Shop OBEX P1 Docs P2 Docs Learn Events
Parallel Processing and Concurrency — Parallax Forums

Parallel Processing and Concurrency

mp0mp0 Posts: 1
edited 2007-08-12 14:51 in Propeller 1
"Me: www.cs.utexas.edu/users/EWD/ewd13xx/EWD1303.PDF
Friend: Heh. I scrolled down and was like, "Why is what this guy says matter? What the heck did Dr. Edsger W. Dÿk... oh."

It's an interesting paper on the importance of separation of concerns as well as trends in concurrency.

It's interesting that the earliest attempts at concurrency are similar to the propeller... separate logical execution threads, each servicing a single device. It's ironic that interrupts that are eliminated from the propeller with the use of separate execution cores were introduced to solve problems with using separate execution threads.

Comments

  • Mike GreenMike Green Posts: 23,101
    edited 2007-08-12 05:14
    In the mid-1970's, I wrote several operating systems based on Dykstra's ideas. One was for the IBM System 360 and the other was for a business computer based on the Zilog Z-80. In both cases, there was a very small low level kernel that took the existing I/O interrupts and turned them into events in multiple execution threads. The rest of the system only saw multiple parallel execution threads with semaphores (locks) synchronizing them. There was never any problem with interrupts. There was never any problem with timing dependencies. For the Z-80 system, the biggest problem was the limited processor throughput. Some of the comm channels would not get serviced fast enough and the hardware would get overruns. Adding additional buffers wouldn't help enough because the system really was overloaded at times. The operating system was having to do synchronous serial I/O with CRC calculation in software, etc.
  • potatoheadpotatohead Posts: 10,261
    edited 2007-08-12 06:53
    Thanks for that link!

    What an excellent read. Gives one a sense of perspective that's thought provoking.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Propeller Wiki: Share the coolness!
  • deSilvadeSilva Posts: 2,967
    edited 2007-08-12 07:43
    @mp0: Thank you for this article which reminds me of the "old days"... I had been a student at my German university when the TR4 was exchanged by a TR440 smile.gif

    I have always been impressed by Dijkstra's clear handwriting which reflects the chrystal clearness of this ideas, mostly ignored outside academic circles. He blames himself in this article for not having published some results a few years earlier. He should not smile.gif
  • Fred HawkinsFred Hawkins Posts: 997
    edited 2007-08-12 14:51
    The base address of that site: http://www.cs.utexas.edu/users/EWD/

    Worth bookmarking, I'd say.
Sign In or Register to comment.