Best for math heavy application, i7-975 or Xeon 5590?

WinuxPCWinuxPC
edited September 2009 in Hardware
Just looking for opinions here, and the question seems obvious, but I have an application that is just a number cruncher and need to to go as fast as possible. I am running a 975 at 4.2, but I am wondering if two 5590 Xeons in a workstation environment would be substantially faster. It seems obvious that it would but I just wanted to make sure that I am not thinking incorrectly.

The workstation build would be 1333 ECC DDR3 @ 24GB and the 975 would be at 2028 DDR3 @ 24 GB. Both would be on x-25 SDDs.

It seems that 2 HyperThreaded 5590s would have to pump out much faster results but when I look at the numbers, they each seem to have advantages over the other.

What would you do? Money is not an object here so please, be creative.

Comments

  • ThraxThrax 🐌 Austin, TX Icrontian
    edited September 2009
    Are you sure the application is SMP-aware? That is, are you positive that it will scale well out to 16 threads?
  • WinuxPCWinuxPC
    edited September 2009
    It will be single threaded but running 16 instances. I'm looking to take advantage of each core individually.
  • _k_k P-Town, Texas Icrontian
    edited September 2009
    If it scales up to the extra cores: intel for a dual socket board and amd for a quad socket board or higher.

    EDIT: You posted while I was typing. AMD. And if money really is not an issue.... 8 socket board with 6-core procs. This is under the assumption that more instances are better. Or get a bunch of VIA nano's, they do crypto faster.
Sign In or Register to comment.