A SolidWorks Performance Story.

We are a medium sized progressive die shop who has been trying to migrate over to SolidWorks from Autocad for about a year. The ride has been slow and frustrating. We design large highly surfaced stampings with a majority of our data coming in IGES. We chose Solidworks over the competitors on a basis of price and what seemed to be very good functionality through our salespersons demos. We were promised a 40% increase in productivity from what our current system was. If you ever take a look at SolidWorks adds they say about the same. My first few designs were run on Solidworks 2004 with all the required service packs installed (seems like their were around 100 of them) with very slow performance. I contacted my AE with my problems: Slow Assembly performance, slow mate times, slow rotation, extremely slow 2D Performance. My system requirements were at the top of SolidWorks recommendations. How could this be? We tried everything. Large Assembly Mode, Lightweight Components, Detached Drawings. If a CAD system has to have all these options it should say something about its performance out of the box. Autocad, the last time I checked, doesn't have a large drawing dumb down button. By the end of all the checking and testing they notified me that it must be the way that I modeled my assembly and it had nothing to do with SW performance. They hadn't heard of any problems similar to mine. I must have been dreaming. SW is better. SW is better. SW is better. That's what their salespeople are trained to force-feed you. I was still feeling pretty good about the product until our SW salesperson told myself and a colleague that he wasn't going to waste any more valuable time diagnosing our problems if were weren't committed to purchasing the remaining 3 seats we were looking to buy. OUCH!!!! Strike #1.
Low and behold the introduction of SW2005. They have a new feature. Can you guess what it is? It's a new slider bar under the PERFORMANCE settings called Level of Detail. That solved about 30% of my problems in the 3D performance arena. My other performance increase came by the way of mate times. When I had my AE do testing with me in SW 2004 I was having mate times upwards of around 30 seconds. It must have been the way I modeled my assembly I was told. Wrong! I loaded the same design into 2005 and tested the same mates. They were instantaneous in 2005. No waiting. I contacted my AE and told him about it. He tested it and had the same results. Hmmmmmm, SolidWorks doesn't state anything in the list of improvements for 2005 regarding increased mate performance. Was I dreaming again? SW is better. SW is better. SW is better. I would challenge any SW user to test mating between 2004 and 2005 on a large complex assembly. Have fun! Strke #2.
I think my third and final strike with the software is coming as I write this. It is something that has plagued me with SolidWorks from the onset and it regards 2D performance. SW2004 and SW2005 have a bad case of the "I don't do 2D" syndrome. Our industry still relies heavily on prints. Our diemakers build from these sheets and rely heavily on notes and section views. We cut anywhere from 6-12 sections through a die assembly. We label each component in the section views and the Autoballoon function works really well to start. Once you load up that many 2D sheets in this software your in for the long haul. Expect 10 minute load times and gathering any information is like pulling teeth. And just wait until you find a screw up. Reload the model and change. Now wait another 10-15 minutes for all your 2D sheets to update. I could have stretched and pulled in 2D in a fraction of the time. The only way this software can function is if the 2D is translated into a 2D system. I think Solidworks knows this and that's why they have products like E drawings and now they have the 2D emulator. Which I might add doesn't work very well with detail unless you're trying to converse a box. If any of you SW jocks out their are still brainwashed to the SW is better thinking take a look at PROE's new document titled: The 10 Ways that SolidWorks Slows You Down. Check it out. It underlines everything I've just stated. Don't get me wrong. I can model just about anything in SolidWorks pretty fast now with the mysterious improvements in SW2005. The package as a whole however still needs work. Up to this point utilizing SW for doing everything I am 125% slower over my 2D package. If you can muster doing 3D in SW and 2D in Autocad I think you'll have a winning package. The search still continues for us on an all around solution however.

Sorry to hear. I had a professor that I trained back in the day, and he loved SolidWorks for progressive die design as well as his students. You can only imagine that their computers were not top notch. They actually did complicated stuff for their final designs. Wish you luck. Have you had your VAR bring in the regional technical manager from SolidWorks? Or try briniging in another VAR to help. They may have a fresh idea to help.
Ken M.

snipped-for-privacy@hotmail.com wrote:
  Quoted Text Here  
students.
I agree the regional guys need contact, and then maybe the crew @ Swks Hdqrtrs.
If I were going to design planes or complicated multi-step blanking dies, I think I would look to the 3D software salesman for each product to take me to 2 vendors who already do such products, to interview their CAD or IT guys (not to steal technology). In other words, I want to see results before I buy the software.
Bo

Brian,
The first thing new users should understand is that they don't know everything. The second thing you need to know is that the application engineer at the local reseller is not always the best person to diagnose problems. Even the SW direct guys are just going to tell you when you have found bugs, and otherwise may try to talk you out of whatever problem you're complaining about. None of these people are really well equipped to help you with modeling best practice, although they can usually criticize things they don't understand pretty well. Some resellers have someone who is highly experienced in both the software and real world design, and that's the kind of person you want to find. Keep in mind that about 20% of people who call themselves experts really are experts.
100 service packs? Surely you mean something closer to 1? All of the Service Packs can be installed in a single shot if the people installing them have any clue what they're doing.
You say you tried "everything", but you didn't mention anything about working locally instead of on the network, understanding in-context relations, anti-virus software, wise use of subassemblies for mates or using multiple drawings instead of multi-page drawings.
I'm curious how your highly productive 2D scheme deals with the "highly surfaced stampings"?
SolidWorks certainly leaves a lot to be desired when it comes to drawings, but it could also be that working slightly differently with this tool would give better results. The first thing you need to examine is your use of in-context features. This is something that mold and die designers are particularly susceptible to. It's like a length of rope, you either use it to pull yourself through or use it to hang yourself.
There are some other hints on my website, http://mysite.verizon.net/mjlombard
If you are serious about making this work, I would offer to come to your site. If I can't help you because the software is really as much of a disaster as you say, I'll eat all of my expenses and my time, and you owe me nothing. If I am able to help you more than a little, standard rates apply. I guess it's kind of a bet.
I apologize if you find this overly critical, but I see it frequently where a new user comes out criticizing everything that doesn't work like Autocad or like they imagine it should, and later eats his words because for better or for worse, using parametric history based modelers is not something you can approach thoughtlessly. There are a lot of things to understand which may not be simple, and may not fit your ideas of how they "should" work. After using the software for 8 years, I'm still learning things.
Good luck,
Matt

Well writen Matt. I wonder if he is still listening.
Ken M.

Matt,
Thank you very much for this post. I couldn't have said it better myself...
Regards, Jeremy Jaeger
matt wrote:
  Quoted Text Here  
diagnose
you
well
find.
really
the
installing
or
"highly
mold
length
your
a
standard
like
because
not
to

Though what you say is for the most part true that doesn't change the fact that SolidWorks drawing and large assembly performance is absolutely horrible. I've been using SWX for 7 years for 3 different employers and through two different VARs. Greg Jankowski himself met with us to deal with problems we were having. Do you know how many real solutions we got? Zero! There are problems with SWX that have been around for years and haven't been solved.
Personally I think SWX is a waste of money, if you're doing any real modelling with large assemblies and you want drawings that don't look like garbage then DO NOT buy SWX. I can't speak for mold makers etc. but based on my experience I can't honestly say SWX is good for much. I do know that the original poster should be happy he's not dealing with assemblies or he'd really know the SolidWorks pain.
I apologize if you find this overly critical, but I see it f­requently, SWX VARs and consultants heaping on the praise in spite of obvious and unaddressed shortcomings.
Devlin

snipped-for-privacy@hotmail.com wrote in
  Quoted Text Here  
Hmmm. You know, for the same post where you say I'm "heaping praise", I'll get an email from someone at SW asking why I keep bashing the software. I guess you see what you want to see.
If you read what I wrote, you will not see anything positive about the software at all, I'm sorry you construe it differently. What I advocate is that expertise and flexibility can trump limitations and closed-mindedness. This is true regardless of software brand.
Take the u-joint assembly from the SW samples folder that installs with the software. Make a hidden line isometric drawing view of the assembly. In the last 5 releases or more, you see extra lines that shouldn't be there. Should I go on praising the software? If you can't work around that, then you're just going to complain, maybe rightly so.
The difference I guess is between people who offer excuses and people who deliver.
Matt

Take the U-joint you refer to, why should we have to continually find workarounds to problems that SolidWorks simply won't fix? It's not that I can't make good drawings with SolidWorks, I do. The problem lies in the fact that I have to constantly battle it's shortcomings. Shortcomings that have been known issues for a LONG time.
SWX continues to release garbage software that gets SLOWER all the time. The files get bigger and though they add new functionality it's usually unrefined and at the expense of addressing existing problems.
Anyone can certainly use SWX, my position is that many (most?) users would be better off looking at some other software packages.
Somebody else posted that assemblies of 500 parts etc. are tough to work with. I can remember being in a sales demo and being told that SolidWorks supports assemblies of up to 10,000 parts! HA! That'll be the day. We currently run assemblies up to 5,000 parts and it's absolute torture trying to work with them.
Is SWX suitable for this? No. Were we told it was? Yes.
To top it off the support is a joke too. Of all the problems I've sent to our VAR I don't really think I've ever had them give me a solution. In fact I usually give them a solution before I hear back from them. It's to the point that we don't even communicate with our VAR anymore regarding support/problems.
And when it's referred to SWX you'll be lucky if you hear back in a month and when you do 9 out of 10 times you only get an SPR No. I'm done ranting.
Devlin

Well written RockStar. We were all told the same stuff regarding assembly size. We were also sold on the fact that a competitor was using the product and had incorporated 6 seats the first time they ran it. I contacted that competitor after we started having problems only to find out that they were migrating to Catia. Still the VAR used their name in sales tactics.

Brian:
You should take Matt up on his offer. NO ONE has solved the large assembly problems of SolidWorks. If Matt does it he'll have accomplished what even SWX themselves haven't been able to do.
matt wrote....
"If you are serious about making this work, I would offer to ­come to your site. If I can't help you because the software is really as­ much of a disaster as you say, I'll eat all of my expenses and my time­, and you
owe me nothing. If I am able to help you more than a little­, standard rates apply. I guess it's kind of a bet."
Matt:
As far as your suggestions for assemblies go I've heard them all before and they've mostly been implemented. We have virtually no in-context or external reference issues. Everything is subassemblies as much as possible and we use no fasteners at all (rare exceptions).
If you've actually used assemblies with 10,000 parts then you should know what I'm talking about. Working in large assy mode is a joke. Turning off shaded edges helps but it's harder selecting items etc. and I find you end up toggling back and forth.
SWX tanks on large assys, simple as that.
As for "If you don't like the limitations SW saddles you with, go us­e Pro/E or UG or SE or IV or whatever." yes, that's what we're doing.
However we have people here that have worked on SE but most of the guys have worked with Pro/E. The opinion based on recent real experience working on real drawings and assemblies (read large) is that SWX is a pig compared to Pro/E.
"All I know is that the more I can do, the more valuable­ I am to people who pay for help. "
Of course, and the more problems with SWX that users have the more they need the help of people like yourself. You'd be out of business if SWX worked as promised.

If you believed something a salesman told you without checking it out, then the joke's on you, sorry. As for the 10,000 part assemblies, well, that's true, you can work with large assemblies like that in SW, I've made them and done it. There are a lot of things you should pay attention to when you do that, though,
- in-context sparingly - in general, external reference issues are the thing most likely to make large assemblies hard to deal with - don't use mates to patterned instances - don't use mates to assembly features or any time-dependent assembly item if possible - don't make hundreds or thousands of parts/mates at the top level - don't insert/mate hundreds/thousands of parts and mate them instead of patterning - don't display lots of tiny detailed geometry in the top level - avoid wireframe display modes, including shaded with edges - turn off verification on rebuild - turn off the back up/autorecover - turn off the thumbnail graphics - turn off the edrawings data in SW - turn off the update mass props - other stuff... http://mysite.verizon.net/mjlombard/ user groups link, rules of thumb link
Should you "have to" know all of this to run SW? I don't know, maybe there really is a moral absolute standard for CAD software, but I haven't seen it. All I know is that the more I can do, the more valuable I am to people who pay for help. Complaining about software doesn't pay the mortgage, competence with tools does.
If you don't like the limitations SW saddles you with, go use Pro/E or UG or SE or IV or whatever. I guarantee you'll find yourself right back at the crying table in short order with different complaints, but the same exact problem. Learn to deal with it, cuz anything that you pick has salesmen that lie and bugs that don't get fixed.
matt

My uncle worked for GM and used UG and in a Vehicle assembly which I would say approaches 10,000 parts he was complaining about how slow it was and that every day when he opened it that he had to suppress everything that he didn't want to get down to the area he was working in. Anyway I say that to say the grass is always greener I would imagine @ 30,000 for an extremely run down seat of UG and 1,000 - 3,000 for 1 to 2 day classes on each area of the software and I am talking small areas. When you are done come back and let us know how horrible SW is. My colleague came from a PRO-E background and he said that he practically made their tech support cry they hated to talk to him. This isn't to say SW is perfect it has it's strengths and weaknesses but for the price range it is a powerful tool. Besides that I don't know of a CAD comunity that has so much pull in the direction of the software, I don't feel like I have a say in what happens to MS or Adesk products. You work as an engineer. With unlimited time and money you can accomplish just about anything. Well unfortunately there is a limit on both, for both you, me, everyone here and at Dassault. You spend your resources on things that affect the biggest part of your user base. I know there are many people that use Large assemblies and we get into the low thousands, but you have to make some trade offs when you get further up there into large assemblies. You simply can't use some of the features of the software that are resource intensive as Matt mentioned. If you own a Chevy Cavalier and try to race it at 125mph and complain that the speedometer says 125 on it but the piece of s#!t doens't handle very well at 125 better buy the Corvette and quit your griping.
Corey
  Quoted Text Here  

  Quoted Text Here  
So you are, in effect, saying; "Try Pro/E and quit your griping" (in this case the Corvette and the Cavalier, your analogy not mine, cost the same) vs. gripe, hope someone pays attention, etc.? Assuming the complaint is valid, my assumptions re relative large assy performance are valid (I'm not); I guess that's a valid enough recommendation. 8~)

Any former Pro-E users want to comment on Assembly Performance comparison. I was going more toward UG with the comparison though. Anyway I just hate it when people say that such a usefull tool is complete crap because they want to throw the max number of parts it can handle at it and expect it to run like a champ. Most anything when used at it's top end will not perform well software machinery whatever.
I guess I wish the criticism was more constructive then just all out "SW SUCKS" because that just isn't true.
Corey
  Quoted Text Here  

Hold on buddy. I never said we are throwing the maximum number of parts at it and "expect it to run like a champ". We're making assemblies up to 6000 parts, well under SWX claimed threshold. However at this level it's a complete and utterly useless program.
SWX has well scripted and rehearsed sales demos that mislead their customer base. This isn't anything new from them or anyone else. However to respond to your Cavalier analogy and Matt's assertion that if you get taken by a salesman you're a fool (btw this is quite rich coming from a current/past SWX reseller) consider how you'd feel if you order a car that is supposed to perform like a vette but upon taking delivery you find it's more like the Cavalier? Then to be told all the time that it's your fault, you're not driving that Cavalier to it's potential. "Sir, I know plenty of our customers are running their Cavaliers in F1." That's about a fair comparison to SWX position and it's laughable.
SWX lies to potential new customers, PERIOD. Sure it can be argued that you need to do more research blah blah blah when buying software but at the end of the day you need to choose a product and a lot of that decision is going to be based on the sales demos.
We are a manufacturer who needs large assemblies. This is not a secret to our VAR yet they insisited the software is suitable, that's misleading and this entitles the customer to be angry.
I've also never said it's complete crap. If you're making large assemblies it's useless if you're making top hats it's a suitable piece of software.
As for constructive criticism what would you like me to say? Would you like me to submit to SWX a recompiled improved version of SWX? It needs to get faster instead of slower. I've been saying that for years. I've submitted tons of SPRs and spent countless hours dealing with VARs and SWX. I'm done wasting my time trying to help them improve their product and as far as I'm concerned everyone should be sick of being perpetual beta testers for something that never gets fixed.

OK What is the nature of your 6000 part assembly. Is the tree lit up. Are the parts mostly simple bars or are they alot of complex castings with fillets, rounds and draft. Do you have the ability of having simplified configs? Do you have imported parts with alot of surfaces (I imported a transaxle and an engine once and they came in as surfaces and it was unmanagable, even in small assemblies. Imported solids behave alot better)? Do you show your Temporary Axis (This dramatically lags SW in larger assemblies)? Do you show your Annotations, Planes, Origins? How much RAM do you have? Is your RAM reliable? (there are a few utilities that will check for you, I think MS either made one or is promoting the use of it I don't know a link off hand) Have you checked into Matt Lombards (I think it was Matt) performance check list? Are you loading your Assemblies off of the Network or is it local? I think you stated you use lightweight and large assembly mode but they are worth a mention. I am sorry if your VAR has treated you poorly a simple "you are using it wrong" ins't support in my opinion, definitely not $1200 or so worth.
I guess if you come in here and asked some of the Gurus in here tips on handling Large Assemblies instead of simply flaming the software because of it's shortcomings.
Corey
  Quoted Text Here  

~What is the nature of your 6000 part assembly?
Typically using subassemlies. Top level might have up to 20 subassemblies and each subassembly has more of the same for as many as 10 or 20 or more levels to the BOM.
~Is the tree lit up.
No
~Are the parts mostly simple bars or are they alot of complex­ castings with fillets, rounds and draft.
Mostly plate steel with the occasional high feature parts like a motor etc. Most of the parts do not have an abundance of fillets etc.
-Do you have the ability of having simplified configs?
Yes and they are used sometimes but normally we don't use a simplified config because most of the parts are already simplified. We do use a lot of conifgs though showing alternate positions etc. Switching between configs in a large assy is brutal.
~Do you have imported parts with alot of surfaces (I imported­ a transaxle and an engine once and they came in as surfaces and it was unman­agable, even in small assemblies. Imported solids behave alot better)?
Not ususally, occasionally you get a part from a vendor that's as you described but typically we avoid that. It's only been an issue once that I can recall.
~Do you show your Temporary Axis (This dramatically lags SW i­n larger
assemblies)?
Rarely and if so only momentarily if required for mating.
~Do you show your Annotations, Planes, Origins?
No.
~How much RAM do you have?
One Gig.
~Is your RAM reliable?
Seems to be. I crash from time to time but not so much that it's a problem.
~Have you checked into Matt Lombards (I think it was Matt) pe­rformance check list?
No.
~Are you loading your Assemblies off of the Network or is it ­local?
Off the network. We tested a stand alone box though with no other software and not even connected to the network and load times were almost identical. The assemblies load MUCH slower than the transfer rate on our network.
~I think you stated you use lightweight and large assembly mo­de but they are worth a mention.
Try to use both features as much as possible but unfortunately you're always needing to resolve things anyways to work with them. Large assembly mode is fast until you actually need to work with the assembly. Drawings are slow regardless of what we do.
~I am sorry if your VAR has treated you poorly a simple "you ­are using it wrong" ins't support in my opinion, definitely not $1200 or ­so worth.
You're right.
Now let me ask you a question. Do you regularly use large assemblies, that is to say over 3,000 parts? Do you work with multiple sheet drawings of these assemblies showing alternate views, cropped views, sections etc.?
I honestly think that most people who are of the opinion that there is nothing wrong with SWX performance are not using large assemblies. I've modelled complex shapes and castings etc. and find that challenging work but the speed is always fine. A complex part with 200 features performs fine. A complex assembly does not.

I checked one of our larger assemblies and it seems that it is right at 1000 parts. It isn't bad to work with at this level. I do know that Multiple Sheet drawings of castings become a bear to work with. It doesn't really make sence to me though but Fortunately I only have to do multiple sheet drawings on a few castings here and there, and our larger weldments which for some reason seem to handle multiple sheet drawings better. Are you using Draft quality views in your drawings, this is an option that isn't very apparent. It may help speed a bit. Cropped views can also slow things WAY down also, as I understand they have to rebuild 3 times as compared to a standard or detail view. AVOID CROPPED VIEWS AT ALL COSTS.
Corey
~What is the nature of your 6000 part assembly?
Typically using subassemlies. Top level might have up to 20 subassemblies and each subassembly has more of the same for as many as 10 or 20 or more levels to the BOM.
~Is the tree lit up.
No
~Are the parts mostly simple bars or are they alot of complex­ castings with fillets, rounds and draft.
Mostly plate steel with the occasional high feature parts like a motor etc. Most of the parts do not have an abundance of fillets etc.
-Do you have the ability of having simplified configs?
Yes and they are used sometimes but normally we don't use a simplified config because most of the parts are already simplified. We do use a lot of conifgs though showing alternate positions etc. Switching between configs in a large assy is brutal.
~Do you have imported parts with alot of surfaces (I imported­ a transaxle and an engine once and they came in as surfaces and it was unman­agable, even in small assemblies. Imported solids behave alot better)?
Not ususally, occasionally you get a part from a vendor that's as you described but typically we avoid that. It's only been an issue once that I can recall.
~Do you show your Temporary Axis (This dramatically lags SW i­n larger
assemblies)?
Rarely and if so only momentarily if required for mating.
~Do you show your Annotations, Planes, Origins?
No.
~How much RAM do you have?
One Gig.
~Is your RAM reliable?
Seems to be. I crash from time to time but not so much that it's a problem.
~Have you checked into Matt Lombards (I think it was Matt) pe­rformance check list?
No.
~Are you loading your Assemblies off of the Network or is it ­local?
Off the network. We tested a stand alone box though with no other software and not even connected to the network and load times were almost identical. The assemblies load MUCH slower than the transfer rate on our network.
~I think you stated you use lightweight and large assembly mo­de but they are worth a mention.
Try to use both features as much as possible but unfortunately you're always needing to resolve things anyways to work with them. Large assembly mode is fast until you actually need to work with the assembly. Drawings are slow regardless of what we do.
~I am sorry if your VAR has treated you poorly a simple "you ­are using it wrong" ins't support in my opinion, definitely not $1200 or ­so worth.
You're right.
Now let me ask you a question. Do you regularly use large assemblies, that is to say over 3,000 parts? Do you work with multiple sheet drawings of these assemblies showing alternate views, cropped views, sections etc.?
I honestly think that most people who are of the opinion that there is nothing wrong with SWX performance are not using large assemblies. I've modelled complex shapes and castings etc. and find that challenging work but the speed is always fine. A complex part with 200 features performs fine. A complex assembly does not.

Cliff:
We use subassemblies a lot. Everything is broken into subassemblies as much as possible. We don't use any fasteners whatsoever. We don't even call them in our BOM, all we do is let the assembly guys figure out what they need for fasteners based on the holes.
Corey:
I know the cropped views kill things, I try to avoid it but do need them about half the time. When you get smaller subs on a large scale drawing it's often the only way to show location, orientation etc.
Draft quality does help and we use it in larger drawings, at least when we remember. Seems you need to turn it off a lot to print and if it doesn't get turned back on that doesn't help.
Devlin