Sasidhar Paruchuri

Title: Sasidhar Paruchuri and Data Transform Acceleration

Introduction

Sasidhar Paruchuri is an inventor based in San Diego, California. His patent work focuses on accelerating data transformations by coordinating commands, metadata, and processing pipelines between host computing units and data transform accelerators. He is associated with MaxLinear, Inc., and holds 2 patents.

Latest Patents

Paruchuri’s latest patents address methods for configuring data transform pipelines using metadata associated with data transform commands.

“Data Transform Acceleration” describes a method in which a host computing unit identifies an address for a command in a container data structure and accesses the command using that address. The system obtains metadata from information in the command and uses it to configure a data transform pipeline. The metadata may be shared, in whole or in part, among multiple grouped commands.

“Data Transform Acceleration Using Metadata Stored in Accelerator Memory” describes a related approach in which the container data structure is in the data transform accelerator, while the command is in the host computing unit. The method obtains metadata from the command and uses it to configure the pipeline. The metadata may reside in accelerator memory, host memory, or across both, and may be shared among grouped commands.

Career Highlights

Paruchuri’s patent portfolio reflects work on data-processing systems that use metadata to coordinate command execution and pipeline configuration. His inventions explore different arrangements of commands and metadata across host and accelerator memory.

Collaborations

Paruchuri’s listed coworkers include Pinaki Shankar Chanda and Michael Ray Ham. His work is associated with MaxLinear, Inc.

Conclusion

Sasidhar Paruchuri’s two patents focus on metadata-driven data transform acceleration, with approaches that distribute commands and metadata between host computing units and accelerators.