Survey: Inventor Performance

[Neil_Cross]

Q.  Would you fully support the Autodesk Inventor product team in temporarily stopping work on any new features for Inventor, and instead, dedicate a couple of years and all their resource to working ONLY on making Inventor perform much faster than it does now?

Specifically referring to converting more of the application to being multi-threaded, so it can use more than 1 of the many increasing CPU cores that modern architectures have now.

Please note this is a personal survey and in no way has this been actually discussed with Autodesk, there is nothing to read into here!

[Alberto_Pérez_Parra]

Surprisingly Intel Optane is making a great improvement in the performance, even in i3 laptops!

[Frederick_Law]

Replace the HDD with SSD will have better result.

Intel Optane is fast SSD cache.

1TB SSD is around $200.

32GB Optane is around $70.

[jletcher]

They have Inventor crashing at the fastest rate I have seen do you really think they can get it to crash faster?

http://forums.autodesk.com/t5/inventor-ideastation/don-t-add-anything/idi-p/3685008

[jletcher]

P.S.

 Don't matter how fast you make the CPU if the user interface is not user friendly the CPU spend is worthless.

 And after seeing the hole feature UI change and what is in the Beta you will not have to worry about CPU speed you will be so slow molasses will pass you.

[Neil_Cross]

I'm regretting making this thread.  I should have expected this.

[DRoam]

My answer would be no.

I would see a much bigger productivity gain from streamlining Inventor's workflows (less clicks to do things) than from increasing Inventor's performance (less downtime between clicks).

I don't work with big or complex datasets, so I don't find myself sitting and waiting for Inventor very often. Rather, I find myself frustrated by how clunky it is for Inventor to do certain things (such as managing view reps, or selecting components within a pattern or another assembly, or moving between editing and creating constraints), or its lack of ability to do something I need (such as keeping components mirrored, or actually adapting Adaptive parts).

And of course, I suffer from Inventor's bugs and "quirks". In my opinion, fixing these should be prioritized at the very top, above even new features or performance enhancements. If the tools we have don't even function as expected, there's no sense in piling on more tools or making them faster at breaking.

But I can certainly appreciate that some people do work with big and complex datasets, and for them increasing performance is just as critical to being able to get work done as fixing bugs is.

[johnsonshiue]

Hi Derek,

Like I mentioned earlier, improving performance is an endless pursue. The performance enhancements we have been working on, though targeting large assemblies and drawings, are broad base. Even if you don't work on such large scale, you get the same benefit, albeit not as dramatic.

I am glad that Neil started this thread. He is not the only expert having the same question. This thread allows us to discuss issues in public at a transparent level not possible a while back. There is always room for improvement in every corner of the product, just like any product on the market. We are making progress (I know some people would disagree; I only hope they can do so respectfully). I hope you see the change.

Many thanks!

[Neil_Cross]

This topic is a big one for me as it's born out from my own circumstances which I'm sure are not unique, even if I don't manage to reach those people here. 

Once I moved into working in industry, it's been exclusively in companies with hundreds of engineers fighting with big data sets.  And in comparison to something as subjective as the UI, this is an objective measurable consistent limitation/bottleneck that some of us cannot look past and can't avoid, and the downtime caused by this isn't negligible... it massively transcends complaints about hole dialog boxes being unfriendly.  I'm not going to retaliate by dismissing other peoples complaints as being irrelevant, but the downtime I see daily is so significant, a time code was needed for the engineers to allocate the time to it.  

Can some of this be helped by smarter modelling? Yes and no, but the performance limitations are objective and I guess at this point only Johnson really knows exactly what I'm referring to.  But I didn't make this thread to discuss the specifics and explain my issues, I just wanted to see how others felt about it even if its just a small catchment of the community. 

[GeorgK]

Inventor is not developed by engineers in my view but by programmers. The problem is also that Autodesk does not listen enough to the wishes of the customers. Genius 14 (Autocad Meachnical) was developed by a German engineering company and was very successfull. This clearly shows who understands something about his work - engineers who uses it every day. For many functions, I would wish that the developer must use its own function 500 times and that they have to work as a designer and drafter for some weeks before they going to program anything for Inventor. Then many things would work differently than they are doing at the moment.

[johnsonshiue]

Hi Georg,

With all due respect, let's not generalize things. Genius series was a great product. I used it before. I was very impressed with it. Some of the workflows and functionalities trickled down to Inventor and AutoCAD Mechanical. We have a pretty good balance of mechanical engineers and developers on the team. We learn from each other.

Inventor or any CAD program is a tool. It is not perfect. It has room for improvement. We cannot make everybody happy but we try our best ensuring we deliver on value, usability, and stability. If you are not seeing it, we are doing something seriously wrong. It does not really matter who builds it.

Many thanks!

[GeorgK]

I see that they are working on Inventor and there are improvements. But in many places an improvement is only half-heartedly implemented. You can see that very clearly on the drilling tool and other areas. It's little things that need to be improved.

[johnsonshiue]

Hi Georg,

Many thanks for getting back to me! Like I said before and repeatedly, Inventor is nowhere near perfect. There is a lot of room for improvement. It sounds like you have good ideas where those areas need to be. Have you signed up Inventor Beta program (https://bit.ly/InventorBeta)? You can try out the latest internal build on a no-install required environment (single download or browser-based). You can make suggestions to the project teams real time. They are eager to hear your feedback.

Thanks again!

[DRoam]

@Neil_Cross, thanks for sharing a bit of your side. I know you said you're trying to just hear others' feedback, but if you don't mind, I'm really curious what specific areas you see the most performance issues with? Which actions tend to result in the most downtime (sitting and waiting for Inventor to finish something) or lag?

I'm simply curious, since I really haven't experienced much trouble with performance myself.

[Neil_Cross]

@DRoam It's an environment where you've got 200+ Inventor users, there can be teams of 30+ people working on the same GA models with staff turnover to factor in, massive variety of skill sets, and with the best will in the world you can't train every one to create data 100% to best practice... corners are cut every day to meet deadlines and there can't be anyone watching these guys at the time to prevent it.

The problems are usually with the top level GA assembly for a contract, it ends up with dozens of derived parts in it, replicas of those derived parts at different levels, thousands of parts worth of supplier content converted from Pro-E or wherever, and mix in Vault conflicting with Inventor due to released states you get crippling open times of anything between 10-30 minutes... drawing manipulation is sluggish, almost everything you try to do with these big top level assemblies is a write off and really frustrating for the end user.  

My point of view, looking at it from another perspective, is that I see only 15-20% of the PC resource actually being used during this time... I know why, I know it isn't easily fixed, but that doesn't mean it's OK and should be ignored, excused or dismissed.  But I've chatted at length with various developers regarding this and it is what it is, I just made this thread to see if I could find any one else in a similar position.

Don't get me wrong I know Inventor isn't a "slow" program, not by any stretch, I've made 350+ Inventor videos on YouTube using small datasets and it's never slowed me down or been an issue at all... but when you push the limits in a large collaborative environment it becomes a real problem and a lot of people end up wasting thousands of dollars buying upgrades that actually make zero difference to anything.

[Frederick_Law]

I guess its about time we move to/invent a 'distributed file system'.

No more file locking. Complete project/assembly work as one.  User can edit part in the project without locking other out of the project.

[jletcher]

@Neil_CrossWrote:

 "and with the best will in the world you can't train every one to create data 100% to best practice... corners are cut every day to meet deadlines and there can't be anyone watching these guys at the time to prevent it."

Then I say you have the wrong cad manager and people need to be fired for not doing it.

Why do you want to reward those that do things incorrectly?

People don't constrain right so Autodesk puts in options that are a waste to 95% of the users because people don't constrain right and messed up workflows making good users go slower.

But it did not teach them that fully constrain parts is better for performance no it helps them avoid doing the right thing..

You want Autodesk to stop putting features is till they can figure out how to make inventor perform for users lack of discipline in modeling.

When I was manager I had rules in modeling #1 rule don't use projected geometry it is the #2 performance killer in Inventor when Autodesk changed how it worked, it was great till they changed it.

#2 rule No offset constraints it is also a performance killer.

#3 rule everything is fully constrain, floating parts is also a performance killer.

#4 rule no iparts and no iassemblies  these are a major performance killer #1 one in my book and I can prove it very easily.

You figure out why people don't model right and work on that.

I will tell you people cut corners when software is not user friendly and in the past few years that is the #1 complaint I hear about Inventor.

I had modeled a assembly well over 200,000 parts back in Inventor 6 never had issues with performance. I could not make a drawing was the only issue.

Not understanding the issues have people fixing the wrong issues with the wrong fix

"remember last"  is a good example.

[dgorsman]

---

@Frederick_Law wrote:

I guess its about time we move to/invent a 'distributed file system'.

No more file locking. Complete project/assembly work as one.  User can edit part in the project without locking other out of the project.

---

There's a host of downsides to that as well.  A strictly "peer to peer" system doesn't scale well to that many users, and the server-based "professional level" database architecture required to make it work has it's own administrative overhead and problems. 

[leowarren34]

Being honest what the Inventor team and community do is nothing short of outstanding and I feel we have 99% of everything we want right now, what would be better for everyone rather than just the few who use the newest feature would be to optimise current features so it's better for everyone not just a minority

[Neil_Cross]

@jletcher Although I'm not at all interested in discussing modelling practices here, the irony here is that you've just listed off a load of things that you note as "performance killers", but the whole point of this thread and the only thing I'm interested in right now is the root cause of the performance being killed.  The physical act of doing those things you mentioned is not the performance killer, the killer of performance is the fact that Inventor isn't using the processing power of the workstation efficiently.

I'm a firm believer that you should be able to use whatever the software has to offer and the software should handle whatever it lets you do in it, it doesn't though and that's a reality I've accepted, and that's what I set out to discuss addressing here.  

[GeorgK]

@johnsonshiue

I am a participant in the beta program. But it is difficult to make suggestions over and over again, which in my opinion only half-hearted, bad or not at all implemented.