Deep neural network accelerator including lookup table based bit-serial processing elements

A deep neural network accelerator comprising:

• deep neural network cores each configured to generate an output feature map based on an input feature map and a weight;
• and
• an aggregation core configured to receive the output feature map from each of the deep neural network cores and to accumulate the received output feature map to generate a final output feature map,
• wherein each of the deep neural network cores includes:
• a weight memory configured to store the weight;
• feature loaders each configured to store input features being a portion of the input feature map;
• and
• processing elements each configured to, in a current operation:
• receive the input features from one of the feature loaders, the input features comprising at least one input feature utilized in a previous operation prior to the current operation;
• obtain weight values;
• generate results according to the weight values by applying the weight values to the input features including the at least one input feature utilized in the previous operation;
• receive, by at least one lookup table bundle, at least one target weight corresponding to the input features;
• and
• generate output features by:
• selecting a target result corresponding to the received at least one target weight from among the results;
• and
• accumulating the target result,
• wherein, based on the obtained weight values having a 1-bit precision, each of the at least one input feature utilized in the previous operation are utilized in the current operation,
• wherein, based on the obtained weight values having a precision that is greater than 1-bit precision, fewer than each of the at least one input feature utilized in the previous operation are utilized in the current operation, and
• wherein each of the processing elements includes:
• a first lookup table bundle configured to generate first results corresponding to first features of the input features, to receive a first target weight corresponding to the first features from the weight memory, and to output a first target result corresponding to the first target weight from among the first results;
• a second lookup table bundle configured to generate second results corresponding to second features of the input features, to receive a second target weight corresponding to the second features from the weight memory, and to output a second target result corresponding to the second target weight from among the second results;
• and
• an accumulator configured to accumulate the first target result and the second target result to generate the output features.