IP Library › Granted Patent US 11,436,186
Granted Patent B2
US 11,436,186 · App. 16/625,220 · Granted Sep 6, 2022

High throughput processors

Inventors: Robert D Catiller (Terre Haute, IN); Daniel Roig (Terre Haute, IN); Gnanashanmugam Elumalai (Terre Haute, IN)
Assignee: ICAT LLC
G06F15/80G06F9/3867
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 11,436,186
App. No.
16/625,220
Granted
Sep 6, 2022
Kind
B2
Abstract

An algorithmic matching pipelined compiler and a reusable algorithmic pipelined core comprise a high throughput processor system. The reusable algorithmic pipelined core is a reconfigurable processing core with a pipelined structure comprising a processor with a setup interface for programming any of a plurality of operations as determined by setup data, a logic decision processor for programming a look up table, a loop counter and a constant register, and a block of memory. This can be used to perform functions. A reconfigurable, programmable circuit routes data and results from one core to another core and/or IO controller and/or interrupt generator, as required to complete an algorithm without further intervention from a central or peripheral processor during processing of an algorithm.

Claims (16)

1. A high throughput processor, comprising:

an algorithmic matching pipelined compiler; and

an array of reconfigurable tiles, wherein the array of reconfigurable tiles are programmed by the algorithmic matching pipelined compiler, such that each of the reconfigurable tiles in the array of reconfigurable tiles is a reconfigurable core, and the array of reconfigurable tiles is arranged such that a result passes from one of the reconfigurable tiles in the array of reconfigurable tiles to another of the tiles in the array of reconfigurable tiles on each clock, wherein the algorithmic matching pipelined compiler configures the array of reconfigurable tiles such that the array of reconfigurable tiles performs algorithmic calculations as a computer processor at a throughput rate of one result per one clock without propagation delays.

2. The high throughput processor of claim 1 , wherein each reconfigurable tile is an 8-bit reconfigurable tile.

3. The high throughput processor of claim 2 , further comprising a precompiler wherein the precompiler processes a standard higher level software language written not for the high throughput processor but for a type of conventional non-reconfigurable processor, and the precompiler provides instructions for the algorithmic matching pipelined compiler for programming the array of reconfigurable tiles.

4. The high throughput processor of claim 1 , wherein each of the reconfigurable tiles within the array of reconfigurable tiles comprises at least one LUT.

5. The high throughput processor of claim 4 , wherein the at least one LUT implements a logic module.

6. The high throughput processor of claim 5 , wherein each of the reconfigurable tiles within the array of reconfigurable tiles is reconfigurably coupled to one or more adjacent reconfigurable tiles within the array of reconfigurable tiles, such that data may be processed by the array of reconfigurable tiles as 8 bit data, 16 bit data, 32 bit data, 64 bit data, and any other multiple of 8 bit data.

7. The high throughput processor of claim 1 , wherein the array of reconfigurable tiles are configured to complete tasks without any further overhead from the processing unit.

8. The high throughput processor of claim 1 , wherein each of the reconfigurable tiles within the array of reconfigurable tiles is reconfigurably coupled to one or more adjacent reconfigurable tiles within the array of reconfigurable tiles, such that point-to-point connectivity is provided without having to share any communications channel or bus with any other logic module.

9. A system comprising a high throughput processor according to claim 1 , comprising the steps of:

initially inputting data during a latency period; and

calculating a result on each clock after the latency period.

10. The system of claim 9 , wherein the clock rate for the high throughput processor configured with 1-bit reconfigurable tiles is eight times faster than the clock rate for the high throughput processor configured with 8-bit reconfigurable tiles.

11. The system of claim 10 , wherein output data precision for the high throughput processor configured with a plurality of 1-bit reconfigurable tiles or a plurality of 4-bit reconfigurable tiles is the same as output data precision for the high throughput processor configured with 8-bit reconfigurable tiles.

12. The system of claim 11 , further comprised of the step of reconfigurably disposing a plurality of the high throughput processors according to claim 1 within a single ASIC chip package.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Feb 11, 2020
From: CATILLER, ROBERT D; ELUMALAI, GNANASHANMUGAM
To: ICAT LLC
Reel/Frame 051788/0365 →
Continuity (2)
Provisional Application 62523528 · Jun 22, 2017
Related Publication 20200142857A1 · May 7, 2020