We have a Dell quad core with 8 gig ram + 2 gig swap running red hat 5.0 and an HP core i5 with 8 gig ram + 10 gig swap running red hat 6.2. The core i5 keeps having memory allocation errors. It had the same errors when running fedora 16. The simms tested fine.

Both systems are running the same cam software with 2 to 4 users connected from pcs using realvnc at any time. The Dell runs without error with up to 8 users connected but the HP on a seemingly random basis has memory allocation errors with as few as 2 users.

Using tiger vnc on redhat 6.2 not sure what vnc is running on 5.0

The cam software is said to use 1 core per user.

The HP with red hat 6.2 seems to try and distripute load evenly across the 4 processors and has the errors.

The Dell with red hat 5.0 doesn't evenly distribute the load. 1 processor will be at %100 and another at 0.

Is there a way to confiqure the HP to work like the Dell.

cgroups allows you to classify workloads (i.e. processes) to particular subsets of available cores using containers. I use it to allow the monitoring process to keep running when a benchmark is maxing out the rest of the machine.
Should do the job for you. Documented in the (kernel) source tree - not sure if RH offer something on it as well.