Autodesk Inventor

[#1]

Here are the two arguments you will find whenever you search this topic:

Quadro because that’s what Autodesk says

Geforce because it’s cheaper

So after days of endless research and testing I figured Id help those looking for this answer by posting my experience and test results with both.

In a design department of 12 guys who work with 5000 to 10000 part assemblies (Using Inventor 2012) usually designing a machine around some complex “Surface” part in Inventor we run Nvidia Geforce  GTX 560 Ti video cards except for 1 designer who has a Nvidia Quadro FX 4800. He was convinced all his problems stemmed from the Geforce video card and even after switching it out he experiences the same few small issues that everyone else does, (See signature for rest of system specs) which include Inventor crashing once in a great while or the video card driver stops responding. The drivers crashing I believe to be an issue with Windows 7 only because it happens even when the designers are not using Inventor.

Using the EngineMKII.iam I ran the Inventor 6 benchmark switching out video card  here are the results:

Single Nvidia Geforce GTX 560 Ti

Shaded = 124.5 FPS
Hidden = 36.8 FPS *
Wireframe = 58.8 FPS *
Multiple Windows = 195.3 FPS *
Average = 103.9 FPS *

Single Nvidia Quadro FX 4800

Shaded = 130.9 FPS
Hidden = 27.2 FPS
Wireframe = 41.6 FPS
Multiple Windows = 134.7 FPS
Average = 83.6 FPS

Dual Nvidia Geforce GTX 560 Ti (SLI)

Shaded = 141.3 FPS *
Hidden = 31 FPS
Wireframe = 49.8 FPS
Multiple Windows = 165.8 FPS
Average = 97 FPS

With these results, countless days of research, and my small case study here is what I can conclude and recommend:

Inventor is DirectX not OpenGL and Geforce is for DirectX 

Using the Benchmark test for FPS Geforce is the winner 

Cost Geforce GTX 560 Ti ($250) Quadro FX 4800 ($1500)

Go Geforce and take the money you saved and put it into a processor 

If you have an questions feel free to ask 

Thank you

[#2]

Nice work, It confirms my opinion at least. I don't think its just about frame rates though (i remember reading about this about 18 months ago), I'm no expert and can't remember what the other considerations are. But a Quaddro advocate would surely fill us in. Every single one of the 11 PC's I had built for my design office have GeForce cards, I put the money saved into SSD's, CPU & triple channel memory. I think the SSD's make the most difference.

cheers for posting your efforts.

[#3]

Thanks, I agree as well. I used to be a Quadro fan, back when IV used OpenGL. The move away from OpenGL has allowed for the use of cheaper "Gamer" cards.

The slowest segment of your system is the hard-drive. Anything you can do to increase the performance with show up in program load times as well as any data stored.

The latest HD's in the SATA-III format will produce read times in the 500-550 Mb. It really speeds up your system boot and program load times.

[#4]

The decision for some gas been muddied for some with the advent of the suites, I believe inventor is the only one which uses direct x, the rest are still open gl. But again they run just fine with the GeForce cards which support open gl to a certain degree anyway

[#5]

Your research is very wrong you are comparing a gtx 560 ti fermi based architecture that supports direcx 11 and it supports tessalation (http://www.nvidia.com/object/tessellation.html) on a benchmark against a very old fx 4800 which only supports directx 10

the lowest profile quadro card you can compare is gtx 560ti is quadro 600 which is a directx 11 card

here is some benchmarks

NVIDIA Quadro 400 VS GTX560ti

https://www.youtube.com/watch?v=phZ3XMfzL5I

and here is quadro 600 vs quadro 400

https://www.youtube.com/watch?v=DpUNa_C2ldE

and quadro 600 is only $150 US dollars

[#6]

---

ComputerGuru666 wrote:

Go Geforce and take the money you saved and put it into a processor 

---

Search for old post here (written about the time of switch from OpenGL to DirectX) by Norbert.  (find the "lengthy" response)

[#7]

Please See the below link and then help me understand what your trying to say?

http://www.videocardbenchmark.net/high_end_gpus.html

Please see that the Quadro 6000 at $3999.98 scores 3067

The Geforce GTX 560ti at $199.99 scores 2985

Your Quadro 600 for $119.99 scors a whopping 706

"Your research is very wrong you are comparing a gtx 560 ti fermi based architecture that supports direcx 11 and it supports tessalation (http://www.nvidia.com/object/tessellation.html) on a benchmark against a very old fx 4800 which only supports directx 10"

Right in your quote you are argue direct x compatability so here is a question what cards are made to run direct x apps???? (dont over think it) answer is GEFORCE! Question 2 is Inventor 2011 and newer an Open GL or Direct X app???? Answer DIRECT X!

Finally help me understand the point of your entire post?

Thank you

[#8]

*Double facepalm* did you even watch the link I posted????

You are totally misguided, the benchmark results and tools used that you are showing are optimized for real-time rendering/computing for games and processing.


OFF Topic:

What is the fastest nvidia GPU Card?

Answer:

GTX 690

Yes it even has a lot of Cuda cores but still the computing/simulation process of a Geforce and Quadro is very different.

But why is the gtx 690 did not rank higher than other gtx 600 series  according to your link http://www.videocardbenchmark.net/high_end_gpus.html it’s because this website is ranked to “Price and Performance” “”ratio”” so even if gtx 690 ranked 9999th placed it doesn’t mean its very slow. I know this site because I also personally use this site whenever I plan to buy a Geforce card to render videos using cuda

So even if we are comparing geforce cards for example like:

scores

gtx 690  = 3,537
gtx 680 = 4,033

 just because gtx 680 got 4,033 points and ranked 2nd as of October 2. 2012 and gtx 690 scored 3,537 and ranked 9th  it doest mean gtx 680 is faster and everyone knows for a fact gtx 690 is faster than 680. Again the website averages a “Price and Performance” “”ratio”” to score……..

About the directx 11 you said let’s say you have gtx 295 a directx10 card vs gts 450 directx11.

Gtx 295 is a very very fast and older card and it will beat the newer series gts 450 on a directx 10 software but if the applications requires unique tessellation computing that exist in direct11 card the gtx 295 will surely lose an or wont even run it it doesn’t have the technology how to approach it.


same situation like

quadro fx 4800 is faster than quadro 600 but different architecture certain features that quadro 600 has that fx4800 doesnt. so its not always about specs.. it really depends how you utilize your workstation old card to a newer software that is not fully optimized because of older cards, vice versa

 
Back To Topic

 
what do you mean they don’t use directx??? A lot of windows applications use directx even some simple windows application uses a simple api like directdraw uses directx. Autodesk products even requires you to install/to have directx runtimes before you are allowed to install any autodesk product….


If you want to make a proper, professional benchmarking opengl for a workstation then go to this site:  http://www.spec.org/gwpg/gpc.static/vp11info.html


And here are some benchmarks results I googled for your gtx 560 ti


Quadro 600




GTX 560 ti




2X GTX 560 Ti SLi

 

http://mbah.net/2012/04/08/specviewperf-11-benchmark-results/


Even a Dual gtx 560 ti that costs $450+ barely beats a $150 quadro 600 card..........


and again watch these videos because it recreates the performance/simulation of your graphics card in the viewport of professional software like autodesk


https://www.youtube.com/watch?v=phZ3XMfzL5I
https://www.youtube.com/watch?v=DpUNa_C2ldE

 

[#9]

Oh god, not this again...  Thought this "lifelong" topic had been put to bed ages ago (with many a topic suggesting gaming > workstation for Inventor, and Nvidia > ati)

Think I need to do a facepalm...

esponert105 -  I understand what you're saying but, by your own admission, you're using an OpenGL benchmark there.  You deem it worthy to mock the OP by suggesting the 560ti isn't comparable to the fx4800 but are then are quoting benchmarks which are irrelevent for Inventor - and we're all here for Inventor-specific information.  If you're going to outright claim someone's work is wrong then at least research the history and requirements of Inventor first..

Sure, for OpenGL apps like SolidWorks then I would agree that a Quadro is ideal, but Inventor moved away from OpenGL something like 6 years ago and with that any arguements for Viewperf benchmarks went out the window.

Bottom line, there are no specific "workstation" DirectX gpu benchmarks that we can rely on - the best being Passmark's site because it lists both GeForce and Quadro cards together.  (PS I've never heard, or read, any suggestion it's graded as a performance/price basis - the price is there as a reference, otherwise all the "NA" priced cards would be grouped together.)

Whether a card is DX11 or DX10 in a comparison doesn't matter as Inventor is only DX10 (atm) - just the fancy tessalation etc wouldn't be used in Inventor.  Passmark's benchmark is DX9 by default (and Dx10 optional), afaik, so any DX11 features on a card which could boost its score in any benchmark wouldn't be used either - so (atm) it's fair to compare both Dx10 and Dx11 cards.  Also, CUDA cores and DX11 gpu-processing doesn't come into it either as Inventor doesn't use this and neither does Passmark (afaik).

As to whether to use a dual 560ti setup - from memory I believe AD have said SLI doesn't really benefit Inventor but I have a feeling some user's benchmarks have suggested a slight improvement.

Saying all this - There is an argument that driver support is typically better with a workstation card but the majority of us have been using gaming cards for years now generally problem-free (certainly no worse than with a workstation card).  As for warranty and life-expectancy - some claim gaming cards won't handle the constant demand of day-to-day use - personally think that's a questionable statement too as look at all the gamers worldwide - some gaming day after day during college holidays and those literally dying from gaming non-stop for 48+ hours straight, thus I'd say the hourly-demands of a decent gaming card are just as bad (if not worse) than a typical working week.

Sorry if it seems like I'm having a go at you, I'm just trying to suggest some of your comments might be a little out of place when considering the specific software we're using.

[#10]

Thank you Sam_M couldn't have responded to that better myself