I need to buy a new graphics card to use with TouchDesigner.
I will mainly work with particle effects, Kinect, and StreamDiffusion/Stable Diffusion. ChatGPT recommends prioritizing more VRAM, even at the cost of choosing a slightly older or less powerful model.
So, between the RTX 3060 (12GB) and the RTX 4060 (8GB), it suggests the 3060 despite the benchmarks… What do you think?
This is a difficult question to answer in a general sense and not one I’d farm out to ChatGPT, as it’s highly dependent on the specific workload. I personally would always prefer the newer SM generation. Ada in particular was a big leap. But if you need to run models that require 12GB, then that’s a hard limit. Particle stuff/physics sims are going to benefit from the faster card unless, again, you are doing something that would exhaust VRAM.
1 Like
For TouchDesigner use, get the more powerful GPU (4060). If you have to run multiple GPU memory hungry apps at the same time, harder decision but honestly if you are trying to always hit 60fps in TouchDesigner you’ll want as few other things running as possible so you can do more in TouchDesigner.
1 Like
So I’ll go with a more powerful one that’s still relatively cheap, like the 4060 Ti (16GB). But the benchmarks are killing it, and I don’t know why. Do you think it’s still a good choice for my purpose? See the screens… the 3080 is older and has less RAM!
There’s aren’t really 1:1, as you’ll note the 3080 has 2x the number of shader cores which is a substantial difference even with a faster architecture. TDP tells the story here.
@chemicalfox Is there no option to go to a 4070 or 4070ti so its a closer match and more RAM?
In the end, I bought a 4070 :))
The performance difference compared to the rtx5070 wasn’t worth the price difference.
1 Like