/Patent Tracker
MicrosoftGranted patentChips & hardwareInterest 4/10

Microsoft patents chip method to save power by adjusting math precision

Microsoft has been granted a patent for a method to make neural processing units (NPUs) use less energy. The system dynamically lowers the precision of calculations in specific parts of the chip when high accuracy is not needed, based on data gathered during the AI model's training. This allows the hardware to run more efficiently by avoiding unnecessary high-precision processing.

Why it mattersThis is interesting to tech enthusiasts and hardware designers because it offers a specific way to reduce the energy cost of running AI models on dedicated chips.

In the filing’s own words

Methods, systems and computer program products are provided for improving performance (e.g., reducing power consumption) of a hardware accelerator (e.g., neural processor) comprising hybrid or analog multiply and accumulate (MAC) processing elements (PEs). Selective variation of the precision of an array of MAC PEs may reduce power consumption of a neural processor. Power may be conserved by dynamically controlling the precision of analog to digital (ADC) output bits for one or more MAC PEs. Dynamic control of ADC output bit precision may be based on precision information determined during training and/or post-training (e.g., quantization) of an artificial intelligence (AI) neural network (NN) model implemented by the neural processor. Precision information may include a range of dynamic precision for each of a plurality of nodes of a computation graph for the AI NN model.

What exactly they claim (claim 1)

1. A neural processing unit (NPU), comprising: a memory storing bit precision information determined during at least one of training or quantization of a neural network model; a precision determiner configured to dynamically determine a precision indication based on the bit precision information; and a power conservation controller configured to selectively vary a precision of a digital output based on the precision indication, the selective variation of the precision of the digital output reducing power consumption of the NPU.

Facts

Company
Microsoft (Microsoft Technology Licensing, LLC)
Official title
Analog MAC aware DNN improvement
Patent number
US12737026B2
Application number
18400749
Filed
Dec 29, 2023
Granted
Sep 15, 2026
Inventors
4
Claims
20

Classified under

G06F 1/3234 Physics & computingG06F 7/5443 Physics & computingG06F 13/102 Physics & computingG06N 3/02 Physics & computing
Read the full filing (official PDF) ↗ Open in the Patent Tracker

More from Microsoft in Chips & hardware

Summaries are simplified from the filing’s own abstract and first claim. A patent filing is not a product announcement.
Patent data and drawings: U.S. Patent and Trademark Office (USPTO). Nebula is not affiliated with or endorsed by the USPTO.