CPUFreak91 Member Posts: 2337 From: Registered: 02-01-2005 |
What kinds of HPC projects have you guys worked on? Have you used any code that makes use of the omp atomic mode (prevents the OS from interrupting the thread) in C++ or an equivalent in a different language? What have you used them for?
------------------ "Oh, bother," said the Borg. "We've assimilated Pooh." Any fool can know, the point is to understand. -- Albert Einstein |
|
MastaLlama Member Posts: 671 From: Houston, TX USA Registered: 08-10-2005 |
I have not. That reminds me of early computing when you could write a program that would cause the monitor to refresh too fast and actually physically melt because it got too hot. | |
PFC Member Posts: 29 From: Canada Registered: 10-16-2007 |
hmm.... not too sure what that does? when would you want to use that? ------------------ |
|
CPUFreak91 Member Posts: 2337 From: Registered: 02-01-2005 |
quote: Atomic mode could be used for something like I/O or some critical calculation that you don't want interrrupted by the OS. ------------------ "Oh, bother," said the Borg. "We've assimilated Pooh." Any fool can know, the point is to understand. -- Albert Einstein |
|
supercoder Member Posts: 37 From: Registered: 08-20-2007 |
i would think this would have to be compiled into the kernel, I don't think most os's would allow this, do you know what os supports it? my prof mentioned a real-time os (unlike ms, linix, unix, mac, etc..) that runs the linux os as just a user process (kinda treats it like a program). so you are presented with the linux os to do all your developement and use it as normal. but since behind the scene you have the REAL os, which you can send your time critical code to run on you get real-time perks like guaranteed execution times, which other os's cannot give. I found an article on the real time os ------------------ [This message has been edited by supercoder (edited December 07, 2007).] |
|
CPUFreak91 Member Posts: 2337 From: Registered: 02-01-2005 |
quote: No I'm not sure. The more I look around, the more I see a lack of evidence that supports this (or maybe it's because i'm not too familiar with multithreaded programming. I did find this:
I wonder if "special hardware instructions for better performance" means the OS can't interrupt it. (Since performance decreases every time the OS interrupts the thread). [/B][/QUOTE] I wonder how the atomic synchronization clause works on these kernels/kernel extensions. ------------------ "Oh, bother," said the Borg. "We've assimilated Pooh." Any fool can know, the point is to understand. -- Albert Einstein My Programming and Hacker/Geek related Blog [This message has been edited by CPUFreak91 (edited December 12, 2007).] |