more than 1 processor for CATIA

[jrochte]

I have CATIA V5 R15. Will having two processors do any good with CATIA V5 R15? Is the software setup to accept multithreading?

What about hyperthreading? Does having hyperthreading enabled help when working in CATIA V5 R15?

 

[solid7]

Since your thread is vague in the asking I don't feel like I'm hijacking it to take your question to another level. In addition to what jrochte has asked, I would also like to know if dual processors in 64-bit operation is considered "hyperthreading", or if it's seen as one processor. (64-bit Catia on x64)

I have recently acquired an Intellistation A Pro with 2.2 GHZ proc, and 4GB RAM. I have 2 of these machines, and will be running SMP once my second processor option board arrives. This will be a mirror of the first machine - 2X 2.2 GHz proc, 2X 4GB RAM, etc. Even if it's not double throughput, it will still be useful for running multiple session, or multiple intensive applications.

BTW - off topic, but does the second processor need its own graphics adaptor?

---
CAD design engineering services - Catia V4, Catia V5, and CAD Translation. Catia V5 resources - CATBlog.

 

[jrochte]

Im sorry, I want to know if the software CATIA V5 R15 will take advantage of 2 processors instead of just using one and leaving the 2nd idle. Is it designed to multithread. I heard talk about some applications not being designed to multithread. Is this one of them?

 

[PeterGuy]

DMU and the Caching process can make use of multiple theads as can the FEA mesher & solver. I have a recollection that the MC'in workbenches do but I am unsure.

The rest as far as I am aware in uniprocessor.

However, I am a fan of SMP / dual core machines as they enable you to still be productive (on some other task email, calcs, etc.) whilst Catia has tied up one processor /core thinking about something hard.

In the windows SMP world you do not need as seperate graphics adaptor one will be fine.

There are different ways of implmenting SMP at a board level as far as I am aware intel use a shared memory model (both cores address the same memory through the same bus) where as AMD has seperate memory buses for each core. But it is a while since I took an active intrest in achitecture so I could be wrong with current processor.

Dual 64 bit processor will be treated in the same way dual 32bit are. Again as far as I am aware hyperthreading is a additional CPU instructions for assisting in the priotising & processing of other CPT instructions.

 

[solid7]

So the memory and processor utilization are status quo in 64-bit Catia? That is, DMU and elfini only?

---
CAD design engineering services - Catia V4, Catia V5, and CAD Translation. Catia V5 resources - CATBlog.

 

[PeterGuy]

Not exactly

The SMP behaviour between 32bit & 64bit is the same with DMU (includes caching for assembly workbench) and FEA making use of multiple processors as these bits of catia are muti-threaded.

Execution of 64bit Catia is likley to be slower that 32bit Catia as 64bit uses 64bit pointers and these are longer the 32bit pointers and therefore require more processing to manipulate them.

I wote a reasnoble explanation of Catia memory behaviour in this thread

http://www.eng-tips.com/viewthread.cfm?qid=158538&page=1

 

[JacobL]

In general, only "behind the scenes" operations are multi-processor enabled.
The same goes to any 3D package, like maya, XSI and 3dmax, where the rendering uses as many cpu's you have, but the workflow uses only one, with few exceptions.
the possible advantage in multi-cpu system in this case is the ability (in some cases) to seperate the applications on a cpu level - CATIA on one, UG on the other, and run them in 2 monitors on the same workstation

multi-cpu shines in 3D rendering, video/audio encoding, CAE, and all other situations you send the computer to work while you wait.

 

[jrochte]

JacobL, Can you be a little more specific in behind the scenes? Give a couple examples, please.

 

[JacobL]

"behind-the-scene" means all the operations you don't really interact with.
In cad it could be automatic feature recognicion, in FEM it's the analysis part (not the modeling), and the same goes to CFD and rendering.

all interactive operations are NOT multi threaded. The iea os multi-processing requires certain algorithems which benefit form the division of labor.