University of Illinois

Leland Stanford Junior University

Deming Chen

Subhasish Mitra

Alexander Gerhard Schwing

Jian Huang

David Lin

Keith Campbell

Youjie Li

A programmable switch includes an input arbiter to analyze packet headers of incoming packets and determine which of the incoming packets are part of gradient vectors received from worker computing devices that are performing reinforcement learning. The programmable switch also includes an accelerator coupled to the input arbiter, the accelerator to: receive the incoming packets from the input arbiter; asynchronously aggregate gradient values of the incoming packets, as the gradient values are received, to generate an aggregated data packet associated with a gradient segment of the gradient vectors; and transfer the aggregated data packet to the input arbiter to be transmitted to the worker computing devices, which are to update local weights based on the aggregated data packet.

Accelerating distributed reinforcement learning with in-switch computing

Disclosed are improved methods and structures for verifying integrated circuits and in particular systems-on-a-chip constructed therefrom. Our methods&#x2014;which we call Quick Error Detection&#x2014;Hardware (QED-H)&#x2014;advantageously quickly detect and fix anomalies (bugs) within SoC hardware components&#x2014;and in particular customized SoC hardware components that are not necessarily software programmable. Of further advantage, methods according to the present disclosure are compatible with existing Quick Error Detection (QED) techniques while being extensible to target software-programmable components as well. In sharp contrast to prior art methods, method(s) according to the present disclosure represent a new system validation methodology that builds validation checks in both software and hardware components seamlessly and systematically, thus enabling extremely quick error detection and localization for all digital components of the entire SoC advantageously producing productivity and time-to-market gains.

System-level validation of systems-on-a-chip (SoC)

Growing community of inventors

Urbana, IL, United States of America

Deming Chen