Inventor

AMBARDEKAR AMOL ASHOK

US27 patents

Patents

27 patents

US10628345B2Apr 21, 2020

Enhancing processing performance of a DNN module by bandwidth control of fabric interface

MICROSOFT TECHNOLOGY LICENSING LLC6 citations92

US9619488B2Apr 11, 2017

Adaptable image search with computer vision assistance

MICROSOFT TECHNOLOGY LICENSING LLC24 citations86

US11476869B2Oct 18, 2022

Dynamically partitioning workload in a deep neural network module to reduce power consumption

MICROSOFT TECHNOLOGY LICENSING LLC2 citations84

US11405051B2Aug 2, 2022

Enhancing processing performance of artificial intelligence/machine hardware by data sharing and distribution as well as reuse of data in neuron buffer/line buffer

MICROSOFT TECHNOLOGY LICENSING LLC2 citations84

US11256976B2Feb 22, 2022

Dynamic sequencing of data partitions for optimizing memory utilization and performance of neural networks

MICROSOFT TECHNOLOGY LICENSING LLC2 citations84

US11182667B2Nov 23, 2021

Minimizing memory reads and increasing performance by leveraging aligned blob data in a processing unit of a neural network environment

MICROSOFT TECHNOLOGY LICENSING LLC2 citations84

US11100390B2Aug 24, 2021

Power-efficient deep neural network module configured for layer and operation fencing and dependency management

MICROSOFT TECHNOLOGY LICENSING LLC3 citations84

US10795836B2Oct 6, 2020

Data processing performance enhancement for neural networks using a virtualized data iterator

MICROSOFT TECHNOLOGY LICENSING LLC2 citations84

US10540584B2Jan 21, 2020

Queue management for direct memory access

MICROSOFT TECHNOLOGY LICENSING LLC3 citations84

US11010315B2May 18, 2021

Flexible hardware for high throughput vector dequantization with dynamic vector length and codebook size

MICROSOFT TECHNOLOGY LICENSING LLC2 citations82

US11722147B2Aug 8, 2023

Dynamic sequencing of data partitions for optimizing memory utilization and performance of neural networks

MICROSOFT TECHNOLOGY LICENSING LLC0 citations73

US11341399B2May 24, 2022

Reducing power consumption in a neural network processor by skipping processing operations

MICROSOFT TECHNOLOGY LICENSING LLC0 citations73

US11205118B2Dec 21, 2021

Power-efficient deep neural network module configured for parallel kernel and parallel input processing

MICROSOFT TECHNOLOGY LICENSING LLC0 citations73

US11176448B2Nov 16, 2021

Enhancing processing performance of a DNN module by bandwidth control of fabric interface

MICROSOFT TECHNOLOGY LICENSING LLC0 citations73

US11100391B2Aug 24, 2021

Power-efficient deep neural network module configured for executing a layer descriptor list

MICROSOFT TECHNOLOGY LICENSING LLC1 citations73

US11030131B2Jun 8, 2021

Data processing performance enhancement for neural networks using a virtualized data iterator

MICROSOFT TECHNOLOGY LICENSING LLC0 citations73

US10996739B2May 4, 2021

Reducing power consumption in a neural network environment using data management

MICROSOFT TECHNOLOGY LICENSING LLC2 citations73

US10963403B2Mar 30, 2021

Processing discontiguous memory as contiguous memory to improve performance of a neural network environment

MICROSOFT TECHNOLOGY LICENSING LLC0 citations73

US11909422B2Feb 20, 2024

Neural network processor using compression and decompression of activation data to reduce memory bandwidth utilization

MICROSOFT TECHNOLOGY LICENSING LLC1 citations71

US11750212B2Sep 5, 2023

Flexible hardware for high throughput vector dequantization with dynamic vector length and codebook size

MICROSOFT TECHNOLOGY LICENSING LLC0 citations71

US11528033B2Dec 13, 2022

Neural network processor using compression and decompression of activation data to reduce memory bandwidth utilization

MICROSOFT TECHNOLOGY LICENSING LLC1 citations71

US11514648B2Nov 29, 2022

Aligning input image data with model input data to generate image annotations

MICROSOFT TECHNOLOGY LICENSING LLC3 citations68

US11487342B2Nov 1, 2022

Reducing power consumption in a neural network environment using data management

MICROSOFT TECHNOLOGY LICENSING LLC0 citations62

US11526581B2Dec 13, 2022

AMBARDEKAR AMOL ASHOK

Patents

Enhancing processing performance of a DNN module by bandwidth control of fabric interface

Adaptable image search with computer vision assistance

Dynamically partitioning workload in a deep neural network module to reduce power consumption

Enhancing processing performance of artificial intelligence/machine hardware by data sharing and distribution as well as reuse of data in neuron buffer/line buffer

Dynamic sequencing of data partitions for optimizing memory utilization and performance of neural networks

Minimizing memory reads and increasing performance by leveraging aligned blob data in a processing unit of a neural network environment

Power-efficient deep neural network module configured for layer and operation fencing and dependency management

Data processing performance enhancement for neural networks using a virtualized data iterator

Queue management for direct memory access

Flexible hardware for high throughput vector dequantization with dynamic vector length and codebook size

Dynamic sequencing of data partitions for optimizing memory utilization and performance of neural networks

Reducing power consumption in a neural network processor by skipping processing operations

Power-efficient deep neural network module configured for parallel kernel and parallel input processing

Enhancing processing performance of a DNN module by bandwidth control of fabric interface

Power-efficient deep neural network module configured for executing a layer descriptor list

Data processing performance enhancement for neural networks using a virtualized data iterator

Reducing power consumption in a neural network environment using data management

Processing discontiguous memory as contiguous memory to improve performance of a neural network environment

Neural network processor using compression and decompression of activation data to reduce memory bandwidth utilization

Flexible hardware for high throughput vector dequantization with dynamic vector length and codebook size

Neural network processor using compression and decompression of activation data to reduce memory bandwidth utilization

Aligning input image data with model input data to generate image annotations

Reducing power consumption in a neural network environment using data management

Compression-encoding scheduled inputs for matrix computations

Selectively controlling memory power for scheduled computations

Compression-encoding scheduled inputs for matrix computations

Real-world object recognition for computing device