18052774. MODULO-SPACE PROCESSING IN MULTIPLY-AND-ACCUMULATE UNITS simplified abstract (QUALCOMM Incorporated)

From WikiPatents
Jump to navigation Jump to search

MODULO-SPACE PROCESSING IN MULTIPLY-AND-ACCUMULATE UNITS

Organization Name

QUALCOMM Incorporated

Inventor(s)

Edwin Chongwoo Park of San Diego CA (US)

Ravishankar Sivalingam of San Jose CA (US)

MODULO-SPACE PROCESSING IN MULTIPLY-AND-ACCUMULATE UNITS - A simplified explanation of the abstract

This abstract first appeared for US patent application 18052774 titled 'MODULO-SPACE PROCESSING IN MULTIPLY-AND-ACCUMULATE UNITS

Simplified Explanation

The present disclosure describes techniques and apparatus for performing mathematical operations on processing units using data in the modulo space. Here is a simplified explanation of the abstract:

  • The patent application introduces a method for processing binary-space input using a neural network or other processing system.
  • The binary-space input is converted into a modulo-space input using a set of coprimes, which are numbers that have no common factors other than 1.
  • The modulo-space input is used to generate a modulo-space result through one or more modulo-space multiply-and-accumulate (MAC) units.
  • The modulo-space result is then converted back into a binary-space result.
  • The binary-space result is the final output of the processing operation.

Potential applications of this technology:

  • This technique can be applied in various fields where mathematical operations are performed on processing units, such as machine learning, data analysis, and cryptography.
  • It can enhance the efficiency and accuracy of mathematical computations in these applications.

Problems solved by this technology:

  • Traditional mathematical operations on processing units may suffer from limitations in precision, efficiency, or scalability.
  • By utilizing the modulo space and coprimes, this technology addresses these limitations and provides a more robust and efficient approach to mathematical computations.

Benefits of this technology:

  • The use of modulo space and coprimes allows for more accurate and efficient mathematical operations on processing units.
  • It can improve the performance and reliability of processing systems, leading to better results in various applications.
  • The technique can potentially reduce computational complexity and resource requirements, making it suitable for resource-constrained environments.


Original Abstract Submitted

Certain aspects of the present disclosure provide techniques and apparatus for performing mathematical operations on processing units based on data in the modulo space. An example method includes receiving a binary-space input to process (e.g., using a neural network or other processing system). The binary-space input is converted into a modulo-space input based on a set of coprimes defined for executing operations in a modulo space. A modulo-space result is generated through one or more modulo-space multiply-and-accumulate (MAC) units based on the modulo-space input. The modulo-space result is converted into a binary-space result, and the binary-space result is output.