Back to Back Comparisons Don’t Matter
This was a comment from a reader in a recent post:
One of the people who originated SW said in an interview that they’d settled on parasolid rather than ACIS because there were certain blend examples that parasolid would do that ACIS couldn’t. That was a lot of years back, but it implies that base geometry creation was pretty settled, even back then, and that parasolid and ACIS weren’t too far apart for that purpose. IronCAD uses both kernels in parallel, so the differences can’t be too considerable. A glance at user forums shows that some users complain that they can’t achieve this or that result, whereas you, as an experienced and capable user, know that the way they tried to create it was doomed to failure. What that demonstrates is that the way that CAD companies will get the greater number of new users isn’t by adding arcane capabilies to please high-end users, but by exposing the capabilities that they already have to easier use by inexperienced and uneducated (in the application-specific sense) users. A number of users who trip over a glitch frequently will tell you ‘they can’t even get the basics right’, but those users are still using the program. Ultimate capability is mostly less important than ease-of-use. What I’m saying is that back-to-back comparisons are almost meaningless. It comes down to whether you’re a Ford person or a GM person, despite protestations of other standards.
I’m not singling this out to ridicule it. There is stuff I agree with and stuff I don’t agree with in here. I welcome differing points of view because it helps me figure out what’s going on. I really want to discuss a couple of the ideas here.
One thing I wish is that I knew who you are and where you are coming from. Your comments might make more sense if we understood the context in which they were made. Are you a user? What kind of stuff do you design? What software do you currently use? Work for a company or yourself?
ACIS and Parasolid
If the differences weren’t that great, then IronCAD wouldn’t need to use both. There was a general rush away from ACIS, including both SW and Solid Edge. I think the differences might not have been huge, but they were big enough to make a difference to users and to some extent developers. As I remember it it wasn’t just geometry reasons why companies switched, there were some reliability and possibly speed reasons too. Dan Staples has talked about this a fair amount in comments to this blog and other places.
CAD vs User
I definitely agree that many users could get a huge boost in CAD power if they only learned what was in their existing software and how to apply that power. It’s not all the CAD company’s fault. Some users simply aren’t very curious, some are possibly lazy, some would use it if they only knew. I totally agree with this, as well as the idea that basic features and public education are both simpler to add than advanced features. I would argue, though, that I’d be happy to get 1 advanced feature for every 10 basic ones. I’m sure they have their own way of determining a ratio.
At the same time, you might run out of basic features to add at some point. You will never run out of advanced stuff. If you think CAD has been developed as much as it can be, you’re just not paying attention.
A number of users who trip over a glitch frequently will tell you ‘they can’t even get the basics right’, but those users are still using the program.
I guess it all boils down to if you want users to have a positive “experience” using your product, you need to make sure that it can handle the job. If you don’t care about keeping those customers, then it doesn’t matter what they think. My point here is that if you’re going to do something, do it right. Users have to use a highly imperfect tool to get perfect results. We can’t get away with the excuse that “aw, my CAD software sucks, so I can only do about an 80% job for you”. And we can’t run out and buy a new CAD package for every job.
Ultimate capability is mostly less important than ease-of-use.
We’re not talking about “ultimate” power. I would define “ultimate” power as being the ability to do anything I can dream up. We’re just asking for “adequate” power, which I would define as the ability to do what I need to do. I would argue that ease of use doesn’t matter at all if you can’t complete the task you need to complete. SketchUp is very easy. But there is a lot of stuff it just can’t do.
Back-to-back comparisons are almost meaningless
You’re assuming that all CAD tasks are essentially equal, and the only difference between the relative difficulty of any two jobs is the relative abilities of the users. I don’t buy this one at all. Despite all the similarities, there is a significant difference between working in multibody techniques in Solid Edge and SolidWorks. Spaceclaim and Pro/E both do sheet metal, but the methods are worlds apart. NX and Solid Edge are both built on the same kernel, but Siemens goes to the expense of maintaining two separate development teams. This is as strong a proof as any that the differences must have some value.
Even if a comparison did just come down to tomahto or tomayto, GM or Ford, you can still make relevant comparisons. GM has a heads up display that Ford doesn’t. Ford doesn’t have anything to compare with a Corvette (since the GT is out of production). These aren’t cosmetic differences, but even if they were, they are valid decision factors.
Are you saying that you should just go pick up CAD software off the shelf because they are all interchangeable, and it doesn’t really matter which one you use? Don’t waste your time on an evaluation?
Regarding the rush away from ACIS back around ’96, as far as I have understood it, is that Spatial was in a pretty bad boat both with their policies and their core technologies. I don’t like putting credit on just one person but I’ll say that around 2000 they got a big boost by Mike Payne (former PTC/SW) joining them for a few years and helped put them back on the rails.
So I suppose, if one were to say, that TODAY parasolid and acis are probably fairly equivalent in power. But not back in ’95-’98 when SolidWorks and Solid Edge got their beginnings.
Sitting on the side lines watching the reaction to this post is interesting. As a person who on many occasions, in the early days, represented Autodesk with AutoCAD at CAD shoot-outs that involved a plethora of CAD software with much more capability than we had and yet we could always “put on a good show”. We could also defend ourselves because we were seen as not being a solution by many; little did they know?
But more to the point I believe the greatest problem with CAD comparisons – shoot-outs included – is that failure to be able to do “something” a particular way or as good as another was a sign of weakness. I would always take a “shoot-out” task, I knew I could not finish or do correctly, as far as I could. It was no embarrassment to me as it was to other presenters because I believe it was better, for those watching, to see just what AutoCAD could do. It also gave those, who were patient buyers, a way to gauge AutoCAD development. So while there were those who criticized and poked fun there were also those who were watching and making their decisions based on what they saw may happen. So CAD comparisons can be very beneficial if viewed and backed with sensible thought instead of thinking it is some form of race to a finish line.
Now you may choose to dive for cover. CAD development has always been led by the developers. It matters not what we think we contributed, (as the earliest users we did), nothing we wanted got done if it did not suit the developers. As time moved on developers (shareholder) returns took a more controlling interest and that is were we now sit; firmly locked to companies whose major interest is not entirely about us. We feel miffed, right?
The boat industry missed, I believe is this; we failed to ensure those we bought from understood how important it was for them to keep paying attention to our needs. I would suggest a way we could – maybe should have done is define just what CAD software should do; as a minimum for instance. Should we not have “created a standard” a list of features a CAD system MUST have to be acceptable?
Complex for sure but engineering is about applying standards, nut, bolts, pipes, threads, material specifications etc. so why not CAD tools. If it was defined a CAD tool should extrude and “object” then any and all CAD systems could be lined up and compared against a standards list of functionality. They way they achieved that task and how well can then also more easily compared against what is defined and others.
We have “consumer” organizations all over the world comparing all sorts of “stuff’ from toasters to cars. Most, if not all, created by people like us using CAD software and yet do we struggle to find a way to compare the tools we use? I think that’s pretty funny really that after all these years we have failed to build a “tool” CAD developers knew they had to address; instead we left it to them to tell us “what we wanted” and it ain’t right is it?
Thanks for making your blog available for us all Matt, seasons greeting to you and all your readers. I hope 2012 is a great one for everybody.
I think interopability is great place to start, it can help flesh out what the readership is interested in and what ancillary products are commonly used.
@Mark
Maybe we’ll come up with a comprehensive list of things to compare, or maybe it should be done on a case-by-case basis.
The one collaboration (can we use “interoperability” instead?) thing that I have trouble with more and more is STL files. The tools I have now to work with them are incomplete, and only work up to a certain point. I’m ok with other file formats.
I’m in a somewhat unique position. Many of my customers don’t have CAD at all. That’s what they hire me for. Many of the rest hire me specifically for SW, but I still have version conflicts with them sometimes.
I don’t disagree with that. Some purchases in CAD software are lost to feature to feature comparisons, but this does become a minor amount in large installations. It can become a slippery slope to decisions sometimes, especially when the majority of parts are revolves and extrusions and not complex geometry.
Companies will choose what works for them based on geometry, workflow, and collaboration. Indiviuals will choose what feels comforatable, and easy for them to use.
When you look at the independant user not only should it be about the tools, but also about collaboration as well. Especially working in context of contracts.
@Mark
You know the view from the high level must be grand. If you are isolated in the world of giant CAD installations, I can see where you might be tempted to believe that functional details don’t matter. Down here on the ground, we are the ones that make the whole process work (developing CAD data), and I can tell you that when you get out of the realm of extrudes and revolves, there are significant differences between tools that affect productivity, and for independents, quality of life.
I also say that Head to Head comparisons don’t really address anything other than what is in a software package. I once had a guy count my clicks in a demo and I stopped and pointed out he wasn’t buying me with the software and he should address the concepts I was laying out and not the exact method I got there. As soon as you get in a feature war you lose sight of why you want to upgrade or switch your CAD software.
The reason you should buy software for design is to address a workflow or a process issue in your company. Does this software address what I need to get a design from art to part and anything beyond that. Can I export this to the format my suppliers need? Can I export this to a format my internal secondary users need? How easy and accurate is that translation? What metadata can be carried with the export to reduce rework and errors? Some companies do this type of work better between their products than others.
To Kevin’s interest, I have seen a video series that takes EXPERT users from Inventor, SolidWorks, Catia, and Pro/E and has them accomplish the same design task for 15 different elements of common design. Each video was shown in entirety so all users could comment on if it was the right techique or not in the respective software as well. I am afraid though that most readers on this forum would not like the results of it.
@murray
Murray, thanks for filling in some details. If you’re doing simple parts, and have used a few different programs, I can see where they would all start to look the same.
@Devon Sowell
I agree that there is a largely unaddressed problem with file management in CAD. In my ideal (and non existent) world, when operating systems were rapidly developing from text interface to windows interfaces, file revision management was implemented in the OS. Unfortunately it never happened. Now CAD companies (among others) are trying to solve the problem by selling overly complicated PDM software or using a database structure for files.
I realise that this perspective adds nothing in terms of practical use, but I can’t help feel that until file revision management is handled in the right part of the organism, the animal is going to be an ugly beast. Imaging if your anus was in your head! Actually, probably best not to imagine it.
I’m retired now, but I was a CADD professional from drawing boards into 2D CAD through through to modeling. I haven’t had experience with large assemblies, and now I use “retail” CAD for very occasional contracts. The access that I’ve had to Pro/E, SW, Vellum and other suites has been through employment, so I don’t have current experience, and I can’t claim power user status of any of them. I think that a public, “competitive” evaluation will be inconclusive or misleading. It might be interesting if you, or anyone else, or everyone else, put up difficult tasks, sight unseen, for expert operators of the various systems to bash out as quickly as they can, back-to-back, but is that likely to happen? Political candidates go all conditional when debating season comes around, because they each look like ‘the man’ under specific debate formats, and like the whipping boy in others. My opinion comes from a minor epiphany some years back. The job I was in meant that the way I worked didn’t change much, although each new release had new features and tools that I could benefit from. I realised that I’d stopped trying to use newer versions of one particular program when the UI had been changed. The new UI was unfamiliar and I didn’t feel as competent with it, even doing the same things I’d done previously. I determined to try to work with the new UI and found that not only was I making use of newer tools, but that I was able to understand and work with other programs much more effectively. My other point is that the price of CAD is much less, proportionally, compared to previously, and that it’s much more likely that something that will do some things the way we want to do them is already on the market, and at a price that doesn’t terrify. So what, apart from the money, is stopping us from, as you put it, running out to buy another program? Format exchange. Lack of a neutral format for parametrics is what makes us want our weapon-of-choice to do everything under one roof, and it’s why DS/SW or Siemens or anyone else wants you to think that their offering is most capable, while turning us all into reactionary conservatives so far as CADD’s concerned. I think it’s likely that something like a STEP format with XML tagging is most likely where that sort of advance is going to come from, or where it ought to, but the .dwg and JT wars give you an idea of how much resistance that will face.
Head to head feature comparisons do have a place provided you set down the parameters and use the same models. Back in the 90s this kind of stuff was quite common – with 3D CAD relatively young and developing fast, there were very clear differences between applications and kernels. Ashlar used to top every fillet test around, for example, despite use ACIS based software. Back then I used SolidWorks and a number of ACIS based systems and there were huge differences. ACIS was (and to some extent still is) more flexible with surfaces – allowing open solids to be created for example. Parasolid was more robust and reliable, and tended to be better for shelling.
Add to that other kernels like the Think3 one, VX, Granite One (Pro/E), CATIA etc. and each has different ways of doing things. VX for example gives more control over fillet “upside down smile” cases than parasolid.
These days though, I think developers actually use different bits of code from different kernels to achieve specific aims. For example, the killer feature Fill Surface in SolidWorks comes from CATIA. Developers often license specific bits for translation functions – Spatial libraries for example, Granite One etc. Again, many of the CAD minnows put the big boys to shame in this respect offering native level translation in even the lowest cost products (for example Ashlar licensed the CATIA v4 in/out and Granite One (Pro/E) in out and Parasolid back in the early 2000s. At that time to get the same options in other apps you would pays thousands.
The reality is that these days, if you are primarily working on prismatic type components and assemblies ANY cad system will do a decent job. Kernels are all quite robust now and can all handle this type of stuff with ease. So what you have left is the interface and the performance, and specific geometry creation needs.
Interface – to me – has always been, and always will be the key decision maker in whether to buy software or not. If the interface is difficult (and that can apply to system wide or specific features or tools) it will never be used effectively. I can usually tell withinan hour or two if I am going to “get” software or not. I do get SolidWorks, I don’t get Inventor, for example. Whilst SolidWorks started with a good core interface they have developed it to be efficient and obvious and dumped as many dialogue boxes as possible. Inventor still uses lots of dialogue boxes.
But interface to the toolset is also critical. Great functionality can be let down by poor interface. Ashlar has a fantastic core interface – possibly the best for 3D surface design for setting up curves. But they have some clangers in tools. Their bend tool for example was powerful but it was almost impossible to use due to the way it was interfaced. What sets the big boys apart from the minnows is feature previews. Most tools in SolidWorks give an interactive preview. Most lesser known CAD systems do not use previews because that kind of functionality is difficult and costly to programme.
Performance is obvious isn’t it? Well no. It very much depends what you are doing. Graphic performance again is an area where the big boys tend to excel and the minnows struggle (think SolidWorks vs Alibre for example at large assembly performance). Feature rebuild times, preview display time, functions (shelling, filleting etc) are all dictated by kernel calls and how well the app has been developed.Again the big players focus a lot of time and effort in these areas becuase that is what most users SEE. I use different apps daily and there are still huge differences between apps – even using the same kernel. It all comes down to how well the app has been developed.
Finally, specific geoemtry creation needs. To me, again, this is critical. This is what sets applications apart. I’ll repeat my mantra – if you cannot model it, you cannot edit it, you cannot simulate it, you cannot render it, you cannot prototype it, you cannot machine it (all from CAD data). It really is that simple.
As I said, if I am doing simple parts ANY system will do the job (so I often wonder why many jump straight to SolidWorks or SolidEdge when Alibre would do just as nicely at a fraction of the cost – but there you go). What separates the men from the boys is the difficult stuff – NOT wild surfaces (as favoured by most demos), but the subtle, dimensionally controlled technical surfacing – like controlling a wall thickness as the surface transitions around a curve, or like doing a subtle indentation into a double curvature surface, or ensuring a reflection line is smooth, or creating specific draft features or specific split lines to freeform surfaced parts….no….geometry based CAD still has a hell of a long way to go before it is “done”.
And this is the issue I have with all CAD vendors – no matter who. The only company I have seen recently doing anything to the core modeller is PTC – with Freestyle in Creo Parametric. I have no experience with Creo so can’t say much about it apart from the fact that everyone I know who uses it swears by it.
The only other geometry enhancements to tools are in very high end packages that quite frankly, only very large companies can justify. Some more specialist tools are available but most of these are costly as well – Delcam’s Powershape, SensAble’s haptic modellers for example offer toolsets that can do far more than most mainstream tools.
So for me at least, a CAD olympics is not really a thing that would interest me aside from academic curiosity. Feature by feature comparisons are interesting but don’t really tell you much. No one person can know enough to use all test systems properly (though you could argue that this in itself is a good test, but in reality of course we always migrate towards the interfaces and processes we prefer, so getting unbiased reviews is next to impossible).
No, the one thing I think that would interest me is to see how expert users in different software would model up specific products. If you have, say, a consumer electronics product, and you start froma series of hand sketches and an electronics package, how would a user DESIGN the product using a particular system. Even for a simple product, showing a COMPLETE process (design. modelling, tweaking, detailing, drawings, export options) like this would take perhaps at least 5 hrs of video. Multiply that by x number of cad systems under review, and an indexing system to identify specific stages and processes in each approach….adds up to a lot of effort and delivery bandwidth.
Yet, this, I think, is what is needed. Others have tried this. One that springs to mind is Design Engine in Chigaco. I’ve often lamented the fact that they are based in the USA and don’t offer online training, as those guys seem to know exactly the issues I identify. Their website seems to be down at the moment though, but there is a Core77 forum linkthat relates to a SoldiWorks vs Pro/E shootout.
http://boards.core77.com/viewtopic.php?f=4&t=21864
I think that reliable geometry is most important. Sweeps should sweep every time with every kind of path curve. No wiggles or jiggles that are not on the path curve. If the result self intersects, give the sum of the contours. We learn to work around features that fail in particular circumstances. We protect ourselves from Solidworks corruption, crashes, and hangs by saving files in distant directories with conspicuous names.
File bloat. The 1200X oversize full of crap Solidworks file is junk. This makes managing the data more difficult. An inscutable CAD database will really mess me up.
Of course very advanced users need the broccoli shaped lofts, and loopy sweeps that have no resemblence to the profiles. I prefer solid geometry control.
Here’s are a few observations;
1. My clients are small to medium size R&D and MFG companies making prismatic assemblies and parts.
2. CAD software has become the defacto design tool and is used and taught in most colleges and other schools.
3. In my experience, most users just know enough about their CAD software to make the specific types of assemblies and parts they work on. Usually they don’t want to learn more than is necessary, unlike the people that read this blog and provide intelligent comments.
4. Let’s define the “design process” as geometry creation and design history including document control of the CAD files and Drawings.
5. After years of use most companies have thousands of CAD files and Drawings that are not usable. For example, multi sheet(more than 10 sheets) CAD Drawings and complex 3D CAD files.
6. The “design process” has now become unmanageable. Complex 3D files are very difficult to edit and no one can find the latest or correct Drawing. Duplicate files are all over the place, located on local drives and the network.
7. So CAD companies have chosen to focus on the control and management of the CAD files rather than the geometry creation tools.
8. Why? In my opinion so they can sell new products and because the existing CAD software is bloated and out dated. For example, the SolidWorks application code is over 17 years old.
9. So now expert CAD users are stuck. We can’t create the geometry we want and we can’t manage all the thousands of CAD files we’ve created.
10. The entire process of CAD, Computer Assisted Design, needs a complete overhaul.
Devon Sowell