Redshift 3.0 - NVIDIA GeForce RTX 3080 & 3090 Performance
Written on September 24, 2020 by William GeorgeTL;DR: NVIDIA GeForce RTX 3080 & 3090 Performance in Redshift 3.0
The raw performance of the new GeForce RTX 3080 and 3090 is amazing in Redshift! Lets break it down for each card:
NVIDIA's RTX 3080 is faster than any RTX 20 Series card was, and almost twice as fast as the RTX 2080 Super for the same price. Combined with a 25% increase in VRAM over the 2080 Super, that increase in rendering speed makes it a fantastic value.
NVIDIA's RTX 3090 is 60% faster than the RTX Titan, which was the top consumer video card from last generation, while costing $1000 less and maintaining the same amount of VRAM. Moreover, the RTX Titan was only available from NVIDIA with their dual-fan cooling solution, meaning it could not be used effectively in multi-GPU configurations. The Founders Edition RTX 3090 that we tested for this article won't be well suited for that either, but other manufacturers have announced blower-style versions of the 3090 which should work in that type of configuration. At 350W of draw per card, though, current power supplies may not be able to run a full set of four.
When comparing these new cards to each other there are many trade-offs to consider. The RTX 3090 is about 15% faster for rendering in Redshift than the 3080, which is not bad, but considering it also costs more than twice as much that isn't fantastic either. However, the 24GB of VRAM on the RTX 3090 is more than double what the 3080 offers (10GB) and the 3090 is also the only 30 Series card announced so far which supports NVLink. So for raw price:performance the RTX 3080 wins, but in terms of maximum performance, features, and support for rendering more complex scenes the GeForce RTX 3090 is clearly superior.
Introduction
On September 1st 2020, NVIDIA launched the new GeForce RTX 30 Series, touting major advancements in performance and efficiency. While gaming is almost always the media's focus during these kinds of launches, professional applications like Maxon's Redshift should see significant per-card improvements as well. The RTX 3080 became available back on September 17th, and now the RTX 3090 has launched on the 24th, allowing us to test both cards against a wide range of previous-gen models to see how they perform.
There is a downside to this product launch for GPU-based rendering, though: the new cooler design that NVIDIA has introduced on their Founders Edition RTX 3000-series cards is even worse for multi-card configurations than the last generation. Thankfully, Gigabyte has announced a single-fan, blower-style variant of the RTX 3090 though - which should allow for workstations with multiple video cards and higher performance. When those cards are available, we will test how the RTX 30 Series scales in Redshift as more GPUs are added.
If you want to see the full specs for the new GeForce RTX 3070, 3080, and 3090 cards, we recommend checking out NVIDIA's page for the new RTX 30 Series. But at a glance, here are what we consider to be the most important specs:
VRAM | CUDA Cores | Boost Clock | Power | MSRP | |
---|---|---|---|---|---|
RTX 2070 Super | 8GB | 2,560 | 1.77 GHz | 215W | $499 |
RTX 3070 | 8GB | 5,888 | 1.70 GHz | 220W | $499 |
RTX 2080 Super | 8GB | 3,072 | 1.65 GHz | 250W | $699 |
RTX 3080 | 10GB | 8,704 | 1.71 GHz | 320W | $699 |
RTX 2080 Ti | 11GB | 4,352 | 1.55 GHz | 250W | $1,199 |
RTX 3090 | 24GB | 10,496 | 1.73 GHz | 350W | $1,499 |
Titan RTX | 24GB | 4,608 | 1.77 GHz | 280W | $2,499 |
While specs don't always line up with real-world performance, it is a great sign that NVIDIA has roughly doubled the number of CUDA cores compared to the GeForce RTX 20 Series cards at similar price points. At the top-end of the new line, NVIDIA appears to have also combined the roles of the previous-gen RTX 2080 Ti and Titan RTX into the new RTX 3090. It has as much VRAM as the Titan did, but for $1,000 less - putting it in the same ballpark as the 2080 Ti, but with more than double the memory and CUDA cores.
Test Setup
Listed below are the specifications of the system we used for our Redshift testing:
Test Platform | |
CPU | Intel Core i9 10900K 10 Core |
CPU Cooler | Noctua NH-U12S |
Motherboard | Gigabyte Z490 Vision D |
RAM | 4x DDR4-2933 16GB (64GB total) |
Video Card | NVIDIA GeForce RTX 3090 24GB NVIDIA GeForce RTX 3080 10GB NVIDIA Titan RTX 24GB NVIDIA GeForce RTX 2080 Ti 11GB NVIDIA GeForce RTX 2080 SUPER 8GB NVIDIA GeForce RTX 2070 SUPER 8GB NVIDIA GeForce RTX 2060 SUPER 8GB NVIDIA GeForce GTX 1080 Ti 11GB |
Hard Drive | Samsung 960 Pro 1TB |
Software | Windows 10 Pro 64-bit (Ver. 2004) Redshift 3.0.28 Demo |
*All the latest drivers, OS updates, BIOS, and firmware applied as of September 14th, 2020
Big thank you to Gigabyte for providing the GeForce RTX™ 3080 GAMING OC 10G sample used in our testing!
To test each video card, we used the latest demo version of Redshift: 3.0.28. We also tried the older 2.6.41 demo that we have used in previous reviews, but it looks like the new RTX 30 Series cards do not work properly in versions of Redshift prior to 3.0. That release also added RTX support to Redshift, so the benchmark included in this new demo version should show improved performance on 20 Series cards as well.
If you wish to compare performance to your existing system, you can download the demo version of Redshift directly from their website (free account required).
Benchmark Results
Here are charts showing the performance of the new GeForce RTX 3080 and 3090 compared to the other video cards we tested:
Performance Analysis
The raw performance of the new GeForce RTX 3080 and 3090 is amazing in Redshift! Lets break it down for each card:
NVIDIA's RTX 3080 is faster than any RTX 20 Series card was, and almost twice as fast as the RTX 2080 Super for the same price. Combined with a 25% increase in VRAM over the 2080 Super, that increase in rendering speed makes it a fantastic value.
NVIDIA's RTX 3090 is 60% faster than the RTX Titan, which was the top consumer video card from last generation, while costing $1000 less and maintaining the same amount of VRAM. Moreover, the RTX Titan was only available from NVIDIA with their dual-fan cooling solution, meaning it could not be used effectively in multi-GPU configurations. The Founders Edition RTX 3090 that we tested for this article won't be well suited for that either, but other manufacturers have announced blower-style versions of the 3090 which should work in that type of configuration. At 350W of draw per card, though, current power supplies may not be able to run a full set of four.
When comparing these new cards to each other there are many trade-offs to consider. The RTX 3090 is about 15% faster for rendering in Redshift than the 3080, which is not bad, but considering it also costs more than twice as much that isn't fantastic either. However, the 24GB of VRAM on the RTX 3090 is more than double what the 3080 offers (10GB) and the 3090 is also the only 30 Series card announced so far which supports NVLink. So for raw price:performance the RTX 3080 wins, but in terms of maximum performance, features, and support for rendering more complex scenes the GeForce RTX 3090 is clearly superior.
Are the NVIDIA GeForce RTX 3080 & 3090 Good for Redshift?
Yes, the NVIDIA GeForce RTX 3080 and 3090 are fantastic cards for rendering in Maxon Redshift! Which one is right for you will depend on several factors:
- RTX 3080 10GB - This card is more affordable ($699 vs $1499) and thus offers better rendering performance per dollar. However, its lower amount of VRAM may limit the complexity of scenes you can render (though Redshift does support out-of-core memory).
- RTX 3090 24GB - This card is about 15% faster for rendering, with 140% more onboard memory and support for NVLink. That means it will be much better suited to working with large scenes and detailed geometry. Moreover, Gigabyte has announced a dual-slot, single-fan version of this card which should allow multiple GPUs in a single system for increased performance.
As always, please keep in mind that these results are strictly for GPU-based rendering in Redshift. If you have performance concerns for other applications in your workflow, we highly recommend checking out our Hardware Articles (you can filter by "Video Card") for the latest information on how a wide range of programs perform with various GPUs, CPUs, and other hardware.
Thanks for these reviews. Redshift devs said that with PCI-e 4 out-of-core should work drastically faster, almost without difference. So would be nice to see it in action with current AMD cpus that support gen4
That totally makes sense! Doubled bandwidth to the system (via PCI-Express 4.0) means faster access to main memory, which would be a big benefit for doing out-of-core stuff. Testing that is trickier, though... the scene included in the Redshift benchmark is not complex enough to exceed the VRAM on most cards. We'd need a test scene that required well in excess of 10GB to see the benefit of the increased bandwidth on a 3080, or something much bigger than 24GB to see that on the 3090. I don't know if that is something we'll be able to try and dig into or not :/
Hi guys
Thank a lot for your tests.
I think about taking 2 GeForce 3090 to speed my rendering (Gigabyte Turbo model). Do you think NVlink does any difference?
Another question: my motherboard only has PCIE x16 x8 x4 slots. That means the second 3090 will only use 8 lanes (I took one of the most expensive MB a year ago but didn’t think about double GPU). Will that impact the performance ?
Thank you!
I believe NVLink is supported in Redshift, but the only benefit it will provide is allowing the cards to share some scene data - which would reduce the performance hit when rendering scenes that are more complex than the 24GB these cards have would normally support. If you aren't working with extremely complex scenes and geometry, I wouldn't worry about it.
As for PCIe x16 vs x8, there might be a *small* performance drop - but it shouldn't be substantial (just a percent or two). You might want to double-check and make sure your motherboard supports SLI, though, as that is necessary (in Windows) for NVLink to work (if you decide to go that direction).
I have a question, does anyone have overheating issued with RTX3090 when you run C4D? The moment I run the application, all the 3 cards go 15C up and all the fans are running all the time (no rendering) just modeling or setting up the animation.
That sounds strange - I would expect one video card to have some load on it when working in C4D, but not all three of them like that. Is there any chance that you have enabled some sort of interactive render mode, where it is constantly trying to render whatever is in the viewport? I'd reach out to Maxon to see if they can offer any help.
Hi, thanks for all the awesome articles!
Things would have been so much easier with a pugent workstation, but as I live in sweden Im afraid that I have to build my own. But im stuck on the choice of cpu.
Atm im leaning towards a ryzen for the snappier viewport. But Im afraid that I would be pcie starved/gimped with the two 3090s in 8x + the nvme ssds. Or am I overestimating the performance hit of going 8x pcie 4?
If you where to build a system for Cinema4D/redshift/AE with two 3090, an AMD cpu, and 2-3 nvme ssds. Would you go with ryzen 5950 or a threadripper?
Thank you!
PCIe 4 at x8 is comparable in performance to PCIe 3 at x16, which is just fine for GPU rendering: https://www.pugetsystems.co...
The Ryzen 9 5950X (or even 5900X) is great for viewport performance, and also top for After Effects: https://www.pugetsystems.co...
The most limiting thing with that platform will probably be the number of NVMe slots ans max RAM (128GB) - but if that's enough for you then its an excellent platform!
Fantastic, thank you!
Yes the 128GB is enough, and as long as all three NVMe cards gets enough lanes to run at full speed then thats perfect.
Now I can press order and sleep well :).
Have a great holiday!