IP Library Granted Patent US 10,387,365
Granted Patent B2
US 10,387,365 · App. 15/737,151 · Granted Aug 20, 2019

Method and system for high performance real time pattern recognition

Inventors: Jamieson T. Olsen (Batavia, IL); Tiehui Ted Liu (Naperville, IL); James R. Hoff (Wheaton, IL)
Assignee: Fermi Research Alliance, LLC
G06F15/17387G06F15/76G06K9/62H04B10/40H04B10/801G06F2015/768
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 10,387,365
App. No.
15/737,151
Granted
Aug 20, 2019
Kind
B2
Abstract

Systems and methods supporting high performance real time pattern recognition by including time and regional multiplexing using high bandwidth, board-to-board communications channels, and 3D vertical integration. An array of processing boards can each be coupled a rear transition board, the array achieving time and regional multiplexing using high bandwidth board-to-board communications channels and 3D vertical integration.

Claims (32)

1. A system supporting real time data processing, comprising:

an array of processing boards;

a rear transition board coupled to each of the processing boards achieving time and regional multiplexing using board-to-board communications channels;

at least one mezzanine card per processing board said at least one mezzanine card further comprising a field programmable gate array, data processing circuitry, and a plurality of fiber optic transceivers; and

at least one through silicon via connecting a plurality of two dimensional integrated circuits configured on at least one of said mezzanine cards in order to provide pattern recognition associative memory for identifying patterns according to a 3D vertical integration.

2. The system of claim 1 , wherein the processing boards incorporate a full-mesh architecture permitting, inter-board communication and wherein said processing boards conform to advanced telecommunications computing architecture specification standards.

3. The system of claim 1 further comprising a backplane configured to time-multiplex incoming data in a manner that manages input and output demands of the system.

4. The system of claim 1 wherein said data processing circuitry comprises pattern recognition circuitry implemented in a field programmable gate array.

5. The system of claim 1 wherein said data processing circuitry comprises pattern recognition circuitry implemented in a pattern recognition application specific integrated circuit.

6. A data processing apparatus, comprising:

an array of processing boards;

a rear transition board coupled to each of the processing boards achieving time and regional multiplexing using board-to-board communication channels;

at least one mezzanine card; and

at least one through silicon via connecting said plurality of two dimensional pattern recognition integrated circuits configured on at least one of said mezzanine cards in order to provide pattern recognition associative memory for identifying patterns according to a 3D vertical integration.

7. The apparatus of claim 6 , wherein the processing boards incorporate a full-mesh architecture permitting, inter-board communication and wherein said processing boards conform to advanced telecommunications computing architecture specification standards.

8. The apparatus of claim 6 further comprising a backplane configured to time-multiplex incoming data in a manner that manages input and output demands of the apparatus.

9. The apparatus of claim 6 wherein said at least one mezzanine card further comprises:

a field programmable gate array;

data processing circuitry; and

a plurality of fiber optic transceivers.

10. The apparatus of claim 9 wherein said data processing circuitry comprises pattern recognition circuitry implemented in a field programmable gate array.

11. The apparatus of claim 9 wherein said data processing circuitry comprises pattern recognition circuitry implemented in a pattern recognition application specific integrated circuit.

12. A method for real time data processing, comprising:

configuring an array of processing boards;

coupling said array of processing boards in order to provide inter-board communication thereby achieving time and regional multiplexing using board-to-board communications channels; and

identifying patterns according to a 3D vertical integration.

13. The method of claim 12 further comprising:

connecting a plurality of two dimensional pattern recognition integrated circuits vertically with at least one through silicon via, on at least one of mezzanine card;

connecting said mezzanine card to said array of processing boards; and

coupling a rear transition board to each of the processing boards, said coupling and array achieving said time and regional multiplexing using board-to-board communications channels and 3D vertical integration.

14. The method of claim 12 further comprising:

time-multiplexing of incoming data in a manner that manages input and output demands with a backplane.

Assignments (3)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Dec 30, 2024
From: FERMI RESEARCH ALLIANCE, LLC
To: FERMI FORWARD DISCOVERY GROUP, LLC
Reel/Frame 069795/0347 →
CONFIRMATORY LICENSE Recorded May 31, 2023
From: FERMI RESEARCH ALLIANCE, LLC
To: UNITED STATES DEPARTMENT OF ENERGY
Reel/Frame 063803/0950 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Apr 20, 2018
From: OLSON, JAMIESON T.; LIU, TIEHUI TED; HOFF, JAMES R.
To: FERMI RESEARCH ALLIANCE, LLC
Reel/Frame 045601/0106 →
Continuity (2)
Provisional Application 62181040 · Jun 17, 2015
Related Publication 20190012293A1 · Jan 10, 2019