Off loading Stress Analysis to another machine

[vex]

Is it possible to have the FEA solver/server that runs with Inventor Professional (not Simulation) setup on another machine to either have the mesh off loaded to it and the solution done on that machine alone, or to utilize parallel processing via multiple machines?

I have a feeling that it's going to be a 'no', but that it is a feature in the more simulation oriented software offerings from Autodesk...

[swalton]

I haven't used this, but it might get you what you need:

http://www.autodesk.com/products/optimization-inventor/overview

I don't know if you have to do an optimization study or if you can send complex single study.

[vex]

Yeah, I've seen that... I am unaware if we currently have any sort of subscription to utilize that feature of the software or not. It seems like sort of an overly complex way to do what I would like.

[raviburla · Accepted solution]

Hi,

Inventor Professional / Inventor Stress Analysis does not support distributed computing. So offloading a model to other machine for solution purposes is not possible. However, on a given computer, Inventor Stress Analysis does support shared memory parallelization and the meshing and solving operations (which are most time consuming ones) will take advantage of the many cores in the modern computers.

Please let us know if you have more questions.

Thanks,
Ravi Burla (Autodesk)

[vex]

Really?  When I was reading the help file about this it gave me the impression that it would not take full advantage of the multiple cores since it's not hyperthreading.  Is this changed in 2015?

http://knowledge.autodesk.com/support/inventor-products/troubleshooting/caas/sfdcarticles/sfdcarticl...

[raviburla · Accepted solution]

Hi,

Inventor Stress Analysis leverages multiple cores based on a shared memory parallelization model. This has been added in 2013, the link you had pointed out also mentions this (quoted below).

"In release 2013 we added multi-core processing capability for
* Stress Analysis (no special options or settings are required)"

Please let us know if you have more questions.

Thanks,
Ravi Burla (Autodesk)

[LT.Rusty]

---

@vex wrote:

Really?  When I was reading the help file about this it gave me the impression that it would not take full advantage of the multiple cores since it's not hyperthreading.  Is this changed in 2015?

http://knowledge.autodesk.com/support/inventor-products/troubleshooting/caas/sfdcarticles/sfdcarticl...

---

Distributed computing = multiple cores in separate computers.

Multithreading = multiple cores in a single computer.

Inventor FEA does multithreading, but does not bring in other workstations beyond the one that you've already got running.

[vex]

Ah, I see. So the FEA solver and mesher utilize all CPU cores relative to the memory parallization model you referenced and does not follow the quote immediately following that?

"In order to fully benefit from multi-core processors, you need to use multi-threaded software. Unfortunately, Inventor is currently a single-threaded application.

On a dual-core computer, a CPU-intensive operation that uses 100% of the resources of a single-core processor will only use a maximum of 50% of the CPU for that same operation on a dual-core computer, and only 6% of each CPU on a 16-core computer."

[LT.Rusty]

---

@vex wrote:

Ah, I see. So the FEA solver and mesher utilize all CPU cores relative to the memory parallization model you referenced and does not follow the quote immediately following that?

"In order to fully benefit from multi-core processors, you need to use multi-threaded software. Unfortunately, Inventor is currently a single-threaded application.

On a dual-core computer, a CPU-intensive operation that uses 100% of the resources of a single-core processor will only use a maximum of 50% of the CPU for that same operation on a dual-core computer, and only 6% of each CPU on a 16-core computer."

---

FEA and rendering are fully multithreaded.  Modeling is single threaded.

[vex]

---

LT.Rusty wrote:

FEA and rendering are fully multithreaded.  Modeling is single threaded.

---

thanks!