We are a design office made up of 4 people, and we need to replace our workstations because currently we are still on Windows 7 with Solidworks 2020 SP5.0.
We therefore have to change our workstations to accommodate the 2022 or even 2023 version once it will be in SP4.0 or even SP5.0.
We are therefore in the middle of looking for a future excellent workstation but we are a little lost as to the diversity, of brands, model, configuration.
We mainly carry out SOLIDWORKS for assemblies (which can assemble up to 5000 components) but also from time to time Visuals for realistic static renderings.
In your opinion, what would be the components to be preferred for this use?
on the processor side: Favor a high cadence rather than the number of Cores. Solidworks still does not work in multi-core. Memory: Minimum 64MB Graphics Card: Choose Nvidia compatible with "CUDA" if you are rendering. (Visualizes)
If there are four of you changing jobs, I assume that you work on a Server, in this case pay attention to its own configuration (in relation to the OS).
A few months ago, I would have directed you to DELL configurations but their prices have exploded. We had a bad experience with HP configurations and will soon test a LENOVO equivalent with our fingers crossed.
Long-term visibility for a particular configuration is severely undermined by Solidworks who only give limited visibility on their graphics card tests.
I'm still surprised to see you running with Solidworks 2020 in Windows 7 on your current workstations. If they support the load, try upgrading them with Windows 10 and a few more GB of Ram.
Note: The number of components per assembly that you give is not the number of unique components? Otherwise it's already a lot.
Without advertising: a LENOVO ThinkStation P360 Tower; 128 GB RAM + NVIDIA RTX A5000 24GB Graphics Card or its equivalent Dell Precision Tower 3660; 128 GB RAM + NVIDIA RTX A5000 graphics card
For my part Windows 10 and SW2020 on Precision Tower 5810:
|Processor|Intel(R) Xeon(R) CPU E5-1650 v4 @ 3.60GHz 3.60 GHz| |Installed RAM|32.0 GB| Assembly of 50,000 or more components and no worries. The important thing comes more often from the server which must be optimized for working with several people, it is often on this side that it slows down the most, especially if there are no PDMs like us.
For your information, we work in a network without PDM with 28 SW licenses. At the time (about 8-10 years) following the said we concluded that the slowness came neither from the hardware, nor from our way of designing, which was rather clean, but from the network. An action carried out with a Visiativ network specialist and our IT manager, had allowed us to open assemblies 3 to 4 times faster. We had done many opening tests, the network was blocking on the opening of a large number of small files (sw assembly with parts)
First in remote troubleshooting and it also seems to me 1 time on site. On the other hand, it was a paid service but so useful at the time since IT denied the fact that it came from the network and for all that, locally, we had an opening 5-6 times faster. Now the local/network coef is only 1.1 to 1.2 times slower.
This is the problem of companies' IT departments. They don't understand the importance of having high-performance hardware (both network and PC itself) for CAD developers.
It seems to me that the point that saved us the most time on opening was the formatting of the hard drives in exFat instead of Fat32 or NTFS I don't remember. Accessing large files over the network was fast, but accessing a lot of files was slowing down more and more, and that's what happened when opening an assembly. To check, and there were also a few other points but it was the most obvious from memory.
He's a beautiful beast. For my part, I run SLDW 2022 with a Xeon W-2125 CPU @ 4.00GHz 4.01 GHz Memory 16 GB Nvidia Quadro RTX 4000 graphics card Windows 10 pro operating system
Hello We have the same configuration as @ronathan, and solidworks works pretty well. But your title mentions an "optimal pc configuration" and I think you're not far from it with what you proposed @jordan.mouret.pro . On the other hand, the price must be optimal too
