May 27, 2019
David Cardinal

Nvidia Brings RTX to Mobile With New Quadro GPUs

While most mobile users are looking for the lightest laptop they can get away with, users with heavy compute requirements for applications including scientific work, modeling, media processing, and AI are keen to get as much performance as they can on the road. With this week’s announcement at Computex, Nvidia is giving them access to its Turing architecture GPUs and RTX ray tracing and AI cores.

Nvidia’s New Quadro GPUs By the Numbers

There are six new Quadro mobile GPUs. Three are based on Turing with RTX: the 5000, 4000, and 3000. Three are Turing but without the RTXSEEAMAZON_ET_135 See Amazon ET commerce cores — The T2000 and T1000. Whether you want to fork out the extra money for RTX will have a lot to do with whether your favorite applications are planning to take advantage of them for ray tracing or AI inferencing. The two low-end models, the P620 and P520, are updates to the similarly numbered current Pascal architecture models, the P600 and P500.

The 5000 will supports up to 16GB of GDDR6, like the current P5200, making it more capable than many desktop GPUs. Those hoping to get performance and battery life will be happy to know it only draws up to 110 watts, compared with the 150 watts of the P5200 — despite having 500 more CUDA cores and adding 384 tensor cores and 48 RT cores. Most of the other new models offer slightly smaller spec upgrades than the flagship and come with similar power requirements to the current versions.

Nvidia Quadro RTX mobile GPU feature comparision chart

Nvidia Quadro RTX mobile GPU feature comparison chart.

Nvidia Claims Some Impressive Performance Gains

During its press briefing for the new GPUs, Nvidia showed some compelling comparisons with its previous generation and with the Vega 20 in a MacBook Pro. For example, the new Enhanced Detail capability in Lightroom should run 4x as fast as on a Core i7 or a Vega 20. Similarly, ray-traced rendering in Maya should be nearly 10x as fast. Nvidia also showed how RTX’s RT cores brought photorealistic architectural renderings to life, filling in lifelike shadows and rendering deeper colors in Enscape 3D compared with the same scene rendered without RTX. Use of RT cores should also be more power efficient than similar functions run on the more general-purpose CUDA cores.

Nvidia has also been working closely with high-resolution video camera maker RED on real-time mobile workflows for 6K and even full-featured workflows for 8K video, utilizing Adobe’s Premiere Pro 2019 and DaVinci Resolve 16 in addition to RED’s own tools. Even though most content distribution is currently limited to 4K or 1080p, many videographers shoot in 6K or 8K to give them some post-processing flexibility to pan or zoom — as well as looking ahead to when higher-resolution content becomes more widely distributed.

Even 1080p and 4K content is often shot in 6K or 8K to allow for flexibility in post-processing, includign panning and zooming

Even 1080p and 4K content is often shot in 6K or 8K to allow for flexibility in post-processing, including panning and zooming.

RTX Adoption Continues to Grow

While Nvidia’s innovative RTX capabilities have taken a while to get momentum, the company continues to add to the list of applications that will make use of them. For example, Solidworks will be one of the first to use Nvidia’s RTX-enabled AI-based de-noising for ray tracing. So far it is only a demo, but the companies have said it will be in Solidworks later this year. Nvidia claims that the new feature running on its RT cores will provide as much as a 10x speedup in generating low-noise renderings.

We expect the first mobile workstations based on the new GPUs to be introduced as soon as this week, with announcements continuing during the rest of the quarter, probably from some of the typical vendors including HP, MSI, Dell, and Lenovo. We’ll have to wait until the first workstations ship to get a true sense of performance improvements, as with laptops thermal throttling can quickly become the biggest limitation on graphics and compute performance.

Now Read: