Graphical User Interface for Reconfigurable Processor - PowerPoint PPT Presentation

1 / 23
About This Presentation
Title:

Graphical User Interface for Reconfigurable Processor

Description:

IAmE Tu Le Institute of Advanced Microelectronics ECE/CAMBR University of Idaho le7775_at_uidaho.edu David M. Buehler Institute of Advanced Microelectronics – PowerPoint PPT presentation

Number of Views:52
Avg rating:3.0/5.0
Slides: 24
Provided by: Tu5
Learn more at: http://klabs.org
Category:

less

Transcript and Presenter's Notes

Title: Graphical User Interface for Reconfigurable Processor


1
Graphical Design Environment for a Reconfigurable
Processor
Tu Le Institute of Advanced Microelectronics ECE/C
AMBR University of Idaho le7775_at_uidaho.edu
David M. Buehler Institute of Advanced
Microelectronics ECE/CAMBR University of
Idaho dbuehler_at_uidaho.edu
Gregory Donohoe Institute of Advanced
Microelectronics ECE/CAMBR University of
Idaho donohoe_at_cambr.uidaho.edu
Pen-Shu Yeh NASA GSFC Code 567 pen-shu.yeh_at_gsfc.na
sa.gov
1
223/BOF/MAPLD 2004
2
Overview
  • Field Programmable Processor Array (FPPA)
  • Graphical Programming
  • FPPA Matlab Simulink model (floating point)
  • FPPA graphical interface design flow
  • Examples
  • Summary
  • Future work

2
223/BOF/MAPLD 2004
3
Field Programmable Processor Array (FPPA)
  • An embedded data processor VLSI chip for
    spacecraft.
  • Radiation-tolerant, 0.25m CMOS process
  • 16 on-board, fixed point processing elements
  • Implements a reconfigurable synchronous data
    flow processor
  • Run-time reconfigurable
  • Extensible by tiling multiple chips
  • Serves as accelerator to a host CPU
  • Application development
  • Text base development
  • Configuration and Run-time compilers
  • Standalone functional simulator
  • FPPA Simulink graphical design environment (GUI)

3
223/BOF/MAPLD 2004
4
FPPA Data Path Elements
  • FPPA features
  • 16 configurable on-board PEs
  • Four 16-bit-wide, bidirectional I/O ports
  • One 16-bit-wide dedicated output port
  • On-board program memory and execution unit
  • Interface control signals

4
223/BOF/MAPLD 2004
5
Processing Element (PE) components
5
223/BOF/MAPLD 2004
6
Example of the simple pipeline using multiple
PEs
6
223/BOF/MAPLD 2004
7
FPPA programming complexity
  • Data path descriptions
  • PE
  • Cluster
  • I/O
  • Runtime descriptions
  • Enable / Disable PEs
  • External peripherals
  • Procedural languages like C do not naturally
    describe a dataflow architecture like the FPPA

223/BOF/MAPLD 2004
7
8
Realizing 4 taps FIR filter with FPPA
Data Path
Difference Equation
Routing of output and local bus data
Routing of input data
8
223/BOF/MAPLD 2004
9
Graphical Programming
  • Naturally represents dataflow computational model
  • Quick to learn
  • Familiar Matlab Simulink graphical design
    environment
  • Floating point versus Fixed point implementation
  • Rapid Development and Validation

9
223/BOF/MAPLD 2004
10
FPPA Simulink Model
  • Data path
  • Computational PE
  • Conditional PE
  • Runtime
  • Enable / Disable PEs and modules
  • Data format

10
223/BOF/MAPLD 2004
11
General Model of the Processing Element
11
223/BOF/MAPLD 2004
12
Simulink PE model (Data path)
12
223/BOF/MAPLD 2004
13
Simulink PE model (Data path)
  • Unconditional PE
  • Delay
  • Shift right or left
  • (X Y)
  • (X Y)
  • (X Y) Z
  • (X - Y) Z
  • (XY Z)
  • (XY Z)
  • C0, C1
  • Conditional PE
  • If (condition) then
  • Perform task A
  • Else
  • Perform task B

13
223/BOF/MAPLD 2004
14
Simulink PE model (Run Time)
  • Fire pattern
  • Example
  • Fire pattern 1 PE always enable
  • Fire pattern 1 0 PE enable every other
    instruction
  • Fire pattern 1 0 0 0 PE enable every 4th
    instruction
  • Potential applications
  • Multi-rate Signal Processing
  • Multi-functionality in one FPPA

14
223/BOF/MAPLD 2004
15
FFPA graphical interface design flow
Note SIFOpt tool is a result of David M.
Buehler dissertation at the University of Idaho.
Algorithm
Simulink
  • Model (floating point)
  • Design Data Path
  • Provide Input data
  • Data format
  • Run time

Validation
result
Golden model
result
PERL (floating ? fixed point)
15
223/BOF/MAPLD 2004
16
Example 1 Realizing 4 tap FIR filter
Difference Equation
Input waveform
Output waveform
16
223/BOF/MAPLD 2004
17
Example 2 Wavelet (Multi-rate filter bank)
Source
L1
L2
L3
L4
  • Low Pass filter
  • 4 taps with Debenchies coefficients
  • Debenchies coefficients are normalized by
    sqrt(2) to get the sum equal to one

17
223/BOF/MAPLD 2004
18
Example 2 Model 4 series filters
18
223/BOF/MAPLD 2004
19
Example 2 Model Down Sampling
19
223/BOF/MAPLD 2004
20
Example 2 FPPA view
20
223/BOF/MAPLD 2004
21
Example 2 Simulation result
21
223/BOF/MAPLD 2004
22

Future work
  • Data path optimization
  • Speed
  • Area
  • Cook book for common use circuits
  • Square root
  • Division
  • Trigonometry functions
  • Iterative methods
  • Down / Up sample
  • Matrix computation

22
223/BOF/MAPLD 2004
23

Summary
  • Present FPPA
  • FPPA programming complexity
  • Graphical programming environment
  • Intuitive
  • Fast and easy
  • Validation and Verification
  • Matlab gt Floating point
  • FPPAsim gt hardware specific
  • Compile to FPPA code

23
223/BOF/MAPLD 2004
Write a Comment
User Comments (0)
About PowerShow.com