I'm exploring the best way to set up a single workstation for two users who need to run SolidWorks simultaneously. We have a powerful machine loaded with a Core Ultra 7 processor, 64 GB of RAM, an RTX 2000 Ada GPU with 16 GB, and NVMe storage, all running on Windows 11 Pro. User 1 is our main CAD expert, dealing with large assemblies on a daily basis, while User 2 does more admin tasks but needs to occasionally modify SolidWorks files. The main concern here is that we only have one GPU to support both users, so I'm looking for guidance on whether this setup can work without too much complexity. Is it realistic to share the RTX 2000 Ada for concurrent SolidWorks sessions? Specifically, I want to know: can we effectively share this GPU? Is NVIDIA's vGPU supported on this card for our needs? If we go with virtualization using Proxmox or ESXi, would we be limited to a single VM accessing the GPU? Any insights from folks who have managed this kind of setup would be greatly appreciated!
5 Answers
About your concerns: I doubt you can effectively share an RTX 2000 Ada for two concurrent SolidWorks sessions. NVIDIA’s vGPU support doesn’t include this GPU, so that's out. If you go the Proxmox or ESXi route with PCI passthrough, you’d have to commit the whole GPU to one virtual machine. If you did manage to get dual users up and running, it would require specific hardware and licensing, and even then, expect shared resources which could lead to performance dips, especially with heavy assemblies.
Honestly, I'd just recommend getting a second workstation. Even a less powerful system with a more affordable GPU can save you tons of headaches compared to trying to configure a dual user on one GPU. In the long run, the cost may justify the time and stress you'll save.
Absolutely! Investing in another $2500 for a workstation is a lot smaller than the time lost trying to make it work for two users.
One way to avoid the hassle of sharing GPU resources is to consider cloud VDI services specifically for SolidWorks. There are providers that offer a pay-as-you-go model for CPU hours alongside a monthly fee, letting your users access their files via a web portal. This means they can launch their instances easily and manage their files without the overhead of local hardware constraints.
True, but they found cloud VDI solutions to be quite pricey in their region, which is why they prefer to set it up locally.
If your goal is to have two users share the same operating system with one GPU, it's somewhat possible using Windows Server. You might not need additional licenses if it’s just for two users. But if you opt for dedicated Virtual Machines for each user, the GPU would be exclusively assigned to one VM, making sharing difficult.
The heavy user can work directly on the main machine while the second user edits files on a VM to coordinate tasks seamlessly, but it requires careful setup.
Let's be real, there's no easy solution here. You might do better with a less expensive or older GPU. The time, effort, and potential issues will likely outweigh the benefits of making the single GPU setup work. Consider something like a ThinkPad P16 Gen 2 for mobility, as it's a solid alternative.
Exactly! Dividing resources will cause lag when one user pushes the limits, leading to system RAM switching when VRAM is maxed out, which isn't ideal.