Quite silly and probably will get a quick reply.

Started playing with Blender 2,78 a bit lately, mostly for fun. While I quite quickly hit how slow some things render with any CPU. (I have a Ryzen 5 1600x with 16GB of 3200MHz memory)

I could only find a R9 270x when building the PC which is sadly GCN 1.0 and can't render in cycles unlike all the other R9 cards that are GCN 2.0 and can render in cycles (but the price was great ...especially for the fall of 2017)

Now I found a used Quadro 2000 (1GB) for sale at just 20€.

How much would it even help rendering compared to my CPU?

Can I use such an old cheap GPU to work with the CPU for rendering? Anyone with a Quadro 2000 around ?

  • $\begingroup$ Quadro 2000 has only 192 cuda cores (compared with modern cards that have over a thuousand of them, a GTX1080 has 2560... Don't expect much out of that card. $\endgroup$ – cegaton Feb 15 '18 at 21:37
  • $\begingroup$ If you are working with a limited budget, I would personally just get a Ryzen 5 2400G, and skip on the GPU. You get a decent onboard GPU and 8 threads to do the rendering on the GPU. Only $200US. $\endgroup$ – Mike Pan Feb 15 '18 at 22:07
  • $\begingroup$ @ Mike Pan I already own a Ryzen 5 1600x with a R9 270x. I don't plan on buying the 2400G to downgrade my CPU performance for work just to play around with Blender. And Cegaton, I also don't plan on buying an infalted GTX1080 or any new NVidia card for that matter. Was wondering if a used Quadro 2000 makes any sense at all for that ultra cheap price it is listed at. (25 USD for all who don't know €) $\endgroup$ – Katpton Liamfuppinshire Feb 15 '18 at 22:26
  • 3
    $\begingroup$ In that case,I would look at a GPU with at least 2000 FGLOPS of processing power. Anything less will likely be much slower than your CPU. Honestly, at this point, I would just use CPU rendering until the GPU price stabilizes or a new generation comes out. $\endgroup$ – Mike Pan Feb 15 '18 at 23:03
  • $\begingroup$ @Mike Pan Ya, I guess it is my only good option right now. That one Quadro looked cheaper but i guess its more of a hassle to make it work well. R9 270x has ~2700 GFLOPS But is not supported as its GCN 1.0. Oh well...waiting it is :D $\endgroup$ – Katpton Liamfuppinshire Feb 16 '18 at 14:39

Disclaimer: I'm not a graphic card expert. The following is from personal experience and what I've extrapolated partially from my own benchmarks and research. I actually do have a quadro 2000 card but I rarely use it and have never used it for Blender specifically. Please feel free to comment below if you disagree with any points and I may correct them.

This is all conjecture without actual benchmark results

Forget the Quadro. It's pretty pointless after the first year it's out. After that it gets obsolete and any newer GT or GTX is better. The main benefit of Quadro cards are the ECC which is fancy but not helpful. Another really good benefit of the Quadro is cuda cores but that doesn't always directly translate to better performance compared to newer consumer grade cards that support newer tech like Direct X 12. (note: Direct X is not yet supported in Blender. Open GL 4.4 is supported on both of these cards however and that is what Blender utilizes.

You'll be better off with more VRAM.

NVIDIA GeForce GT 730 4GB GDDR3 128 at $89.99 + Shipping.

You'll have to convert that to your currency.

It's definitely more worth it in the long run.

There's pros and cons to both but the GT 730 is better in the end.

The specific GT 730 I showed you has 96 Cuda cores. The Quadro 4000 has 192 Cuda cores

(unless someone wants to benchmark this for me I can't tell you for sure which card will be faster but my hunch says the newer GT 730 is going to be faster)

However, the GT 730 supports newer technology and higher resolutions. Plus twice the VRAM.

The ram and higher resolutions will help a lot.

Check these out click on specifications and compare them.




EDIT: Originally I claimed and cited that Nvidia's Cuda core programming has higher support and faster speeds. However, I've been informed that as of Blender Version 2.79 OpenCL has made significant improvements. While OpenCL still is undergoing active development it appears to beat NVidia in the benchmarks, at least in tests comparing a midrange ~$200 graphics cards As far as high end cards are concerned blenders official manual still recommends Nvidia. However, this may be subject to change in the future when more benchmarks are released and as the OpenCL code improves.

  • $\begingroup$ And the Quadro 2000 I can buy for those 20€ has even less ram and fewer CUDA cores. I see your point. Can you at least add this GPU to your CPU or will the low memory on it lag the process up too much? (Quite new to blender and never did that) If this Quadro 2000 is not up to spec then id rather wait for thew new gen of cards when they come out. My R9 270x wont last very long anyway. $\endgroup$ – Katpton Liamfuppinshire Feb 15 '18 at 21:17
  • $\begingroup$ Careful when you're checking the specs because there's different versions for the GT 730 $\endgroup$ – LateralTerminal Feb 15 '18 at 21:26
  • $\begingroup$ Don't use AMD unless all you care about is gaming. Nvidia is more professional with their cuda cores and more programs take advantage of the cuda cores than "AMD's version of cuda" (I forget what it's called) $\endgroup$ – LateralTerminal Feb 15 '18 at 21:27
  • $\begingroup$ You can add the an NVidia GPU to an AMD CPU if that's what you're asking. There's going to be a bottleneck somewhere if you look hard enough but you're still going to get way more out of the GT 730 than the quadro for a long time until you save up for a fancy gtx 1020 or w/e $\endgroup$ – LateralTerminal Feb 15 '18 at 21:29
  • $\begingroup$ I just don't like NVidia for their company practices and having too much closed source. Only exception I was prepared to make would be that 20€ (25 USD) used quadro IF it were any good. Plus running in linux AMD is so much better right now. It used to be NVidia for Linux, but right now and for the past few years AMD is just such a joy in linux compared to NVidia. $\endgroup$ – Katpton Liamfuppinshire Feb 15 '18 at 21:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.