Basically Priority Scheduler simply improves the availability of CPU resource to a higher resource group.
If a box is running at 100% when a single low power user is logged on, when a higher power user logs on it will have more bites of the CPU.
To give a basic illustration of this, low power user will be served by the processors and then have to wait 1 second to be serviced again and the higher power user will only have to wait say for one tenth of a second before he's processed again. (I know the figures are not realistic but they serve a purpose). This can be illustrated in a far busier system by allowing the higher power users to jump the queue for CPU resource.
This means that in each second the low power user will receive 1 bite of the CPU power and the high power user 9 bites ...spool and temp space are not released or reduced. It could be that if the low user is already using all of the spool, then as soon as the high user needs some spool he'll get the 'no spool available' error and be aborted - if the low user was first to ask for more spool he'd be the one to suffer!