Cube processing performance

I am facing a unique problem. we have three environments setup for datawarehouse dev,testing,production.

we have our own customized application to process cubes. When I process the cubes in dev it is fast when compared to QA environment and QA is faster than prod.

I conducted a test in all these three environments with same rows of fact table and dimension members. The only difference is dev Analysis Manager db size is 1 GB, QA has 35 GB and Prod has 55 GB of data. The data has been split into many partitions.

The processing of 6 million rows in fact table takes 15 mins with 12 dimensions in prod where as in dev it takes 6 mins.

Any idea why cube processing is slow in QA and Prod while faster in dev.

Are there any machine config differences?
for eg: Processor,RAM.
Also check the OS version and Analysis server service pack level?

Thanks for replying..Below are the details....

Dev and QA db are on the same machine.

DEV and QA Environment
======================

OS : Windows 2000
RAM : 4 GB ( 3 GB swtich not enabled)
Analysis Server Service pack level : sp4
Data stored in local hard drive
CPU's : 4


Prod Environment:
==================

OS: Windows Server 2003
RAM : 4 GB (3 GB Switch enabled)
Analysis Server Service pack level : sp4
Data stored in sans drive
CPU's : 4