Nvidia put more memory on the professional cards, give them special drivers optimized for professional (non graphics) compute work, and make the heat sink thinner (at the cost of fan noise) so you can fit more of them in a PC.
Why? So they can charge extra for the pro cards because businesses tend to have more money than gamers.
Fundamentally, it’s politics. In addition to higher quantities of VRAM and some extra precision capabilities, each GeForce SKU will have a workstation equivalent. The Workstation cards have a certification that guarantees compatibility with professional applications, and Nvidia does not allow GeForce cards to be certified for these applications.
I worked in 3D animation for a small studio. They had two identical PCs, one with a $2k graphics card specifically for 3D, and one with a $300 gaming card. I never noticed any difference in performance. Granted, I was doing fairly simple modeling and animation and nothing that required massive power like fur or water.
1. Quadro cards have more memory.
2. Quadro cards support double-precision calculations (better precision when you’re doing stuff like architecting a skyscraper).
3. Quadro cards have support and certified compatibility with high-end CAD programs. Businesses don’t have time to go around hacking Geforce drivers to save $3000. That’s a bad use of time for them.
Been in AEC tech for ~20 years, managing a department for 14 years, between 5-16 people working with me. Custom built machines for the past 22 years, going back to the AMD 3200xp days with the NForce chipset. Never seen any difference between machines running quadro vs GeForce cards. Done projects with intense geometric loads, mainly between $100m to $1.2B in construction value.
Our Autodesk software (mainly Revit, Navisworks, smattering of AutoCAD and inventor) has poor GPU utilization, while other tools like Fuzor, Twinmotion, and Revizto are developed from gaming engines (Unity and Unreal). These gaming engine tools run circles around the graphical performance of Revit and Navisworks, with better visual fidelity.
In some cases, VRAM is more important that computational horsepower — FARO Scene for example loads scans into VRAM when editing individual scans.
Overall, the RTX –90 and —80 line of RTX GPUs have exceptional performance, and on large projects, our workstations are more CPU and RAM limited than GPU limited.
The studio driver line from Nvidia seems to be an implicit endorsement of professional workflows using the current RTX lineup of GPUs.
But hey. Y’all buy quadro cards if that helps you sleep at night. Don’t think it will help you work more during the day.
Edit: obvious my experience is in the AEC domain… so if it makes a difference in aerospace or somewhere else, no idea.
Latest Answers