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: 

Benchmark for Autodesk Inventor 2012

102 REPLIES 102
Reply
Message 1 of 103
prusas
12150 Views, 102 Replies

Benchmark for Autodesk Inventor 2012

Autodesk inventor 2012 has worked exclusively on Direct3D. At the moment, there is no single test that is likely to correctly identify performance hardware.The results of such tests as Cinebench, 3DMark, SPECviewperf - untrue. But the need to determine the real performance of video cards is growing. According to this offer on their own to determine the performance of different video cards. For the test we will use the internal mechanism of Autodesk inventor 2012 definition of refresh rate.
One benchmark test that you can perform is to run the Post Statistics.

This requires an understanding of registry editing.
Go Start>Run & type “regedit”. Click OK & navigate to this branch: HKEY_CURRENT_USER\Software\Autodesk\Inventor\RegistryVersionXX\System\Preferences\Scene Manager\Debug\Post Statistics, set the REG_DWORD to 1.
The next time Inventor is run, a small information box will appear at the bottom of your screen.
 We twist the assembly . Pattern editing in the assembly "1000.iam", we change the number of parts in the assembly until the refresh rate will be equal to 15 Hz. This is the minimum, and psychologically comfortable refresh rate.We work exclusively in the factory settings
The results report in the next volume:
1.     The operating system and its capacity
2.    Processor
3.    RAM
4.    Video card
5.    Number of parts with assembly at a refresh rate of 15Hz.

Windows 7 64 bit
Intel® Core™ i7 CPU X 990 @ 3,9; RAM 12Gb x 1600
OCZ-Vertex 4 SSD (119.24GB)
Asus NVIDIA GeForce GTX 580; PLE2607WS 1920x1200
102 REPLIES 102
Message 81 of 103
smjanows
in reply to: smjanows

And comparing the old machine the the shiny new one.  On average there is a 35.2% improvement at the same parts count.  In reality, 35% better than a small number is still a small number.

 

Inventor 2012 Performance E8400 vs i7-3820.png

 

 

Message 82 of 103
karthur1
in reply to: smjanows


@smjanows wrote:

....Quality is a bit faster than performance.  Huh?  The smoother versus rougher settings make no difference.

 


When I was doing the tests, I found the fastest setting was "Compatibility, then Quality... Performance was actually the slowest. (look back on page 6 of this thread)

 

Odd that you dont see any difference in smoother vs rough.

Message 83 of 103
jeff.noel
in reply to: smjanows

Someone can correct me if I'm wrong, but I'm surprised you were able to get almost 100% CPU usage from Inventor.  Inventor isn't able to take advantage of multiple cores, so on a dual core machine, the most CPU usage you would see is 50%, a quad core would be 25%.  Did you have other applications running in the background, and were using the overall CPU usage instead of the CPU usage of the Inventor Process?

Message 84 of 103
smjanows
in reply to: jeff.noel

DX11 is a multithreaded API.  It's a driver related issue I think.  I've posted in the past about nvidia vs ATI cards on my old E8400 machine.  I think there's a driver related issue.  It seems nVidia cards scale better than ATI/AMD with CPU using Inventor.  But, I need to get a couple power adapters and throw the GTX480 into the HP to run some tests.

Message 85 of 103
sam_m
in reply to: smjanows

dx11 is multithreaded but Inventor is only DX10.1 (i thought), plus would guess the application would need to be programmed to take advantage of the cores (both cpu and gpu) instead of just letting Dx11 run wild (purely a guess tho).

 

not to mention Autodesk have stated that Inventor presently doesn't use more than 1 processor apart from FEA, rendering and from Iv2012 each view in an idw can use a separate process so a quad core can draw 4 views simultaneously. 



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 86 of 103
smjanows
in reply to: sam_m

A few more data points:  The GTX480 on the i7 and the 7950 and 5870 on the Core2Duo.  On the i7, 2 of 4 cores were fully utilized with the nVidia card.  That is to say that CPU usage per the Inventor process was 50%.  As you can see, the GTX480 is the best performer on both machines.

 

Inventor 2012 Performance.png

Both machines are running at 1680 x 1050 on the primary monitor.  The new machine also has a second 1680 x 1050 display.

Message 87 of 103
smjanows
in reply to: smjanows

Adding 1 more graphics card to the mix:  An old GeForce 8800 GTS.  You will be surprised with the results.

 

Inventor 2012 Performance.png

 

The 8800 GTS is holds its own against new cards once again proving the Inventor is severly CPU limited.  At the 15 fps mark of 600 parts on the i7, Inventor will max out 2 cores while manipulating the model.  The GTX 480 utilization was near 11%, and the 8800 GTS about 38%.  This shows that even an old graphics card has more than enough horsepower.  Save your money and invest in a big fast SSD and a liquid cooling system to overclock your CPU through the roof.  The GTX 480 datapoint on the old machine at 800 parts is the only anomoly and must have been entered incorrectly.

 

That's all the hardware I have to benchmark.

Message 88 of 103
mjmaes
in reply to: smjanows

Ok, I am not 100% sure I am doing this correctly but want to know as I just built a new computer.

Here is what I have

 

Intel i7 2600k (overclocked to 4.5ghz) 

Nvidia GTX 580 (overclocked) GPU

ASUS P8Z68-V PRO/GEN3 MOBO

16gb G. Skill Ripjaws pc3 12800 RAM

RAIDMAX RX-1000AE 1000W PSU

With the speed all the way up on my Space Explorer I was between 15.38 and 15.63 hz with 700 parts.
800 parts about 13 hz, 900 parts 12 hz, and 1000 10.50 hz.  This is all with shaded with edges, 1920x1080, and set to performance with millenium (1 color) for the background.


Does this sound right?  I have yet to get to do any real tests besides this because I just built the machine and it's not set up on our server yet. 

 


Message 89 of 103
karthur1
in reply to: mjmaes

Yes, that sounds about right.  Since around 15 hz is where the rotation gets choppy, we we trying to see what the part count was when we were seeing 15hz.  The values that you have reported are very similar to what smjanows shows on his charts right above your post.

 

How in the crap are you getting 4.5 ghz out of the i7?  I never could get mine stable at that speed. What do you have for a hard drive in this system?

Message 90 of 103
mjmaes
in reply to: smjanows

To answer the question about overclock, I actually had it up to 4.8 but bumped it down a bit for safety.  I run 7 fans inside my enclosure as well as a large heatsink.  This was achieved with Asus's auto overclock feature in their bios.  So simple to use.  As for hard drive I run a Cosair CSSD-F120GB3-BK Force Series 3 Solid State Drive - 120GB, 2.5", SATA III, 6Gbps and a HITACHI Deskstar 7K1000.D HDS721075DLE630 750GB 7200 RPM 32MB Cache SATA 6.0Gb/s.

 

WEI Rating

Processor 7.8

memory (RAM) 7.9

Graphics 7.9

Gaming Graphics 7.8

Primary hard disk 7.9


 



On a side note I just ran the same test on my HP HDX 18 Notebook with Intel Core 2 Duo T9900 @ 3.06ghz and Nvidia GeForce 9600M GT and was only able to achieve 15hz with 250 parts.

 

WEI Rating

Processor 6.6

memory (RAM) 6.6

Graphics 5.3

Gaming Graphics 6.4

Primary hard disk 5.9

Message 91 of 103
prusas
in reply to: mjmaes

http://forums.autodesk.com/t5/Autodesk-Inventor-%D0%A0%D1%83%D1%81%D1%81%D0%BA%D0%B8%D0%B9/Benchmark...
Windows 7 64 bit
Intel® Core™ i7 CPU X 990 @ 3,9; RAM 12Gb x 1600
OCZ-Vertex 4 SSD (119.24GB)
Asus NVIDIA GeForce GTX 580; PLE2607WS 1920x1200
Message 92 of 103
prusas
in reply to: prusas

http://www.tomshardware.com/reviews/best-workstation-graphics-card,3493-7.html 

Windows 7 64 bit
Intel® Core™ i7 CPU X 990 @ 3,9; RAM 12Gb x 1600
OCZ-Vertex 4 SSD (119.24GB)
Asus NVIDIA GeForce GTX 580; PLE2607WS 1920x1200
Message 93 of 103
abreen
in reply to: prusas

Inventor 2013 doesn't seem to have the same registry tree as before, is there something similar to the old PostStatistics flag from 2012?

Message 94 of 103
karthur1
in reply to: abreen

It is the same except for the Version number.

 

For 2013 the path in the registry is this:

 

HKEY_CURRENT_USER\Software\Autodesk\Inventor\RegistryVersion17.0\System\Preferences\Scene Manager\Debug\Post Stattistics

 

 

Message 95 of 103
abreen
in reply to: prusas

Well, it wasn't already there but by adding ...>Scene Manager>Debug>Post Statistics=1 I've got everything working!

Message 96 of 103

Hi ComputerGourou666,

 

You have OC configurations. Could you tell us difference between normal speed 3.2Ghz and OC speed 4.7 Ghz during work on large assembly? Can you make test on 1000.iam file?

 

Thanks you,

Message 97 of 103
Jzeemeijervabh
in reply to: prusas

Has anyone tried the inventor 2014 software? I've come upon something strange. with the benchmark:

 

Computer:

intel i5 i760 quadcore (2,8ghz, turbo 3,3)

nvidia GTX460

got to 700 parts with 15hz

 

Dell Laptop XPS 15

intel i7 4702 quadcore (2,2ghz, turb 3,2ghz)

gtx 750M (much slower than the gtx 460)


got to a whopping 1700 parts with 15hz.

 

Another computer with i5 3470 (faster than both other cpus, max turbo 3,6ghz) and gpu about the same as the 750M (7750 ati) is noticably choppier than the laptop. Another difference might be that the laptop is running windows 8? or that the new generation cpu\s and gpu's is just better in architecture.

Message 98 of 103
karthur1
in reply to: Jzeemeijervabh

I tested my system using 2014 and 2015.

 

2014 is about the same as 2013,  I have to be around 700 parts to get 15Hz.

 

For 2015, I had to increase the part count to 4,500 to get 15 Hz.  I tested 2015 in express and full-load mode and it was pretty much the same.

 

2014-04-14_0721.png

 

 

I really couldn't believe what I was seeing for 2015, so I ran the Inventor6_Bench.ivb (with the Engine MKII.iam) on both 2014 and 2015.  The results were about the same.  Not sure why 2015 scored so high on the 1000.iam refresh rate test.  Here is what I see with the Inventor6_Bench.ivb on the same system.

 

Inventor_2014_and_2015_benchmark.png

 

Kirk

Message 99 of 103
jbnhf23
in reply to: prusas

can someone tell me how to perform this test?

 

I have enabled "Post Statistics", and have downloaded the files 1000.iam and kubas.ipt, but how can I test my system using these files.

 

I have been looking for a macro of some kind, but cant find anyone.

 

pls help  

Spoiler
 

Regards,  

Jesper

Message 100 of 103
karthur1
in reply to: jbnhf23


@jbnhf23 wrote:

can someone tell me how to perform this test?

 

I have enabled "Post Statistics", and have downloaded the files 1000.iam and kubas.ipt, but how can I test my system using these files.

 

I have been looking for a macro of some kind, but cant find anyone.

 

pls help  

Spoiler
 

Regards,  

Jesper


To do this test, open the 1000.iam.  Edit the rectangular array so that there are say 100 items.  Now grap it and spin it around.  As you are spinning it, notice the Post Stat number at the bottom.  You will see a number come up pretty often.  Write it down.  Now change the array so you have 200 parts and spin it, and write the number down.  Keep doing this until the refresh rate gets down to about 15Hz.

 

That is basically how we got the results for the graphs posted above. (Yes, it takes several tests).

 

There is another test that has been around since Inventor 6.  See thread #28 for a macro.  It pretty much runs itself, but you have to have the MKII engine sample files for this test just so your results will be consistent with other's results.

 

Kirk

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

Post to forums  

Autodesk Design & Make Report