Hello everyone,

I am trying to run software that I've developed, which is highly memory-intensive, on a version of Linux (Red Hat 9) that offers Large File Support, with a 32-bit machine. Is this feasible, or is LFS intended only for a 64-bit processor?

Please advise.

JME

Define large

ext2/3 can do LFS for files 2e63 bytes long on 32 or 64 bit platforms.

Thanks for your reply. By large, I mean the ability to process around 10GB, perhaps more. I'm manipulating CT data that is 512 x 512 x 256, .5mm resolution, as well as MR data resampled to be the same, and my processing (a combination of "level set" surface models and Bayesian classification), involves random accesses to individual volume elements, or voxels.

One more thing: I need not only large file access, but the ability to process volumes at run-time via swap spaces of large (there's that word again...) size. LFS should translate into the ability to use swap spaces of that order of magnitude (10-20GB), should it not?

I won't pretend to know what you're talking about here, but 10GB shouldn't be a big deal. I have personally created tar files over 20GB.

Again, this shouldn't be a problem. I really don't think you want to swap 10-20GB of data at a time. Your application is going to be abysmally slow, if it runs at all. Chances are the wait states will rise to a boil and the whole machine will die in the tarpit known as swap.