Community
Inventor Forum
Welcome to Autodesk’s Inventor Forums. Share your knowledge, ask questions, and explore popular Inventor topics.
cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

Can I get more processing power with a Tesla GPU

13 REPLIES 13
Reply
Message 1 of 14
Anonymous
4828 Views, 13 Replies

Can I get more processing power with a Tesla GPU

I use Inventor tube and pipe heavily for my job, and work with very large models that incorporate architectural data imported from a Revit file, as well as using Navisworks (both are 2014 version).  Since Inventor doesn't support mutiple cores on a muti-core processor (except for when rendering in INventor Studio), would using a NVIDIA quadro and Tesla card combo would give me more processing power.  I know there are other programs, (3dsMax, Showcase, FEA, and CFD) that can off load some of the processors number crunching to the Tesla GPU. But I haven't been able to find anything as to whether Inventor or Navisworks cand do that under there "normal" modes.  Any info would be greatly appreciated.

Tags (3)
13 REPLIES 13
Message 2 of 14
mcgyvr
in reply to: Anonymous

Not sure.. I kind of doubt it will work.. I don't even think Inventor supports SLI/Crossfire video card setups yet..

 

Whats the rest of the specs on your computer.

CPU/Memory/OS



-------------------------------------------------------------------------------------------
Inventor 2023 - Dell Precision 5570

Did you find this reply helpful ? If so please use the Accept Solution button below.
Maybe buy me a beer through Venmo @mcgyvr1269
Message 3 of 14
blair
in reply to: mcgyvr

I'm with MCGYVR on this one. Inventor. Just get the fastest single GPU and single CPU processor for Inventor model environment. The only time any additional CPU and GPU's get used is for FEA and Studio.

Inventor 2020, In-Cad, Simulation Mechanical

Just insert the picture rather than attaching it as a file
Did you find this reply helpful ? If so please use the Accept as Solution or Kudos button below.
Delta Tau Chi ΔΤΧ

Message 4 of 14
Anonymous
in reply to: blair

So what is the final verdict about this? Will a Tesla help inventor? 

 

I am in the process of upgrading our drawing office workstations and have 3 options

 

  1. Quadro K6000
  2. SLI 2x Quadro K5200
  3. Quadro K5200 with The Tesla K40 Accelerator

 

All 3 options cost about the same but I am not sure which will really have the right choice. From this post I assume that the Quadro K6000 is the only real choice. Is that true?

 

 

Message 5 of 14
Anonymous
in reply to: Anonymous

I've recently (last week) purchased a new workstation for FEA.  I created a similar thread on here, and received a response from the Autodesk support team:

 

"Hello Curtis,

I wanted to call you to discuss but there was no tel number listed in your profile. I have inquired for opinions on what your last message asked. The responses I have gotten back are from core Inventor team, and the Simulation Mechanical team.

Core Inventor's take on GPU: at this time there is no code that will take advantage of this. With that statement there is no need to purchase one. I would think there would be some type of technical announcement when this comes available.

Autodesk Simulation Mechanical take on GPU: From contact within that support group they had mentioned one of their solvers was able to utilize this but it was linked to causing issues.

In my humble opinion, with the statements above, I would not bother purchasing one of these for our systems. If you need to speak with me regarding my statements please provide a number or let me know this is enough information is enough and I can close out this case.

I try to put myself in the clients shoes and I can't see any bang for your buck here. I hope this helps.
 

Best regards,

Don Warner
Autodesk Support Team"

Message 6 of 14
brian.cranston
in reply to: Anonymous

With respect to Inventor only, it's all about the gigahertz.  I would look at an i7-4790K which will push up to 4.4GHz.  However you are limited to 32GB of DDR3-1600 ram.  You might want more if you are dealing with Inventor and other software concurrently? 

 

If you need more ram, look at a XEON E5-1630 V3 (or 1650, 1680).  You top out at 768GB ram but you only peak at 3.8GHz. There's always tradeoffs, right?

 

I think view creation is multi-threaded too.

 

-Brian

Message 7 of 14
sam_m
in reply to: Anonymous

AFIAK, the statements from Autodesk in the past have been, and I'm yet to see any press to contradict them:

 

1)  SLI isn't used.  Inventor only uses 1 gpu, so there is no benefit with SLI.  Those dual gpu cards like the ATI R9 295X2 might be ok, as long as they are reported to Windows a single device in Device Manager.

 

2) there is no gpu computing used within Inventor, so an extra gpu for that will not be used.  The only statement I know of from Autodesk regarding GPU processing was back in 2009: http://investors.autodesk.com/phoenix.zhtml?c=117861&p=irol-newsArticle&ID=1301645 which announced CUDA support for Moldflow.  I think there was also a similar statement at the time that Ansys FEA would get CUDA support too.

 

So, as of the moment, a Tesla card or SLI setup would pretty much be wasted for Inventor.

 

Whether there will be any changes in the future, no idea, I hope so...  I thought DirectX 11 had some gpu computation support natively, so I guess/hope they could aproach it via that, but as the main program is predominately single-threaded, I would only guess FEA and Studio/raytrace rendering would benefit.

 

While talking about graphics cards - why have you gone Quadro?  Again, I believe the general thought these days was they're a waste of cash.  Inventor is now DirectX and Quadro excel in OpenGL.  So, unless you're using other applications that need an OpenGL gpu then it would be FAR cheaper to stick to gaming GeForce cards.



Sam M.
Inventor and Showcase monkey

Please mark this response as "Accept as Solution" if it answers your question...
If you have found any post to be helpful, even if it's not a direct solution, then please provide that author kudos - spread that love 😄

Message 8 of 14
Anonymous
in reply to: sam_m

Thanks everyone for the inputs. 

 

We are going for ECC RAM, Xeon E5 processor and Quadro k6000 for the same reasons. Stability. It might seem like a waist now but if it can reduce down time and allows the team to focus on the work then it is a win. At the end we sell design hours and if we can get more done with the same people then it is a no brainer. 10% improvement over 5 years counts up to a half a man year. That is double the initial investment.

 

We only upgrade every 4-5 years and in my country we can claim back the cost of such hardware over 3 years against tax. So in 4 years’ time I have the money back to spend again.

 

So at the end it is an investment that pays for itself and more. Making us more competitive at no real cost. 

Message 9 of 14
dgorsman
in reply to: Anonymous

Hate to torpedo your boat, but server-style workstations aren't more or less stable than conventional computers.  Used to be that way waaaay back in the day before playing games was considered a career (... still can't wrap my head around THAT).  But these days there are gamers who push their hardware faster, harder, and longer than *anything* you will be doing, and they do it with relatively sedate hardware.  Thanks to that and vastly improved quality controls there is virtually no difference between a "workstation" and "gaming computer" in terms of stability for desktop applications.

----------------------------------
If you are going to fly by the seat of your pants, expect friction burns.
"I don't know" is the beginning of knowledge, not the end.


Message 10 of 14
blair
in reply to: dgorsman

I agree with the other "EE's" and there posts, spec'ing our ECC memory and Quadro cards is a waste of money. The difference between Quado cards and GeForce cards comes down to more memory on the Quado card and better testing certification with OpenGL. Since Inventor went away from OpenGL over 7 releases ago there is not need.

 

I used to spec my machines with Quadro cards but after doing some bench tests (you could select either OpenGL or Direct3d within Inventor) I found the Direct3D to be slightly faster with the same model on the same machine.

 

ECC memory is great for "mission critical" servers that require a 100% up time. You can cut your systems costs in half by spec'ing out good "gaming" PC's and not notice any difference in performance at half the cost.


Inventor 2020, In-Cad, Simulation Mechanical

Just insert the picture rather than attaching it as a file
Did you find this reply helpful ? If so please use the Accept as Solution or Kudos button below.
Delta Tau Chi ΔΤΧ

Message 11 of 14
dgorsman
in reply to: blair

One of the few reasons to go with a workstation is when you need to source truly large numbers of computers, like our business unit.  In that case you'll be going with one of the big vendors like Dell or HP and they only source those kinds of numbers with their workstation lines.  Bulk and corporate discounts along with support credits bring the numbers down to competetive but for short runs the gaming vendors are a better choice, and for single or very small runs self-build is even more cost effective.

----------------------------------
If you are going to fly by the seat of your pants, expect friction burns.
"I don't know" is the beginning of knowledge, not the end.


Message 12 of 14
Anonymous
in reply to: Anonymous


@Anonymous wrote:

So what is the final verdict about this? Will a Tesla help inventor? 

 

I am in the process of upgrading our drawing office workstations and have 3 options

 

  1. Quadro K6000
  2. SLI 2x Quadro K5200
  3. Quadro K5200 with The Tesla K40 Accelerator

 

All 3 options cost about the same but I am not sure which will really have the right choice. From this post I assume that the Quadro K6000 is the only real choice. Is that true?

 

 


 

What will be the primary purpose of these machines?  That's the important question.

 

Will they be used for Inventor solid modelling/drawings? FEA? Renderings? 

 

And the Quadro K6000 is way overkill for Inventor, from the research that I've done.

Message 13 of 14
blair
in reply to: Anonymous

Tesla won't help Inventor as it's not written to take advantage of it. Software such as 3dMax will take advantage of aditional cores, both CPU and GPU as you can select which cores you with to use.

 

I would look at option 4

 

http://www.dell.com/us/p/alienware-area51-r2/pd?oc=dpcwc02h&model_id=alienware-area51-r2

 

 


Inventor 2020, In-Cad, Simulation Mechanical

Just insert the picture rather than attaching it as a file
Did you find this reply helpful ? If so please use the Accept as Solution or Kudos button below.
Delta Tau Chi ΔΤΧ

Message 14 of 14
Anonymous
in reply to: blair

Hello group,

 

The last time I checked the date this morning its 2018.

 

Can Nastran In-CAD support Tesla Cards for its linear analyses?


We're in the process of upgrading our machines and this is my only chance to get a graphic card.

 

Thanks all.

 

Regards,

Kenneth Gomez.

Tags (2)

Can't find what you're looking for? Ask the community or share your knowledge.

Post to forums  

Autodesk Design & Make Report