US 2022/0035891 A1 · REDUCED RESULT MATRIX

Abstract

Matrix multiple operations may use a reduced result matrix to increase the speed and accuracy of the operation. In one example, each higher precision row/column is decomposed into multiple component rows/columns of the base type that can be combined as weighted sums to form the original higher precision row/column. In another example, the decomposition may be independent for each input matrix and decompose to any multiple of the base type. In another example, the base type for each input matrix could be different. In another example, after decomposition, a matrix operation is performed (e.g. matrix multiply, convolutional layer, or possibly other matrix operation) on decomposed base type input matrices to yield a result matrix that contains components of the higher precision results. The results may be combined together to obtain higher-precision results.

Claims28

Dependent claims

An apparatus comprising:

a memory configured to store a first result;
a processor coupled to the memory, the processor configured to:
decompose a data component into a low first component and a high first component;
perform a first matrix operation on the low first component to generate the first result;
store the first result in the memory;
perform a second matrix operation on the high first component to generate a second result;
and
combine the first result and the second result to generate a final result, wherein the final result is a result of a third matrix operation on the data component.

5 more elements

Dependencies are inferred from the claim text.

Reduced result matrix

Life of this patent

Reduced result matrix

Life of this patent

Abstract

Claims28