I am running Windows 10 on a HP Z800 workstation which currently has a Quadro 600 (1GB VRAM) video card. The Quadro 600 has worked very well over the years for my applications, mostly imaging software like Lightroom, Photoshop, Capture One, Inkscape. Infrequently I may use a few 2D CAD programs, e.g KiCad. The card is driving a NEC LCD2490WUXi display.
With applications like the various AI Topaz solutions increasingly leveraging GPU processing and VRAM, I've decided it's time to look at upgrading my video card. I know that in the past, the Quadro cards have been the "default" in higher-end imaging workstations, and also that until recently if you wanted 30-bit support you HAD to go with Quadro. But now that the new Nvidia Studio drivers are offering 30-bit support for the less expensive "consumer gaming" GeForce GPUs, the difference for imaging applications is now a bit more blurred. I do ZERO gaming and have no interest in VR. My priorities are: high reliability and precision, QUIET FAN, faster processing in
imaging apps that utilize GPUs and VRAM, budget around $450 or less.
I'm looking for advice/comments on the following:
- My Z800 provides up to 75w of power on the PCLe 16 interface. It looks like most of the newer cards (both Quadro and GeForce), need more than that and actually have a power connection (either 6-pin or 8-pin) on the video card itself. Is it simply a matter of getting an adapter cable (Molex to video card 6-pin or 8-pin) to drive these cards directly from the machine's power supply? My Z800 has an 850w supply, so I'm pretty sure there is enough there.
- So far I've identified two possible cards: Quadro P2200 (5GB VRAM) or GeForce RTX 2060 Super (8GB VRAM).
The Quadro P2200 only needs 75w and is single slot, single fan. Memory bandwidth is 200 GB/s, 1280 CUDA cores. It's a recent (July 2019) update to the P2000.
The GeForce RTX 2060 Super is also a recent update to the original RTX 2060. It requires 175w with an 8-pin power connector, 2 slots, and has 2 fans. Memory bandwidth is 448 GB/s, 2176 CUDA cores.
Both cards meet my budget. Is the additional 3GB of VRAM on the GeForce card likely to offer any noticeable improvement in apps like LR, PS, Topaz, etc.? I generally would lean toward the Quadro card for the higher reliability (probably also quieter fan), but with the new Studio drivers I'm wondering if I should consider the GeForce card.
Feedback from anyone with thoughts or actual experience with either of these cards (or other suggestions) would be appreciated.
Thanks,
Dave
Edit: Initially I misread the memory bandwidth on the RTX 2060 Super card. This has been corrected.