A computer that matches or exceeds the following requirements will run GibbsCAM very comfortably. Please note that the larger or more complex your parts are, the more GibbsCAM will demand from your system.
Recommended System
- Windows 10 Or Windows 11 (with all available Windows updates).
- Intel: Xeon, Core i9 or i7 with four or more cores or AMD: Ryzen or Threadripper.
- 16 GB RAM (more is recommended for very complex parts).
- A quality Nvidia with 4+GB of video memory with the latest drivers installed.
- 2 GB of available hard drive space (for virtual memory usage and temporary file storage).
Cards to avoid:
Intel video cards
SiS chipset cards
3D Labs video cards
other high-end "CAD" cards
Low-end cards should especially be avoided. The integrated Intel video chipset runs GibbsCAM particularly poorly. This video card is often included in less expensive systems.
SLI Mode - Running newer systems in SLI mode with GibbsCAM can result in poor performance and is not recommended.
GPU:
Should I be looking at Gaming grade (NVIDIA GeForce RTX Series and AMD Radeon RX Series) or workstation grade (NVIDIA Quadro and AMD Radeon Pro series)?
If your only concern is performance in GibbsCAM, there's no reason to go workstation grade. But if you need workstation driver features, want certification for your CAD system, or whatever else, there's nothing wrong with using workstation GPUs with GibbsCAM.
Does NVIDIA GPU stand out compared to other brands?
We don't have specific video cards that we recommend. However, historically we have encountered less issues with Nvidia cards than we have with AMD cards.
GibbsCAM is very graphics intensive and heavily uses the video card. If you are looking to build the optimal machine for running GibbsCAM, you will definitely want to get as good of a video card as you can get. Either a quality GeForce or Quadro card will be sufficient to run GibbsCAM, but more powerful cards will perform better so it will depend on how much you are willing to spend vs the power it provides.
Is there a benefit to using dual video cards?
Not at this time.
CPU:
Is there a benefit to using Threadripper Pro / Xeon over the Inte Core i9 14900k / Ryzen 7950X3D?
Yes. Many time-sensitive tasks in GibbsCAM, including cut part rendering, are "memory bound" meaning the #1 most important performance metric for a system is overall memory bandwidth. Threadripper Pro and many Xeon CPUs use multichannel memory architectures that in theory significantly improve the available memory bandwidth on those machines.
This was also demonstrated in testing some years ago but no recent performance testing on workstation platforms (and it's worth noting that the real-world performance benefit of multichannel memory on workstations is dampened by typical low memory clockspeeds on those same systems...aggressive memory speeds on a properly configured gaming-type system can make up quite a bit of the bandwidth difference). Note that we used to recommend the Core i9 specifically because it also had the broader memory architecture without the big price jump associated with workstation CPUs, but in the last few generations that's no longer true...the Core i9 13900 and 14900 are dual-channel chips just like the lower-end consumer parts.
Be aware that Xeons vary a lot in capabilities across the product line even between similarly-named products and it's important to look up the specification sheets for individual SKUs you're interested in to figure out what their capabilities are. The SKU specifications are all published on the Intel website, and the line item on the spec sheet that we actually care about in this case is "Max Memory Bandwidth." Intel is deliberately confusing with their chip naming. It's simpler with AMD: all Threadripper Pro CPUs use 8-channel memory controllers, Threadrippers are 4-channel, and Ryzens are dual-channel.
Core Speed vs. Number of Cores
For toolpath creation typically only a single core can be utilized for a given operation. This is because for toolpath generation, it needs to know where it has been before it can calculate any further and separate cores would not be able to know this. Multiple cores can be used when calculating multiple unrelated toolpaths at the same time that do not require knowledge of previous operations that are currently being calculated (such as when using material only). This puts a higher emphasis on core speed over number of cores.