Hi all, I am facing a problem to simulate some 3D problems with huge grid, I am using Pentium 3 with 1 GB ram, it takes long time for the sloution to converge. Can someone tell me what can I do, should I jump to new Xeon workstations, or those based on unix operating system. Thanx in advance.

We would really need to know your definition of 'fast' and your price range (as well as the speed of your current Pentium 3).

I will assume you are using Fluent and have a parallel version (capable of more than 1 processor). 1) Cheaper methods: 1a) You will get good performance with a dual 1.53 GHz AMD MP workstation. This would cost about $3000-$4000. 1b) You can also get good performance with a dual 2.0 GHz Intel Xeon workstation (costing about $4000-$5000). 2) More expensive methods: 2a) You can buy a cluster of less expensive computers and link them together in Linux to get good performance. A cluster of 20-50 computers would probably be the best price/performance ratio. A reasonable cluster might cost $20,000-$50,000. 2b) Unix machines with large numbers of processors. This would cost a minimum of $30,000 to beat any of the computers listed above. $100,000+ is easy to spend for the best performance possible.

3) Future computers. If you are willing to wait a few months some new processors will be out. 3a) The first AMD 64 bit chip (code named Hammer) will be released soon, there is a press release about it today. This has the potential to rival the speed of the Unix machines yet have a much lower price tag. Who knows, this also may be a terrible chip. 3b) The server version of the Intel Xeon will be released in the beginning of next year. This might allow four or eight 2.0+GHz Xeons in a single machine.