Computer/System Requirements to run a large inventor assembly >10,000 parts

[Anonymous]

I am working on this project in inventor and it requires alot of time to open and move around.  Would it be within my best interest to get a more powerful computer, or is it not even worth the money.  Right now i have a DELL Intel(R) Xeon(R) CPU E5-1620 v2 @ 3.70GHz 3.70 GHz:  16GB OF RAM:  WINDOWS 7 OS.  Graphics card is a NVIDIA Quadro K2000. I also work over a network pulling parts from different locations.  Please help me with this issue. Thank you.

[wrw0007]

Cody,

 In my experience better computers are always worth it! ... To what extent, I cannot answer.

Just some reccommendations.

When working with large assemblies it is always best to work within your local machine whenever possible, work in a single root directory rathe than multiple locations, create level of detail representations, and utilize express mode as much as possible.

Another big memory comsumption is usually general workflow. The more efficient the part modeling is, the smaller the memory load from the assembly, although I understand that is a difficult task when using large assemblies.

Also, removing textures, lighting, shadows, edges, and reflections will help.

Which of these takes priority over the rest? I'm not sure, but In my case If I were working with large assemblies, I would put all of them to practice as much as I could.

Maybe another poster can help us realize which is the most important for memory reduction.

[mslosar]

I can attest to networks slowing the process down - we're fighting that now.

As for your machine - it seems fairly well off.

How big of an assembly are you working with? Here, we're having network issues, but, our sizes can be in the vicinity of 1500 unique parts and 14000 total parts. Our machines sound similar to yours, except we've got 32GB (memory usage never passes around 3-4GB) and quadro 4000's..

I think (fingers crossed) that we're going to see a dual processor test machine, but I'm skeptical if it'll help much. The multithreaded aspects of inventor aren't many...

We do notice a large speed increase working local though. Problem is, we've got multiple people working the same project so local is hard to do with that constraint. We're getting vault, but it's still likely a couple months off at best.

[innovatenate]

Hi codyhetke,

One thing you may consider testing is how much impact the network has on performance. You may consisder using the save as > pack and go command to move the data to the local hard drive, and quantify the impact to performance in Autodesk Inventor. A data management solution like Autodesk Vault will enable collaboration over the network while keeping data local to the hard drive. 

Another suggestion is to ensure that all of the reference files have been migrated to the latest released. Further information about migrating data with the task scheduler can be found in the below help topic.

http://help.autodesk.com/view/INVNTOR/2016/ENU/?guid=GUID-873B3D0D-E1D7-4CF3-BFC3-1956B43389F2

If the graphics drivers on the machine are older than six months, it may be a good idea to update the graphics drivers. You can find the latest driver for the Nvidia graphics card on the manufacturer's website, below.

http://www.nvidia.com/download/scan.aspx?lang=en-us

There are also some large-assembly specific recommendations in the below link that may help to optimize the performance of Autodesk Inventor.

http://knowledge.autodesk.com/article/How-to-optimize-Inventor-for-working-with-large-assemblies

Note that the PDF attachment in the above link discusses hardware requirements for large assemblies in better detail. The handout also discuss many different strategies for managing level of details and simplified parts in Autodesk Inventor which will greatly influence the performance of Autodesk Inventor.

I hope you find this information helpful. Please let me know if you have any specfic questions.

Thanks,

[cadman777]

Thanx for that info, Nathan.

Would you be able to find out fro me if Autodesk developers intend to make the modeling app multi-thread, such as w/iProperties and other non-mathematical aspects of this program?

One of the biggest time-uses is changing the BOM at the end of a project. When the model is big, and there's an unimaginable amount of metadata, changing just one BOM entry can take 3 minutes to update. Sometimes it can take 15 minutes.

I have quite a good system for managing the subassemblies and parts, but there is a mega-bottleneck due to lack of multi-threading.

Also, since Intel refuses to work towards faster frequecy processors, we're all stuck in the 3.5-4.0 GHz range, which means we waste all the other cores and threads of our high-tech multi-core processors. Why buy multi-core if you don't use it 99% of the time? Oh, that's right, you can't buy single-core processors!

Please advise ...

[dgorsman]

AutoDesk like other developers cannot publicly comment on future development plans.  Thems the rules for a publicly owned company.  There *is* a wish for implementing multi-threading in the IdeaStation, so vote if you haven't.  But have a look at the various posts I've made on the concept - it's not a magical wand to turn a pumpkin into a muscle car and simply may not be a viable solution.

"Also, since Intel refuses to work towards faster frequecy processors..."  Ermmmm..... no.  There are several technical limitations which get in the way.

[cadman777]

dg, thanx for your opinion.

Hmm ... when I was growing-up, we called that 'making excuses' and 'passing the buck'. If I acted like that at work, I would be on the bread line.

When is someone in the industry going to take responsibility for these basic functions and give them to us?

After all the sales hype, wouldn't it be a good thing to follow-thru w/some benefits to those of us who hammer the keys all day long?

[SteveMDennis]

Chris, dg is right, it is the government and legal system that prohibit us from commenting on future releases much. Believe me it is not passing the buck.

As to multi-threading and multi-core usage we are constantly looking for improvements in these areas.

We cannot simply wave a wand and voila Invetnor is MT or MC. MT/MC is a very complex problem when it is not designed into a software system from the beginning and in fact can do more damage than good. Debugging them is a MAJOR PITA. There are documented cases where if you take an algorithm and make it MT/MC you can actually slow it down! So we pick and choose our work in this area very carefully and you are unlikely to open up the next version of Inventor and go WOW! But every year for the last several we make small imporvements (like in boolean performance or drawing view performance) where we get isolated safe improvements in performance due to MT/MC.

I know it is not the answer you want but I am trying to give you a small peek under the covers.

[Anonymous]

I've got the exact same hardware you've listed with the addition of an 240GB SSD.

We do however you Vault so I'm always working locally.

I would love to go to 32GB RAM but trying to justfify that to upper management isnt a joy.

CPU I really wanted to bump up but after doing some CPU Benchmark comparison, I didn't feel I was going to get the most for my money at that time.

I'm working in assemblies with 2-3k parts and it can get slow here and there but for the most part I'm pretty happy with the setup.

[Anonymous]

---

@Anonymous wrote:

...I would love to go to 32GB RAM but trying to justfify that to upper management isnt a joy.

---

Just inform them that they spend more money on coffee stir sticks in a year, than the minimal cost to upgrade your RAM...

[swalton]

I've got a Dell with a Xeon e5-1620, 16 GB RAM and a SSD too.  Up to 5k component assemblies work well.  It can get slow after that.

The times I want/need more RAM are multi-sheet idws of 5-30k component assemblies and when I run FEA in one IV session and do separate design tasks in a 2nd one.

When IV is slow, Task Manager shows that the CPU is maxed out.  If it is a single thread type operation, I see about 13% cpu (one core's worth).  If it is a multicore operation (FEA, Background view computes, etc) I see 100% cpu.

[wrw0007]

As an engineer currently using 32GB ram, simultaneous 2mil point CFD, FEA, Modeling, and Drafting barely makes my computer break a sweat. I whole-heartedly vouch for the upgrade. Tell your upper management to do it because I say so. i'm sure that will smooth out the bumps?