[ztdep]

Dear friends:
I want to buy a workstation for the cfd parallel computation.
could you please give me some information about the performance comparison of xeon and opteron.
regards.

[bindesboll]

It depends on your software, availability of parallel licenses, memory configuration, model setup etc. Start reading these:

http://www.padtinc.com/blog/the-focu...-vs-intel-xeon

http://www.padtinc.com/blog/the-focu...-vs-intel-xeon

http://www.cfd-online.com/Forums/har...vy-bridge.html

http://www.cfd-online.com/Forums/har...-bandwith.html

http://www.cfd-online.com/Forums/har...eon-w3550.html

Best regards
Kim Bindesbøll

[CapSizer]

Originally Posted by bindesboll

It depends on your software, availability of parallel licenses, memory configuration, model setup etc. Start reading these:

http://www.padtinc.com/blog/the-focu...-vs-intel-xeon

Ah, lovely! Very nice demonstration of what some of us have been saying here, which is that you should start by counting the number of memory channels. Maximise that number, then worry about the rest ...

[flotus1]

Although I would subscribe to this conclusion, I dont like the fact that these results were obtained on two completely different systems (for example linux vs windows)

[CapSizer]

Originally Posted by flotus1

Although I would subscribe to this conclusion, I dont like the fact that these results were obtained on two completely different systems (for example linux vs windows)

Have you seen a significant difference between similar systems running different operating systems? When I went to the trouble of measuring that, I found no difference between Linux and Windows, although different compilers did make a significant difference.

[flotus1]

I was not saying that the huge difference observed in the "benchmark" came from the operating system.
If I had something to say in the last purchase of a workstation for our lab it would have been an opteron-based system.
It is just not a good practice to run such a comparison with different setups.