Sent a message to my var, got a somewhat answer back, so I
thought I would post here, I am taking some time to look at the
archives but thought I would ask anyway just in case someone wants
to offer info
IT is buying a workstation for Cosmos use specifically, they ask:
" Does Cosmos take advantage of dual processors? What about dual core and quad core, any advantages in any of these?" "What is the best setup of processor and ram?"
The answer from my Var:
"Yes COSMOSWorks does take advantage of the dual core. Here is the information from the KB article:
S-02691
COSMOSMotion and COSMOSFloWorks do not make use of Dual Processor or Dual Core.
COSMOSworks, on the other hand, does:
- FFEPlus will take advantage of Dual Processor only.
There is a improvement of 15% on using this.
- Direct Sparse solvers will take advantage of both multi Processor and Dual Core system.
There is improvement of 60% for this solver.
What is a Direct Sparse solver??
Thanks yall
IT is buying a workstation for Cosmos use specifically, they ask:
" Does Cosmos take advantage of dual processors? What about dual core and quad core, any advantages in any of these?" "What is the best setup of processor and ram?"
The answer from my Var:
"Yes COSMOSWorks does take advantage of the dual core. Here is the information from the KB article:
S-02691
COSMOSMotion and COSMOSFloWorks do not make use of Dual Processor or Dual Core.
COSMOSworks, on the other hand, does:
- FFEPlus will take advantage of Dual Processor only.
There is a improvement of 15% on using this.
- Direct Sparse solvers will take advantage of both multi Processor and Dual Core system.
There is improvement of 60% for this solver.
What is a Direct Sparse solver??
Thanks yall
I can't tell you the details between the three, but I can tell you that I run direct sparse almost all the time. I expect you should be able to take advantage of this as well, though there may be a benefit of one of the other ones for certain types of analysis.
Using direct sparse, I had a model than ran on our old P4 2.8 GHz system in 29 minutes, which reduced to 11 minutes on our new C2D 6600 system.