong adhesion, and capability for long-term operation at room temperature.

Abstract: The present disclosure provides a method and apparatus for deconvolving feature data using convolution hardware. The method includes: reading a feature map and deconvolution kernel into on-chip memory, and padding zeroes to the feature map; determining convolution kernels based on the deconvolution kernel; removing a row and/or column of each convolution kernel whose elements all are invalid weights, to obtain an optimized convolution kernel, and removing a corresponding row and/or column in the zero-padded feature map to obtain an corresponding optimized feature map; convolving each optimized convolution kernel with corresponding optimized feature map using the multiply-add array, to obtain convolutional outputs; and interleaving and synthesizing the convolutional outputs to obtain an interleaving synthetic output including at least a deconvolutional output corresponding to the feature map and deconvolution kernel.

Abstract: A processing method includes: obtaining an input feature map; processing the input feature map by using a dilated convolution layer of the convolutional neural network, to obtain a plurality of local feature maps; obtaining a plurality of local output feature maps by performing zero padding on the plurality of local feature maps performing convolution processing on the plurality of zero-padded local feature maps; and fusing the plurality of local output feature maps, to obtain an output feature map processed by the dilated convolution layer. A plurality of consecutive local feature maps can be split from the input feature map. The local feature map can be performed with convolution processing by using a compact convolution kernel. Performing dilated convolution processing on the input feature map under a premise of not increasing computational complexity overcomes limitation of holes on a dilated convolution algorithm, and can realize data reuse between adjacent sliding windows.

Abstract: Disclosed are a method and an apparatus for performing an operation of a convolutional layer in a convolutional neural network.

Abstract: Disclosed are a feature extraction method and apparatus for a three-dimensional feature map, a storage medium, and an electronic device. The method includes: determining an overlay parameter based on depth information of a three-dimensional feature map to be processed; decomposing the three-dimensional feature map into a plurality of target two-dimensional feature maps based on the depth information and the overlay parameter; performing two-dimensional convolution processing on each of the plurality of target two-dimensional feature maps to obtain a plurality of initial feature maps; and determining a target feature map corresponding to the three-dimensional feature map based on the plurality of initial feature maps.

Abstract: Disclosed is a method for convolution calculation in a neural network, comprising: reading an input feature map, depthwise convolution kernels and pointwise convolution kernels from a dynamitic random access memory (DRAM); performing depthwise convolution calculations and pointwise convolution calculations according to the input feature map, the depthwise convolution kernels and the pointwise convolution kernels to obtain output feature values of a first predetermined number p of points on all pointwise convolution output channels; storing the output feature values of a first predetermined number p of points on all pointwise convolution output channels into an on-chip memory, wherein the first predetermined number p is determined according to at least one of available space in the on-chip memory, a number of the depthwise convolution calculation units, and width, height and channel dimensions of the input feature map; and repeating the above operation obtain output feature values of all points on all pointwis

Abstract: A method and an apparatus for adapting feature data in a convolutional neural network. The method includes selecting a plurality of consecutive layers; determining an expected number of subdata blocks and a layout position, width and height of each subdata block in an output feature data of a last layer; determining, for each current layer, a layout position, width, and height of each subdata block of an input feature data for the current layer according to the layout position, width, and height of each subdata block of the output feature data for the current layer; determining an actual position of each subdata block of the input feature data for a first layer in the input feature data for the first layer; and obtaining the expected number of subdata blocks of the input feature data for the first layer according to the actual position, width and height of each subdata block of the input feature data for the first layer.

Abstract: Disclosed are a method and an apparatus for performing convolution operation on folded feature data. The method comprises: reading the folded feature data provided to a convolution layer and an original convolution kernel from a dynamic random access memory (DRAM); pre-processing the folded feature data and the original convolution kernel; storing the pre-processed folded feature data into a static random-access memory (SRAM); folding the pre-processed original convolution kernel in at least one dimension of width or height according to a folding manner of the folded feature data to generate one or more folded convolution kernels corresponding to the original convolution kernel; storing the one or more folded convolution kernels in the SRAM; and reading the pre-processed folded feature data and the one or more folded convolution kernels from the SRAM into a calculation unit for convolving the pre-processed folded feature data with the one or more folded convolution kernels.

Abstract: An acrylic conductive paste is provided, based on 100 parts by weight, including: 30-84 parts of conductive particles, 15˜45 parts of acrylate, 0.5˜2.5 parts of adhesion promoter, 0.5˜3 parts of initiator. The conductive particles include three-dimensional dendritic conductive particles; and the adhesion promoter is a mixture of a silane coupling agent and a phosphate ester. The conductive paste of the present disclosure has good electrical conductivity, short curing time, strong adhesion, and can be used for a long-time room temperature operation. The present disclosure also provides a method for preparing the above-mentioned acrylic conductive paste, which is convenient for operation and industrial application; at the same time, it shows that the acrylic conductive paste of the present disclosure can be applied to semiconductor components for packaging a semiconductor device.

Abstract: A modified epoxy acrylic resin conductive adhesive is disclosed, based on 100 parts by total mass, including the following components: 49-75 parts of conductive particles, 24-45 parts of modified epoxy propylene resin, 0.5-2.5 parts of silane coupling agent, and 0.5-3.0 parts of initiator. The conductive particles include at least 5% conductive particles with a three-dimensional dendritic microstructure among all the conductive particles. A preparation method and application of the modified epoxy acrylic resin conductive adhesive are disclosed. The modified epoxy acrylic resin conductive adhesive of the present disclosure has advantages in good electrical conductivity, short curing time, strong adhesion, and capability being used for a long-time room temperature operation.

Abstract: Disclosed are a method and an apparatus for performing an operation of a convolutional layer in a convolutional neural network. The method includes reading unfolded-feature-data and an original convolution kernel from DRAM, padding the unfolded-feature-data, folding the padded unfolded-feature-data in at least one dimension folded feature data, storing the folded feature data into a SRAM, folding the original convolution kernel in the at least one dimension to generate one or more folded convolution kernels, storing the one or more folded convolution kernels in the SRAM and reading the folded feature data and the one or more folded convolution kernels from the SRAM into a calculation circuit for performing a convolution operation on the folded feature data by using the one or more folded convolution kernels.

Abstract: Disclosed are a method and an apparatus for adapting parameters of a neural network. The method includes selecting one or more dimensions for a weight parameter of each of at least one layer of the neural network, determining a dimension value and a corresponding target value in each dimension of the weight parameter, and padding the weight parameter in a case where the dimension value in at least one dimension of the weight parameter is less than the corresponding target value, the dimension value in each dimension of the weight parameter after the padding being equal to the corresponding target value.

Abstract: A method and apparatus are disclosed to perform the circular addressing to emulate a virtually unlimited memory space despite the fixed capacity of a physical memory by readdressing the portion of the data that exceeds the pre-defined length of the circular addressing region to another pre-defined address in the circular addressing region. Data segments in a data sample can be loaded and computed with recalculated circular addresses for different applications.

Abstract: A front-side conductive paste for a crystalline silicon solar cell chip is provided. The front-side conductive paste for a crystalline silicon solar cell chip includes, in parts by weight, 80.0-93.0 parts of a metal powder, 6.0-15.0 parts of an organic carrier, and 1.0-5.0 parts of an oxide etching agent. The oxide etching agent contains at least 10-40% of MgO, 0.1-5% of PbO, and 5-30% of Li2O based on 100% by mole, with the molar ratio of MgO:PbO being 10:5˜40:0.1, and the mole ratio of MgO:Li2O being 10:30˜40:5. The metal powder forms good ohmic contact with crystalline silicon substrate during the sintering process of the front-side conductive paste applied overlying an insulation film on the substrate. Finally, a front-side electrode of low contact resistance, good electrical conductivity, and strong adhesion is obtained.

Abstract: A front-side conductive paste for a crystalline silicon solar cell is provided. The front-side conductive paste for a crystalline silicon solar cell includes, in parts by weight, 80.0-93.0 parts of a metal powder, 6.0-15.0 parts of an organic carrier, and 1.0-5.0 parts of an oxide etching agent, where based on 100% by mole of the oxide etching agent, the oxide etching agent includes 15-30% of PbO; 25-40% of TeO2; 8.0-15.0% of Li2O; 9.0-20.0% of SiO2; 5.0-15.0% of Bi2O3; 0.5-10.0% of ZnO; and either one or both of 0.1-10.0% of MgO and 0.1-10.0% of CaO; and no more than 5.0% of an oxide of additional metal elements. The metal powder forms good ohmic contact with crystalline silicon substrate during the sintering process of the front-side conductive paste applied overlying an insulation film on the substrate. Finally, a front-side electrode of low contact resistance, good electrical conductivity, and strong adhesion is obtained.

Abstract: A front-side conductive paste for a crystalline silicon solar cell is provided. The front-side conductive paste for a crystalline silicon solar cell includes, in parts by weight, 80.0-93.0 parts of a metal powder, 6.0-15.0 parts of an organic carrier, and 1.0-5.0 parts of an oxide etching agent, where based on 100% by mole of the oxide etching agent, the oxide etching agent includes 15-30% of PbO; 25-40% of TeO2; 8.0-15.0% of Li2O; 9.0-20.0% of SiO2; 5.0-15.0% of Bi2O3; 0.5-10.0% of ZnO; and either one or both of 0.1-10.0% of MgO and 0.1-10.0% of CaO; and no more than 5.0% of an oxide of additional metal elements. The metal powder forms good ohmic contact with crystalline silicon substrate during the sintering process of the front-side conductive paste applied overlying an insulation film on the substrate. Finally, a front-side electrode of low contact resistance, good electrical conductivity, and strong adhesion is obtained.