SAR ATR Hour for the SLAAC Fall 99 Retreat - PowerPoint PPT Presentation

1 / 6
About This Presentation
Title:

SAR ATR Hour for the SLAAC Fall 99 Retreat

Description:

... Module Performance Goals: Brian Bray. FOA: Scott Hemmert. SLD: ... Wrap-up (Good/Bad/Future): Brian Bray. Sandia National Labs. Current Challenge Problem Modules ... – PowerPoint PPT presentation

Number of Views:43
Avg rating:3.0/5.0
Slides: 7
Provided by: brian316
Category:
Tags: atr | sar | slaac | bray | fall | hour | retreat

less

Transcript and Presenter's Notes

Title: SAR ATR Hour for the SLAAC Fall 99 Retreat


1
SAR ATR Hourfor the SLAAC Fall 99 Retreat
Intro/Module Performance Goals Brian Bray FOA
Scott Hemmert SLD Steve Crago CDI Mike
Wirthlin Wrap-up (Good/Bad/Future) Brian Bray
2
Current Challenge Problem Modules
  • FOA (Focus of Attention)
  • adaptive rank order quantization and multi-level
    morphology
  • quad uP 2 Mpixels/sec ACS goal 10
    Mpixels/sec
  • SLD (Second Level Detection)
  • for in-the-clear scenarios
  • adaptive attenuation estimation with binary
    template matching
  • quad uP 16000 templates/sec ACS goal 80000
    templates/sec
  • CDI (Contamination Distribution Indexer)
  • for camouflage, concealment and deception (CCD)
    scenarios
  • epsilon-contaminated mixtures model (10X more
    compute intensive than SLD)
  • initial ACS goal is just the 2X not the 1X
    templates (90 of compute)
  • quad uP 1600 templates/sec Inital ACS
    goal 16000 templates/sec
    End ACS goal (2X and 1X templates at 80000
    templates/sec)

Mpixel in downsampled space
microprocessor PowerPC 750 _at_ 400Mhz
3
The Good
  • ACS CDI Performance
  • JHDL
  • BYU students and faculty
  • Virtex parts
  • on-chip memory
  • large amounts of logic

4
The Bad
  • We are behind
  • Need an embedded Virtex based ACS board
  • ACS parts with large penalty for large precision
    and FP operations
  • What is the third generation ACS part?

5
The Future (As I See It)
  • Bigger second generation ACS parts and better
    tools will come
  • third generation will not be here in near term
  • what will it be???
  • Virtex-like with RISC core, I, D and DRAM
    interface?
  • Will it be an improvement to compute tasks or
    just system on a chip applications
  • PCI accelerator cards with multiprocessor
    workstation based groundstations
  • will reduce the need for embedded VME hardware
  • Annapolis Microsystems PCI
  • WildSTAR is a significant improvement over
    WildForce
  • SLAAC PCI?
  • When compared to the WildSTAR, what design
    features can a SLAAC PCI Virtex board provide to
    overcome not being COTS and available now?
  • Daughtercards for embedded VME multicomputers
  • CSPI, Mercury
  • many Mercury only shops for embedded VME
    multicomputers

6
Mercury Compatible ACS Node
Can SLAAC get access to these chips and enough
info to write SW drivers?
Mercurys PPC Daughtercard
ACS Daughtercard
Can this bus be run at a lower clock rate than
83.3Mhz?
CLK Gen
Boot EEPROM
How do others go about designing Mercury
compatible daughtercards?
64
64
64
64
64
64
Write a Comment
User Comments (0)
About PowerShow.com