Skip to content

Details

PLEASE REGISTER HERE
https://trifork.zoom.us/webinar/register/WN_5OApy9-CRlWg80pdF0p1Gw

Join us along with Daniel Wisehart, inventor of imaging electronics in CCD-based cameras successfully deployed in NASA missions for a discussion and Q&A on moving software to hardware, the why and why not, the problems and how it's done.

Why?
● To keep the CPU free to do what SW is better at and to use HW for what it is better at.
● To filter the data noise the SW has to deal with and provide it only with actionable data.
● To prioritize and sort incoming data into separate channels.
Configure HW triggers to act automatically on specific data.
● To speed up tasks that can afford the cost of communicating between SW and HW.
● To take advantage of the massively parallel nature of HW.
● To run long winded parallel process like genetic algorithm search

Why not?
● CPU is a dynamic, general purpose, fetch/fetch/execute/store machine.
● HW has a set configuration of static structures.

What are the problems?
● HW dev takes roughly five times longer than SW dev, so Agile is difficult.
● A synth/place/route takes one to four hours, so CI/CD is less helpful.
● Simulation is available, but data creation and simulation are slow.
Debugging in hardware is very limited.

How is it done?
● Find the people who know FPGAs and who have experience with PCIe and Network designs.
● Create an interface between HW and SW that works well for both teams.
● Create test adapters for both the HW and SW teams that operate the other side of the interface so they may work in parallel.
● Identify a hardware platform--usually, a PCIe card--with a suitable FPGA, could be an AWS FPGA server.
● Build out servers to support HW development, or rent them from AWS.

More about Daniel:
Daniel Wisehart is a sole inventor of imaging electronics in CCD-based cameras successfully deployed in NASA missions:

● Ultra-low noise, ultra-low light, radiation-hardened electronics flew into space on an earth-sensing rocket called VISIONS.
● Imaging the Mars Lander’s infrared range finder during daylight hours from JPL in Southern California during the Lander’s sling-shot around the earth.
● Custom image sensor for ground testing Hubble’s Wide Field Planetary Camera no. 3 (WFPC3).

PLEASE REGISTER HERE
https://trifork.zoom.us/webinar/register/WN_5OApy9-CRlWg80pdF0p1Gw

You may also like