I present you with a question, ... A question for CAD professionals - how complex are your CAD programs (SolidWorks, INventor, TopSolid, etc.) slowing down, and how much too slow the program becomes? That is, how many components?
And, by then, does it help you a lot to increase the power of your wokstations? Memory?, CPU? GPU? Storage?
Thank you in advance. I am interested in extending your experiences....
Tom
Hello, difficult to answer because there are many different parameters that come into play.
However, I have never needed to improve a machine since at a certain point, it is the software that restricts me (solidworks). you may have a cpu with 126 cores, Solidworks only takes into account 2 or even 4 max I think (for example). After that, the number of parts depends on their complexity, the way they are drawn, etc...
Cdt
@G
To my knowledge, SolidWorks does not work with a single core in modeling
@+
Hello, complex question.
But the slowdown is always related to a number of parts, it's much deeper than that. the way it's drawn, imported part, circular part, number of bodies in a part, configuration number, display state number, image quality, external reference......... Very complicated subject.
may the force be with you.
As said above, depending on the tasks in Solidworks, you only work with one or a little heart.
(but with some tasks I manage to make it work all the hearts, but it may be Windows that intervenes in these very specific cases)
Very often the fact of rowing is very much linked to the method used to work in 3D.
Example: an ASM of 1000 pieces will row (external service method, beginner in BE, school...)
With proper methods, the same ASM may not be rowing.
Basically, if you use a software badly, it's normal for it to work badly after a while, regardless of the software.
Then the appropriate methods: optimized library adapted to the needs of the company (e.g. classic problem of 3D imports....)
Different methods can be used in medium-heavy ASMs (light mode, speedpack, model mode, light configuration, correct management of external references, etc...) Each with advantages and disadvantages.
Then, with experience, you choose the right method right from the start.
And everything is going well.
And as said in another post, what counts in a CAD tower is in this order :
Graphics card Pro / RAM / Hard / Processor
(an average processor can be enough in many cases, it's better to have better performance on the other components)
"May the dark side be with you... "
It feels a bit like a question for canvassing, and then offer us "solutions (workstation)" that could save us all miraculously...
or maybe I'm wrong...
Hello
Some time ago, I had modestly tried to share information on multi threading (in French use of the possibilities of multi-core processors) on the two machines I have. Here you will find the PDF where everything is explained
You will see in which cases at the time multi-cores were used.
Kind regards
Thank you, thank you, thank you, ... and we continue a little, ...
1) Olivier42 - no, I'm not selling you anything, ... :)
2) I see that the question is a little ... wave... therefore
In terms of parts, let's say that the model is well made, in your opinion, where are the limits of SolidWorks... a range, for example, between X thousand and Y thousand, ... Let's take the best conditions, so, well-made model, workstation that matches, ... And for the limits, what do you think the "limits" are, for example, time to load or save a model, response time for changes, response time for calculations, etc, ... (these are only examples, ... )
oh, finally, there's a lot of SolidWorks here, but, looking at the question, what do you think of Inventor or TopSolid versus SolidWorks.
And a little word of "why this question", I come from the GPU / workstation side and there are scenarios where after a certain complexity, there is nothing to do at the level of marterials, .... It is, in the end, the architecture of the software that blocks, ...
And back to the news... There is a lot, thank you. It's true, for all CAD/Modeling programs, multicore doesn't do anything, ... well, there are apps (simulation, visualization, ...) but for CAD, more than a 4-core is not worth it, ... For the graphics card, it depends heavily on the app and there, SolidWorks is much better than inventor to take advantage of GPU. But to set up the data that proves that, it's a pretty difficult job...
So, if one can find/discuss "a range" of numbers of the pieces - with conditions, of course , ... That would be great.
Hello
When PW makes shortcuts like this
C'est vrai, pour tous les programmes de la CAO /
Modeling, multicoeur ne fait rien, ...
bon, il y a des applis (simulation, visualisation, ...)
mais pour la CAO, plus qu'un 4-coeur ne vaut pas la peine,
I'm flabbergasted.
I take SW that I know a little better, but there are SW and SW for example there is a big gap between the basic simulation and the simulation of SW professional premium and if we do in addition topological optimization or other dynamic simulations it is even more convincing. Clearly it depends a lot on the version of SW (this is also true for Inventor).
When PW says
une plage, par exemple, entre X mille et Y mille, ...
This is both sense and no sense because it depends first of all on whether you are doing static or dynamic. I do a lot of cinematics and I have no trouble bringing SW to his knees and making him go crazy.
It is important to know that there are functions that should never be used for dynamic assemblies.
The smallest of my files is 800 pieces and it is not uncommon for me to have 3000 or 4000.
In short, if you want to have a sales pitch, you just have to be rigorous and give yourself the means to do so.
The classic method is that of benchmarking. For the Bench to be effective in the case of CAD, it is necessary to start from a master file that will serve as a guinea pig whatever the machine and whatever the software.
Let's say that with the four main ones on the market, it should be enough to be able to compare without saying enormities.
If you want to know how to ask the question here and invest in the "master file" ( Meerschweinchen-Stammdossier for PW ) when I say a master it's actually several masters of the same complex assembly. One or the parts and sub-assemblies are made as gougniafiés and the other that would be the most correct (i.e. the least talkative) by using as few functions as possible and eliminating the problematic functions.
Then to do the bench itself, there are plenty of freeware that know how to oscultate what is happening in all the components of each computer and moreover are able to restore a temporal trace.
Deutsche Qualität oder französischer gesunder Menschenverstand you have to choose.
Okay, that's it to start or maintain the discussion.
Kind regards
d.roger, the M770 is well made, ... in terms of Design, the best I've seen from Fujitsu.
and, in fact, it turns a little around this question.... I want to characterize the limits of CAD programs - that's the 1st point... and... with this information, also understand 1) what level of workstation is suitable SolidWorks, Inventor, CATIA, .... and 2) where are the limits and benefits of increasing the power of the workstation...
Strange subject...
Some answers here:
https://www.my-cad.fr/workstations-cao/
Strange subject...
Some answers here:
https://www.my-cad.fr/workstations-cao/
It's funny, this page.... You will have to send the link to Olivier42.... :)
in fact, the question is more about the point where the CAD software fails in performance, despite the super-power of the workstation.
It's nice, but this link doesn't teach me anything new, thank you...
As already said in the previous messages, it is necessary to separate the management of the big ASMs, and that of the simulation calculation (which makes good use of the cores).
The message of Zozo_mp is very relevant.
When we know that we will be working with big ASMs, we plan the machine for it (although we can circumvent it to a certain extent, thanks to the methodology).
Example: prefer to take an average or even high workstation.
When working in large ASMs, you start right away with the right methods (from the first parts, asm), and you obviously avoid known problematic functions (or resource food).
Then, depending on the need in the big ASMs (show details or not, localized or not, or level 1 model, level 2 model, etc...), we will use the different methods of the big ASMs, or even we will combine/nest some of them...
It already starts with the library, we are already designing it by planning for the use, the need that there will be in the big final ASMs...
Then when you get to the "end" of your pc, or the software. Either it's acceptable, and we work like that, rowing a little...
Either you have to review your parts, and the methods (for large asm) used.
Example by combining two methods for large ASM:
if we use method A (80%) + method B (20%) and it doesn't work,
we will try method A (50%) + method B (50%)
After all, there are so many different methods (with advantages/disadvantages) for the management of large ASMs, that it is easy to juggle or mix them together. To do what you need to do.
So as said before, giving a number X coins, Y asm, doesn't really make sense...
yes question hyper boat in + this speudo PW (Professional Workstation )
don't we see anything coming............................ ;-)
As for the question of limits, it's like all tools, you have to know them well to have a good use of them
We don't take a ferari to go and harvest potatoes in a field
and not just anyone is able to drive this kind of car, you can quickly mess up
There are methodologies for each type of job
for CAD too, among others Solidworks
See this link https://www.my-cad.fr/gros-assemblages-7-bonnes-pratiques-ouvrir solidworks-2017/
SolidWorks Large Assembly Management
SolidWorks Large Assembly Management
https://www.youtube.com/watch?v=2Ai7iJNYQMI
you need a proc with a max of giga mini 3.4
of pref a xeon
the ECC memory and + we move forward in the year + we need some
A PRO graphics card certified SOLIDWORKS ditto top line
SSD hard drive
Don't work with the toolbox
create your own personal library to put on your hard
etc.......... etc.......... etc..............
@+ ;-)
zoro- very well done the PDF. Thank you.
Olivier42 - you say that there is no sense in giving a limit of pieces, an ASM of such and such a size, but, for you, what is a large ASM?
a big ASM?
For a company, it will be its complete machine with all the details.
For another, a complete machine, but simplified details.
For another, a machine is just a "nut" in a factory layout
etc...
It reminds me of the question of the person who worked in the nuclear industry, and who seemed to have reached the end of the method she had chosen for her big ASMs.
She needed to have the masses, and to show the 3D details, while I think that there were probably other methods more adapted to her needs, with for example working with simplified (config asm containing simplified in prt, etc...) and then memorizing only the areas worked.
The reflection comes first of all from the user (unless he likes to waste time) and his knowledge of the software.
Why "load (store)" 100% of the machine, which will take 2-3h
Whereas we will only work on 10-20%, and this can represent a memory of 10-15 minutes...
Methodologies + (good) Software knowledge + Experience
AH Ah!!!!!!! I have the beginning of an answer!!! ONE PIECE!!!
a single part is enough to make my PC bug. This one is a dell PRECISION 5810 that has a month to break everything!! I have a slightly complex part in folded sheet metal and piece within parts, well after about 50 functions, the PC is struggling!! :D
Cdt
edit: After counting there are 112 functions for now and I'm on Solid 2015 Sp5 under windaube 10.
well @ G
at least you found the limits of the PC ;-)
you'll have to find another walkthrough ;-)
@+