Breaking News
Design How-To

Dodging Amdahl's Law with message passing, FPGA-based, parallel processing

NO RATINGS
Page 1 / 4 Next >
More Related Links
View Comments: Newest First | Oldest First | Threaded View
devbisme
User Rank
Rookie
re: Dodging Amdahl's Law with message passing, FPGA-based, parallel processing
devbisme   3/11/2010 9:48:15 PM
NO RATINGS
I thought Amdahl's law related the maximum speedup of a program to the percentage that was inherently sequential. So a program that was 5% sequential could never be sped-up more than 20x, no matter how many FPGAs, microprocessors or fast MPIs were thrown at it. Or is this article referring to some other form of Amdahl's law?

TheMerc
User Rank
Rookie
re: Dodging Amdahl's Law with message passing, FPGA-based, parallel processing
TheMerc   3/11/2010 9:22:49 PM
NO RATINGS
Sorry guys. We solved this issue at NASA back in 1982. Unfortunately, we were so far ahead of the rest of the world (256 processor heterogeneous system) back then that no one ever read the papers we published and you probably cannot find them now. The only thing that makes it worse is that software developers have become so reliant on "canned" Operating Systems and Development Environments (IDE's)that they don't think for themselves about how the overall system really plays together.

stephendoyle
User Rank
Rookie
re: Dodging Amdahl's Law with message passing, FPGA-based, parallel processing
stephendoyle   3/1/2010 12:02:29 PM
NO RATINGS
I don't think that you are really dodging Amdahl's law. There is still overhead associated with the message passing, but by keeping the number of nodes down you are reducing the impact of the overhead on the overall system ... exactly in line with Amdahl's law.

Top Comments of the Week
August Cartoon Caption Winner!
August Cartoon Caption Winner!
"All the King's horses and all the KIng's men gave up on Humpty, so they handed the problem off to Engineering."
5 comments
Like Us on Facebook

Datasheets.com Parts Search

185 million searchable parts
(please enter a part number or hit search to begin)
EE Times on Twitter
EE Times Twitter Feed
Radio
LATEST ARCHIVED BROADCAST
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.
Flash Poll