IP Library Granted Patent US 11,556,390
Granted Patent B2
US 11,556,390 · App. 16/591,402 · Granted Jan 17, 2023

Efficient high bandwidth shared memory architectures for parallel machine learning and AI processing of large data sets and streams

Inventors: Phillip Alvelda, VII (Arlington, VA); Markus Krause (Oakland, CA); Todd Allen Stiers (Great Falls, VA)
Assignee: Brainworks Foundry, Inc.
G06F9/5066G06F9/5083G06F9/544
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 11,556,390
App. No.
16/591,402
Granted
Jan 17, 2023
Kind
B2
Abstract

The present disclosure relates to systems and methods to implement efficient high-bandwidth shared memory systems particularly suited for parallelizing and operating large scale machine learning and AI computing systems necessary to efficiently process high volume data sets and streams.

Claims (47)

1. A method for implementing parallel processing with shared memory comprising:

receiving, by at least one processor, a data stream object produced by one or more remote user devices including at least one digital image capture device;

analyzing, by the at least one processor, the data stream object for validity;

applying, by the at least one processor, security measures to the data stream object; saving the data stream object, by the at least one processor, as a sequence of individual elements into the shared memory;

analyzing, by the at least one processor, meta data describing process steps to be applied to the data stream object and preparing a sequence of algorithms to be run on each of the individual elements; and

performing, by the at least one processor, each independent parallel processing task of a plurality of independent parallel processing tasks concurrently and independently using each individual data element of the sequence of individual elements in the shared memory.

2. The method of claim 1 , wherein the data stream object comprises at least one of a plurality of images, a video stream, and a sound stream.

3. The method of claim 1 , wherein the data stream object comprises meta data describing process steps to be applied to the data stream object.

4. The method of claim 1 , further comprising executing, by the at least one computing device, the process steps on the each of the individual elements using the sequence of algorithms.

5. The method of claim 1 , further comprising generating, by the at least one computing device, a report based on a current state of the shared memory.

6. The method of claim 1 , further comprising balancing, by the at least one processor, a plurality of independent processing pipelines across the cluster of computing devices of using a predictive load balancer.

7. A system for implementing parallel processing with shared memory comprising:

a shared memory providing read and write access to one or more remote user devices and one or more computing devices;

the one or more remote user devices configured to:

receive a data stream object produced by at least one digital image capture device; and

save the data stream object as a sequence of individual elements into the shared memory; and

the one or more computing devices configured to:

read the shared memory to access each individual data element of the sequence of individual elements concurrently and independently by a plurality of independent parallel processing tasks;

analyze the data stream object for validity;

apply security measures to the data stream object

analyze meta data describing process steps to be applied to the data stream object and preparing a sequence of algorithms to be run on each of the individual elements; and

perform each processing task from the plurality of independent parallel processing tasks concurrently and independently using each individual data element of the sequence of individual elements in the shared memory.

8. The system of claim 7 , wherein the data stream object comprises at least one of a plurality of images, a video stream, and a sound stream.

9. The system of claim 7 , wherein the data stream object comprises meta data describing process steps to be applied to the data stream object.

10. The system of claim 7 , wherein the one or more computing devices executes the process steps on the each of the individual elements using the sequence of algorithms.

11. The system of claim 7 , wherein the one or more computing devices generates a report based on a current state of the shared memory.

12. The system of claim 7 , wherein the one or more computing devices balances a plurality of independent processing pipelines across the cluster of computing devices of using a predictive load balancer.

13. A method for implementing parallel processing with shared memory comprising:

receiving, by at least one processor, a data stream object produced by one or more remote user devices including at least one digital image capture device;

saving each data stream object, by the at least one processor, as a sequence of individual elements into the shared memory;

analyzing, by the at least one processor, incoming data stream objects for validity;

applying, by the at least one processor, security measures to the incoming data stream objects;

analyzing, by the at least one processor, meta data describing process steps to be applied to the data stream object and preparing a sequence of algorithms to be run on each of the individual elements;

accessing, by the at least one processor, the sequence of individual elements within the shared memory for at least one processing task of a plurality of independent parallel processing tasks;

performing, by the at least one processor, the at least one processing task using the sequence of individual elements in the shared memory; and

saving, by the at least one computing device, results of the at least one processing task into the shared memory.

14. A system for implementing parallel processing with shared memory comprising:

a shared memory providing read and write access to one or more remote user devices and one or more computing devices;

the one or more remote user devices configured to:

receive a data stream object produced by one or more remote user devices including at least one digital image capture device;

save each data stream object as a sequence of individual elements into the shared memory;

analyze incoming data stream objects for validity;

apply security measures to the incoming data stream objects;

analyze meta data describing process steps to be applied to the data stream object and preparing a sequence of algorithms to be run on each of the individual elements;

access the sequence of individual elements within the shared memory for at least one processing task of a plurality of independent parallel processing tasks;

perform the at least one processing task using the sequence of individual elements in the shared memory; and

save results of the at least one processing task into the shared memory.

Assignments (4)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Jan 31, 2024
From: MEDIO LABS, INC.
To: AMPLIMER GENOMICS
Reel/Frame 066310/0001 →
CHANGE OF NAME Recorded Mar 23, 2023
From: BRAINWORKS FOUNDRY, INC., A/K/A BRAINWORKS
To: MEDIO LABS, INC.
Reel/Frame 063154/0668 →
SECURITY INTEREST Recorded Mar 23, 2023
From: MEDIO LABS, INC.
To: BULL, MATTHEW NORMAN; HYGROVEST LIMITED; PHEAKES PTY LTD ATF SENATE; WIMALEX PTY LTD ATF TRIO S/F; AUSTIN, JEREMY MARK; SUNSET CAPITAL MANAGEMENT PTY LTD ATF SUNSET SUPERFUND; REGAL WORLD CONSULTING PTY LTD ATF R WU FAMILY; DANTEEN PTY LTD; NYSHA INVESTMENTS PTY LTD ATF SANGHAVI FAMILY; BLACKBURN, KATE MAREE; COWOSO CAPITAL PTY LTD ATF THE COWOSO SUPER FUND; TARABORRELLI, ANGELOMARIA; JAINSON FAMILY PTY LTD ATF JAINSON FAMILY; PARKRANGE NOMINEES PTY LTD ATF PARKRANGE INVESTMENT; XAU PTY LTD ATF JOHN & CARA SUPER FUND; XAU PTY LTD ATF CHP; LEWIT, ALEXANDER; GREGORY WALL ATF G & M WALL SUPER FUND; MICHELLE WALL ATF G & M WALL SUPER FUND; BRIANT NOMINEES PTY LTD ATF BRIANT SUPER FUND; ZIZIPHUS PTY LTD; AGENS PTY LTD ATF THE MARK COLLINS S/F; FPMC PROPERTY PTY LTD ATF FPMC PROPERTY DISC; VAN NGUYEN, HOWARD; THIKANE, AMOL; MCKENNA, JACK MICHAEL; ALLAN GRAHAM JENZEN ATF AG E JENZEN P/L NO 2; ELIZABETH JENZEN ATF AG E JENZEN P/L NO 2; JONES, DENNIS PERCIVAL; JONES, ANGELA MARGARET; S3 CONSORTIUM HOLDINGS PTY LTD ATF NEXTINVESTORS DOT COM; RUBEN, VANESSA
Reel/Frame 065021/0408 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Dec 13, 2022
From: ALVELDA, PHILLIP, VII; KRAUSE, MARKUS; STIERS, TODD ALLEN
To: BRAINWORKS FOUNDRY, INC.
Reel/Frame 062071/0514 →
Continuity (2)
Provisional Application 62739901 · Oct 2, 2018
Related Publication 20200104185A1 · Apr 2, 2020