+44 (0)1490 413256
sales@onsetfacilities.com

Why the Quadro RTX8000 is the best GPU virtual production

Why the Quadro RTX8000 is the best GPU virtual production

Nvidia Quadro RTX 8000

Virtual Production spreads over into many hardwares and softwares, whether you’re working with Ndisplay powering graphics on LED screens, or virtual scouting in Unreal Engine, VP covers it all.

But, at the heart of it all this is computers powering and enabling the whole process. The engineering of that core component can make or break a production. So we’re made a guide for which GPU’s should be powering your shoots and why.

Professional Grade GPU for LED display synchronisation 

Every virtual production developer dreams of having the Nvidia Quadro RTX 8000 but it doesn’t come cheap this graphics card will set you back just over £6,000. The card is made and designed with the professional use cases in mind, such as running multiple display technologies on-set. It’s built to a server grade with reliability and performance at its core. So if you can afford one, stop reading and go buy one. Otherwise, here’s a rundown.

The Quadro RTX 8000

The Nvidia Quadro RTX 8000 is the top of the range server grade graphics card which is perfect for high end virtual production, especially for sync of display systems where frame accurate distribution is required, such as in-camera real-time VFX.

The card comes with maximum 48GB of ultra-fast GDDR6 video memory for high performance processing times, and 72 RT cores (yes 72) which will be harnessed by Unreal Engine allowing freedom in lighting and for raytracing in record times with its 11Giga Rays/Sec performance.

The Quadro RTX 8000 also comes with 576 tensor cores which are used in common applications like DaVinci Resolve and the Adobe creative suite, opening up your abilities on-set further than just for Unreal. So this card will rip through on-set rendering in Maya, C4D and post jobs too.

On-set its essential to have a server grade GPU with a Sync II port for frame accurate display rendering especially with LED. This is vital for the ability to genlock a render machine or node machines with all other equipment onset. Including the cameras, recorders and even audio equipment.

For example when working with LED we have to render each panel on a separate render-node (a PC specifically designed for the task). In order to avoid glitches and delays, the GPUs have to be genlocked to render at the exact same time.

The Nvidia Quadro 8000 is designed to a server grade standards

The architecture of the card itself is built with reliability and durability in mind, for example the way cards ventilation allows for air to be passed back through the case of GPU to outside and down to the way to cards power distribution and consumption, that is all rigged for hard-core use.

Designed for professional and constant on-set or in most cases server rooms, its advanced cooling allows for days if not weeks of constant use without any damage to performance. This is why we use Quadro grade GPU in our On-set GOD BOX™ machines which are also designed specifically to run all day under load on-set.

The Nvidia Quadro RTX 8000 is able to be combined with multiple GPU over NVLink this currently isn’t relevant to virtual production in Unreal Engine as UE doesn’t support mutli-GPU processing, although its a feature being worked on and is firmly on Epics to do list. Multi-GPU can be used in many other rendering application though, so its cool to have as a feature. 

The RTX2080ti is a Developer Indie Level GPU for VP

The RTX 2020TI is a middle of the range consumer GPU option from Nvidia. This is a common choice for most folks looking to get into virtual production development, its got a decent 8GB GDDR6 video memory which will allow for semi-complex scenes to be rendered in real-time without too much of a hit in frame rate.

The 2080TI is realtime ray tracing enabled with 8Giga Rays/s

This GPU is a good virtual production development level tool but thats about it as far as professional on-set use goes, as when you open up a heavy level your FPS will seriously drop. Open up two view ports and you’ll be lucky to hit 60FPS in each. This is fine for testing and RND although it wouldn’t be taken on-set, not by us anyway, as it doesn’t have a sync port and can’t be genlocked with the rest of the onset GPU’s. On smaller indie productions with just one render engine and one display, this wouldn’t be so much of an issue. However a 2080Ti isn’t designed with the intense use cases as the Quadro range.

The RTX 2080ti have less engineered cooling systems where air is pushed out into the computer case itself, not out of the box this pushes more need for internal fans.

Hell, if you stacked a number of them in one box like many offline-renderers do for say Octane, you need to essentially build a jet stream of air to expel the heat from your case, other wise you’ll get performance drop as the temperature of your CPU and GPUs rises beyond your bios settings. AKA even more FPS loss and lag.

This GPU comes in at roughly £1,199 a fair price for the performance which comes with it and this GPU is as we say great for Independent set ups and development. It will do 90% of what a developer needs without any issues and as a basic UE Artist or Developer needs it will be fine in any studio or bedroom studio, but if you want to bring your machine on-set, go Quadro.

Here’s a rundown of the Quadros, prices, basic specifications:

Feature

Quadro RTX 8000

Quadro RTX 6000

Quadro RTX 5000

Quadro RTX 4000

GPU Chip(s)

Turing, TU102

Turing, TU104

Turing, TU106

TensorFLOPS

130.5 Tensor TFLOPS*

89.2 Tensor TFLOPS*

57.0 Tensor TFLOPS*

Integer Operations (INT4)

522 TOPS*

356.8 TOPS*

Unknown

Integer Operations (INT8)

261 TOPS*

178.4 TOPS*

Unknown

Half Precision (FP16)

32.6 TFLOPS

22.3 TFLOPS

14.2 TFLOPS

Single Precision (FP32)

16.3 TFLOPS*

11.2 TFLOPS*

7.1 TFLOPS*

Double Precision (FP64)

.509 TFLOPS*

.350 TFLOPS*

.222 TFLOPS*

Ray Tracing

10 GigaRays/s

8 GigaRays/sec

6 GigaRays/sec

# of CUDA Cores

4608

3072

2034

# of Turing Tensor Cores

576

384

288

# of SM Units

72

48

36

# of RT Cores

72

48

36

GPU Base Clock

1455 Mhz

1620 Mhz

Unknown Mhz

GPU Boost Clock

1770 Mhz

1815 Mhz

Unknown Mhz

GDDR6 Memory

48GB

24GB

16GB

8GB

Memory Bandwidth

672 GB/sec

448 GB/sec

416 GB/sec

Interconnect

PCI-E 3.0 + optional NVLink 2.0 (2 bricks)

PCI-E 3.0 + optional NVLink 2.0 (1 brick)

PCI-E 3.0

Theoretical transfer bandwidth (bidirectional)

100 GB/s NVLink
32GB/s PCI-E x16 3.0

50 GB/s NVLink
32GB/s PCI-E x16 3.0

32GB/s PCI-E x16 3.0

Achievable transfer bandwidth

TBC NVLink, ~12 GB/s PCI-E x16 3.0

~12 GB/s PCI-E x16 3.0

 

%d bloggers like this: