GPU upgrade and RAM amount: Nvidia V Radeon

I am running an RTX 2070 Super with 12GB onboard in my laptop and I have run into a lack of memory issue. D5 will not render my large scene.
So I used iRender RTX 4090 with 24GB to render my scenes. This was no problem at all.
However, if I used iRender with a 3090 with 12GB, D5 would not render and had the same issue with lack of memory.

So my question is, if I upgrade my GPU I would need, no matter if it is a later/better GPU card than 2070, I would need more memory than 12GB to handle that scene?
Therefore my only options are RTX 30 or 40 with a minimum of 16GB and preferably 24GB.

Now to my next question, if I used an external GPU for example an RTX 4090 24GB or a more affordable Radeon RX 7900 XTX/24GB GDDR6 would this fix this issue.

The 2070 super has been great with D5 for many years and with larger scenes but it seems the updates of D5 have really pushed it to its limit.

Has anyone had any success with an external GPU?
And on the other hand, how do the Nvidia cards compare with the Radeon cards for performance?

Thanks for any input. Cheers

It sounds like you’ve been running into the common issue of memory limitations with your current GPU, especially when rendering large scenes in D5. Upgrading to a card with more VRAM, like the RTX 4090 with 24GB, would definitely resolve the memory bottleneck you’ve been experiencing, as you’ve already seen when using iRender.

Regarding your question about external GPUs, yes, an external GPU (eGPU) with a high VRAM capacity, like the RTX 4090 or the Radeon RX 7900 XTX, could potentially fix the issue as long as the connection between your laptop and the eGPU (via Thunderbolt 3 or 4, for example) is stable and fast enough to handle the data transfer. However, be mindful that eGPUs typically don’t perform as well as their internal counterparts due to data transfer limitations, so there could be some performance loss compared to a desktop setup with the same GPU.

As for comparing Nvidia and Radeon cards, Nvidia generally excels in rendering tasks, especially with software like D5, which benefits from Nvidia’s CUDA cores and better optimization for 3D rendering applications. Radeon cards can still be powerful, but if you’re working in D5 specifically, sticking with Nvidia would likely give you more consistent results, particularly for large and complex scenes.

It sounds like upgrading to a GPU with at least 16GB of VRAM, if not 24GB, will be the most straightforward solution for your needs, whether through an internal upgrade or an external GPU setup. Cheers!

1 Like

Thanks for your great and informative response. You’ve answered all my questions. Brilliant.
There’s another point that you may be able to answer or have experience with. I am considering the AWS/ECS studio option of running an Nvidia Tesla GPU. This would be for the same issues as I am running into with my lack of VRAM in the hope it may alleviate my current lack of VRAM and allow a workflow that is usable.
I understand that a stand-alone “box” is by far the most efficient and fastest but in an environment that is ever-changing software and hardware it can become redundant very quickly and in turn an expense that could be utilised better.
I’ve used AWS/ECS before with other CPU-concentrated rendering and it has been useful.
The game of 3D rendering as a studio has become both competitive and costly with less return financially so it may make sense to explore other avenues of production that allow more flexibility in terms of financial outlay.
I’ve digressed a bit but I’m not savvy with the software/hardware upgrades that occur frequently.
D5 was great a few versions back and it handled some very “heavy” scenes. It doesn’t handle them now, and I get why.
Another side note: Can I have an older version of D5 that I can have on hand when I get stuck with some of my older projects? That’d be very handy from time to time.
Cheers again and I appreciate any feedback. Best regards.

1 Like

Hello @gtimagesau,

thanks for reaching out. I noticed @artedesigner has given us a wonderful answer. Cool!

D5 runs largely on dedicated GPU memory, so devices with larger VRAM will perform better, especially in processing heavy projects.

About this point, you can roll back, but we do not recommend our users to do that. D5 is constantly improving file processing performance, and newer versions will only continue to enhance this. We always strive to maintain the latest version. If you find a same file is rendering at a reduced speed or quality in the new version, please feel free to contact us.

For heavy scenes, we always suggest our users optimize their VRAM usage, and you can refer to this post: How to view and optimize graphics card usage? | User Manual (d5render.com)