Why Linux?

Hi All!

I’m a complete newbie to parallel programming, I’ve recently discovered the world of CUDA and its a Fascinating place!

I’ve got a home project which i’m eager to start, but before i can do that I need to set up my environment…

Starting from the bottom up i have a Linux question - what are the benefits of programming CUDA for linux? is it the small memory footprint of the OS, or are there additional benefits?

I came across ‘Yellow Dog Linux - for CUDA’ - it looks like the last release was built in 2010, but does anyone have any experience with this?

Thanks all - I’m looking forward to getting my hands dirty in some CUDA code soon!


If you value your time, don’t do it.

CUDA can be programmed in Windows as effectively as in Linux.

Linux has no benefits other than it’s “free” (some people mistakenly believe their time doesn’t cost a nickel) and it has no malware (because malware authors aren’t interested in a platform which is run by 1% of people in the world - where most of them are IT specialists).

All other benefits are quite questionable (“more secure”, “more stable”, etc.)

As to why Linux is less desirable - read this page.


nonsens, i work with CUDA on openSUSE 12,2 Linux , before on Windows 7, CUDA is effectivley, stable, and faster on Linux as on Windows, we have change in our engineering office from Windows to Linux on all Servers and Workstation, the reason, costs, stability, performnce since 2005 we work with Linux and all developer enjoy Linux

Ha… So no clear winner! :)

Thanks for your answers birdie and bernasek, if there was a clear winner it would have persuaded me, but I think I will start on windows for now Since i already have it… Maybe if I upgrade my system I’ll have a dabble in Linux then.

Right, I’ll get my environment setup and start getting my cuda on…!