Inventor
KHAITAN HARSHIT
US32 patents
⚠️ This page may combine multiple inventors who share the name “KHAITAN HARSHIT”. Patents are grouped by organization below to help distinguish them — per-person disambiguation is on the roadmap.
GOOGLE LLC
21 patentsUS10175980B2Jan 8, 2019
Neural network compute tile
GOOGLE LLC38 citations98
US10504022B2Dec 10, 2019
Neural network accelerator with parameters resident on chip
GOOGLE LLC22 citations94
US9959498B1May 1, 2018
Neural network instruction set architecture
GOOGLE LLC21 citations94
US9946539B1Apr 17, 2018
Accessing data in multi-dimensional tensors using adders
GOOGLE LLC19 citations86
US10802956B2Oct 13, 2020
Accessing prologue and epilogue data
GOOGLE LLC8 citations84
US10534607B2Jan 14, 2020
Accessing data in multi-dimensional tensors using adders
GOOGLE LLC10 citations84
US10496326B2Dec 3, 2019
Hardware double buffering using a special purpose computational unit
GOOGLE LLC9 citations84
US10248908B2Apr 2, 2019
Alternative loop limits for accessing data in multi-dimensional tensors
GOOGLE LLC10 citations84
US10108538B1Oct 23, 2018
Accessing prologue and epilogue data
GOOGLE LLC7 citations84
US12061968B2Aug 13, 2024
Neural network instruction set architecture
GOOGLE LLC1 citations73
US11422801B2Aug 23, 2022
Neural network compute tile
GOOGLE LLC2 citations73
US11379707B2Jul 5, 2022
Neural network instruction set architecture
GOOGLE LLC1 citations73
US10175912B1Jan 8, 2019
Hardware double buffering using a special purpose computational unit
GOOGLE LLC2 citations73
US11176493B2Nov 16, 2021
Virtualizing external memory as local to a machine learning accelerator
GOOGLE LLC2 citations70
US12455737B2Oct 28, 2025
Neural network compute tile
GOOGLE LLC0 citations62
US11816480B2Nov 14, 2023
Neural network compute tile
GOOGLE LLC0 citations62
US11727259B2Aug 15, 2023
Neural network accelerator with parameters resident on chip
GOOGLE LLC0 citations62
US11501144B2Nov 15, 2022
Neural network accelerator with parameters resident on chip
GOOGLE LLC0 citations62
US11099772B2Aug 24, 2021
Hardware double buffering using a special purpose computational unit
GOOGLE LLC0 citations62
US10885434B2Jan 5, 2021
Alternative loop limits for accessing data in multi-dimensional tensors
GOOGLE LLC1 citations62
US12417406B2Sep 16, 2025
Virtualizing external memory as local to a machine learning accelerator
GOOGLE LLC0 citations60
META PLATFORMS INC
7 patentsUS11709783B1Jul 25, 2023
Tensor data distribution using grid direct-memory access (DMA) controller
META PLATFORMS INC14 citations85
US12001893B1Jun 4, 2024
Distributed synchronization scheme
META PLATFORMS INC4 citations71
US11704562B1Jul 18, 2023
Architecture for virtual instructions
META PLATFORMS INC4 citations70
US11954580B2Apr 9, 2024
Spatial tiling of compute arrays with shared control
META PLATFORMS INC0 citations52
US12265492B2Apr 1, 2025
Circular buffer for input and output of tensor computations
META PLATFORMS INC0 citations51
US12197362B2Jan 14, 2025
Batch matrix multiplication operations in a machine learning accelerator
META PLATFORMS INC0 citations51
US11922306B2Mar 5, 2024
Tensor controller architecture
META PLATFORMS INC0 citations49