Hello everyone,

I'm need to have 64-bit addressing, due to the size of my programs and especially data, but I have a Pentium 4 machine. Is it feasible to use a 64-bit distribution on a 32-bit machine, and if so, can anyone recommend one?

Best regards,

JME

I don't believe that's possible...

64 bit OSes are written specifically to work with 64 bit hardware...

What exactly are you doing that requires so much computing power?

It might work, but it wouldn't utilize the 64-bit areas of the system. It might even crash, I would upgrade or get a 32-bit distro.

Thanks for the replies. I'm putting together a patient-specific neurosurgery simulator, based on image processing on CT and MR volumes that are resampled to consist of isotropic voxels, and these are very large volumes of data.

I developed this software up to now on smaller volumes (128 x 512 voxels x 512 voxels) that had the original CT sampling, which was not isotropic (.489mm x .489mm x 1mm), but some of what comes next requires isotropic spacing (.489mm x .489mm x .489mm), which entails a huge amount of data (254 x 512 x 512 voxels). If not for this, then I have to rewrite my code extensively, I believe. I can't use isotropic voxels that are more sparse, because some fine bone structures are not resolved well then.

Cheers,

JME

Sounds like it's time to spend a few hundred bucks for a 64 bit machine. Better than re-writing code for 32bit - that's for sure. Keep up the great work!!

WOW, I wonder how many words in here I did not understand... . Sorry for my ignorance... lol

CT is computed tomography, a method for producing an image volume of a patient that displays bone and soft tissues, as well as air.

MR is magnetic resonance, which produces image volumes of a patient that displays different soft tissues with different intensities: for a certain kind of imaging protocol, grey matter in the brain looks grey, white matter looks white, the cortico-spinal fluid looks dark, and so on.

In order to use the two sets of data together for my tissue classification, I have to register them (transform the MR data to CT space), then redo the sampling of MR data so that it conforms to CT sampling. In general, the CT sampling is dense along the x (left-right) and y (front-back) axes, but less dense along z (top-bottom), whereas for some of the processing that I'm doing, I have to have them with the same spacing between samples along all 3 axes: isotropic spacing. Resampling them increases the number of total points that I have to process, which means that I'm increasing the size of the CT volume, in terms of # of points, and likewise with the MR data, which reflects the sampling of the former.

How's that?

Amazing, thanks for clarifying mate!. Guess it more for me