How fast is your Inventor PC really?

[Raider_71]

Hi guys,

We have had to do some testing on a bunch of Inventor PC's recently to determine which of the PC's needs to be replaced. Obviously we needed to find out which of the PC's are the worst eprformers as there was only budget to replace 50% of the design PC's. So we thought the Darwin theory will come in handy right... 🙂

Anyways I started searching on the net for toppics on how to benchmark an Inventor PC. Then I thought whats the point of using gaming benchmark tools because Inventor is not a game and there are more aspects than just graphics performance when it comes to percieved performance on an Inventor PC right.

So we decided to create our own Inventor benchmark tool which tests various aspects of an Inventor PC to give us an overview of our PC's performance. This then helped us make a decision as to whcih pc's to replace.

We have made the tool available free of charge to anyone interested in checking how their PC stacks up to their peers or friends. 🙂

Please download it here and post your results here as well if you want. Would be interesting to see what beast workstations are out there.

I would like to say thanks to Kirk #karthur1, for helping in testing the app.

Please feel free to send any suggestions our way. There is an email link in the app.

Download and Install

The application will work with Inventor 2014 to 2016 only.

IMPORTANT: After installation there will be an Inventor Bench icon on your desktop that looks like this: 

My resluts:

HP Elitebook 8560w with an SSD upgrade.

[leowarren34]

@JoschaWondraschek 2 or 4 DIMMs wouldn't make any difference other than the max capacity for the number of slots you have since the 9900kf only supports Dual Channel Memory. As for timings I am doubtful that it makes any difference but it's one of those things where you can get as fast as you can without a massive price hike - I'll be interested to see any data.

[JoschaWondraschek]

I've just read through various sites and discussions and it's mostly recommendet to use a higher RAM clock speed as it seems that they have higher transfer rate (even though the 9900-KF only uses 2666MHz), but I'm not sure if i understood everything correctly, as english isn't my native language and there are a lot of technical terms.

Maybe the only way to find out would be to test with trial&error.

Interesting that @bwatson1967 got such a good result with his config, even though we have nearly the same configuration only that he has 2400MHz RAM speed. (maybe intels XMP plays a role here)

[leowarren34]

Hi @JoschaWondraschek Faster RAM does make a difference but it can be subtle. In general, it's best to buy the fastest you can without spending too much as often faster RAM comes at little cost - often just a few a bucks difference between 2666 and 3000.  There are high-performance kits that do absurd speeds but the price to performance really falls off a cliff so is really unjustified in this application. Just checking PC Part Picker and I can see a set by the same manufacturer with 3200MHz that is $3 more than 2666MHz which isn't too bad IMO. Yet if you want let's say 4000MHz RAM then it's a further $60 for the same capacity DIMM which isn't worth it.

The final thing to say amongst this waffle is that you can run memory faster than the maximum specified by your processor however you will be running the processor 'outside of spec' which can cause issues if you need customer service. It's incredibly sketchy especially when Intel makes the XMP Spec see Neil's above post on it. Also, check out this video explaining the XMP issue: https://www.youtube.com/watch?v=ZA47U8hAD_E

[JoschaWondraschek]

Just for my understanding. If I use RAM with faster clock speed, I have to do some OC on the CPU to use RAMs full potential, cause otherwise the CPU only would use its standard clock speed. (in my case 2666MHz)

I'm really keen on knowing if different RAM timings (with same clock speeds) would speed up Inventor, as I imagine, Inventor does a lot of read/write work on the RAM.

[leowarren34]

Hi @JoschaWondraschek,

To answer, no you do not have to OC your CPU to run your memory above the CPU Spec - they're independent and controllable in the UEFI/BIOS.

[smokes2998]

Really intresting experiment that I did

I ran Catia v5 and  Inventor 2021 on the same machine using an a-surface consisting of 6000 faces. Catia was 70% faster at loading and editing the surface than Inventor, I also notice Catia ultised the GPU more for  doing the 3d calculations than Inventor did therefore I think it is not just a hardware issue but a sorfware issue.

64gb ram

4,2 ghz core I7 chip

quadro rtx 4000 8gb ram

SSD

windows 10

[Neil_Cross]

---

@smokes2998 wrote:

Catia was 70% faster at loading and editing the surface than Inventor

---

Catia is roughly 450% more expensive than Inventor so I'd consider that a huge win for Inventor worthy of much fanfare.

[Raider_71]

Hi @smokes2998 

Yeah, I would never compare Inventor to Catia when there is surface data involved. Catia was built from the ground up (back in the day) to be a surface modeller. Also, NOT as a parametric surface modelling tool. Solid modelling functions were added to Catia in the 90's. 

Inventor was developed as a parametric solid modelling application. Parametric "construction" surface functions were added to Inventor because of user demand but to this day I would never pitch it against the likes of Rhino or Catia if the user has a heavy surface modelling requirement.

So, if you just want to compare Inventor to Catia then compare it with generic modelling and drawing creation workflows in mind to see which one performs best on a given pc. It's going to be pointless isolating one type of workflow because your results may be skewed completely in one direction. There is huge overlap in the two applications but each one has its place out there in the industry that's for sure.

Also, you would be surprised as to how well Inventor performs against Catia in certain fields and from what I have seen it can outperform Catia by quite a bit sometimes.

[Anonymous]

Hi,

i don't use Inventor, just curious on what my PC would do (after watching video from TFI). Installed Inventor 2021 trial and ran benchmark, so why not post it?

Ryzen 5 3600@default, 16g RAM 3200@3600 cl16 (tuned), kingston A2000 500g nvme

[leowarren34]

Hi @Anonymous,

That's a pretty decent score, anything below 10 is considered on the low end, 13 is pretty good, for some perspective the highest I've seen is in the 18 range. The number just gets bigger as the score improves which is cool as you can track the highest score over time, I remember when around 15-16 was highest and that wasn't too far in the past.

[smokes2998]

It is about 6K for basic workbench license.  I find Catia way more productive as a tool with the basic licenses as it faster doesn't crash and it much easier to achieve the same task in catia than inventor as the workflows are designed for engineers and not cad jockeys.

[brunnic]

Anyone tried the new HP Zbook Firefly?

https://store.hp.com/SwedenStore/Merch/Product.aspx?id=1J3P8EA&opt=AK8&sel=NTB

New intel i7-10510U 4 cores with turbo up to 4.9gzh? Still low-end graphic but good enough? No VR plans for future..

[leowarren34]

@brunnic, I've not run the system but a quick spec review,

CPU: Speed is key and 4.90GHz is nothing to sneeze at especially when the top of the generation is only 0.3GHz faster. 4 cores will be alright - perhaps watch the multitasking when you're running Inventor and being more demanding.

RAM: 16-32GB depending on the spec - 16 will be plenty for smaller work and 32 GB will be fine for almost any project.

GPU - Inventor isn't GPU accelerated but does require the VRAM - 4GB is enough for light to medium work but depending on what you are making could be a bottleneck.

SSD: 512GB somewhat small but will leave around 400 GB after windows, Inventor and Over provisioning - so if you're cloud-based as much as possible you should be fine but depends on what other programs you may install.

That's a spec rundown on paper - more specific features such as monitor, keyboard, and other features will require more research/reviews.

[Anonymous]

Posted with the wrong account...

Thanks for input.

We have the HP zbook 15v i7 9750H with 32gig ram as standard machine but it's out of stock so..

Will give it a try and post the resolut.. 

[Anonymous]

Just ran this on one of our Stryker II workstations.

Delid'd 10900K All core clock speed 5.2Ghz (No C-State)

Direct Die Cooling - 360mm Cooler

DDR4 3600 CL16 RAM

3.8TB NVMe Drive

2080Ti

Does this benchmark account for monitor resolution, number of monitors, and cores? For benchmarks to be true test controls have to be in place, but many of our users multitask. It would not make sense for an 8 core that ran the test and scored a 15 and a 6 core that ran the test and stored a 15 to be considered a tie. One has more resources. Or does this benchmark account for that? 

[leowarren34]

@Anonymous If you look at the Inventor Information section of the results it does display the resolution.

Do bear in mind that this is a benchmark which for a maximum score should be run with the minimum apps open. However, I do see the merit of running the benchmark in a real-world scenario as it can indicate issues down the line especially with storage.

In regards to an 8 core vs 6 core being considered a tie that would make sense as it's only measuring what Inventor needs and how it performs, with Inventor being mostly single-threaded having a surplus of cores doesn't impact performance and so shouldn't affect scores despite the disparity. If you want to test whether you 6 or 8 cores will benefit you set up a typical workstation with apps like email, chrome, teams, etc in the background then run the benchmark until you get somewhat consistent results. A benchmark is only there to measure performance quantitatively so you can compare against other systems.

I would also look at TFI's project in the works: https://www.youtube.com/watch?v=QZzokpE-jx0

[Neil_Cross]

Thanks Leo.

@Anonymous  it's impossible to summarise several years worth of benchmarking evaluation using this InventorBench test, but one of the reasons why I've been working flat out for several months on creating a new suite of tests is because this test is too CPU frequency bound.  But in defense of the guy who made the test, he didn't anticipate it being used in this way when he conceived it.

Apologies for the poor quality picture, but to put this into perspective, to prove a point, 5 years ago I built a "low budget workstation slayer" PC with the total parts list being $350, that was everything including case, PSU, GPU, everything.  It had 2 cores, but they ran fast, and this $350 build was scoring higher on this test than almost every HP, Dell & Lenovo workstation on the market.

Obviously you wouldn't buy a 2 core system to run Inventor, but I built this to prove a point, at the time the highest scoring builds containing the 6700K where maxing out at around 11-12, you could spend $5-10k on a Tier 1 workstation and see it score a 6-7 at the time, and here's this $350 2 core DIY budget PC pulling a 9.  Point being, this test has done a great job at identifying performance differentiation at a very fundamental level, but it wasn't built to go beyond that.

The test that I'm working on (and I'm sure the V2 of InventorBench will also) has a wide variety of routines which utilise the areas of Inventor that do make use of higher core count CPUs, along with using bigger datasets to bring RAM into the picture, among many other things.  

But without making this a massive essay, the only total true multi-core optimised routine is ray tracing, all other modules in Inventor that are labelled as "multi-core" are subject to Amdahl's Law, therefore it isn't a 100% all cores working at once throughout the entirety of the operation kind of a function.  What that translates into is you could chuck a 64 core threadripper at a multi-threaded task in Inventor (excluding ray tracing), and it won't finish that task 6x as fast as it would have done on a 10 core system.

That was a long winded way of saying you could have disabled 6 of the cores on your 10900K and InventorBench would finish in roughly the same ball park!

[Anonymous]

Thanks for the share. @Anonymous - Just released it yesterday?!! Congrats
brother!!

[Anonymous]

Should have watched the video first. Did not know it was out there. Can't wait to test this out!!  Cheers. 

[Neil_Cross]

😎

It's been a much bigger project than I first anticipated but will be worth it.  I'm currently sat here tweaking the formulas and algorithm for the generated results.  Trying to decide what the final score number should be, what's the baseline, how to give good separation between the tiered modular scores etc.  

I forgot to mention this in the video, but as well as an "InvMark" score, my test will also generate separate secondary level "Modelling, Assembly & Drawing" scores along with tertiary scores per individual test i.e. FEA test, Dynamic Sim test.