Nvidia unveiled its new Turing structure and RTX 2080 graphics playing cards final month with an enormous promise of “six occasions extra efficiency” than the earlier era, and the brand new capability to help real-time ray tracing in fashionable video games. We’ve seen efficiency claims, some particulars in regards to the Turing technical structure, and lots of PC avid gamers making an attempt to calculate how these new playing cards will really carry out, however I sat down with Nvidia final month to get an in-depth take a look at the brand new Turing structure that’s powering this subsequent era of playing cards. It’s truthful to say that there’s rather a lot occurring.

The large new characteristic of Nvidia’s Turing structure is the power to help real-time ray tracing in video games. Ray tracing is a rendering method utilized by film studios to generate mild reflections and cinematic results. Vehicles, launched again in 2006, was the primary extensively ray-traced film, and lots of studios now use ray tracing in fashionable movies. Even rendering instruments like Adobe AfterEffects, Maya, and 3ds max all help some type of ray tracing, and it’s a way that’s thought-about the “holy grail” for video video games.

Nvidia has spent 10 years engaged on ray tracing for its Turing structure, and it’s clear from the varied demos that this isn’t simply advertising jargon. Nvidia is beginning to introduce ray tracing with the RTX 2080, and a few of the early demonstrations have been spectacular however restricted to sure video games. Nvidia’s Battlefield V demo confirmed off some real-world examples of how the sport shall be rendered extra movie-like, with reflections from explosions on close by vehicles and flames that may even mirror off of the metallic on weapons and even diffuse into the picket elements of older weapons. It’s spectacular stuff once you see it up shut in a selected a part of a recreation or evaluate it aspect by aspect with a scene that isn’t ray-traced.

Turing sees Nvidia introduce new Tensor cores and RT cores. The RT cores are devoted to ray tracing, whereas the tensor cores shall be used for Nvidia’s AI work in video games. Turing additionally consists of some large adjustments to the way in which GPU caches work. Nvidia has moved to a unified reminiscence construction with bigger unified L1 caches and double the quantity of L2 cache. It’s basically a rewrite of the reminiscence structure, and the corporate claims the result’s 50 p.c efficiency enchancment per core. Nvidia has additionally moved to GDDR6 with its RTX 2080, which suggests there’s as a lot as 50 p.c increased efficient bandwidth for video games.

All of those large platform adjustments make it greater than a bit of tough to find out how vital any efficiency positive factors shall be for normal PC avid gamers. “It is a new computing mannequin, so there’s a brand new approach to consider efficiency,” stated Nvidia CEO Jensen Huang on the firm’s unveiling occasion in Germany final month. That new approach to consider efficiency underlines lots of the adjustments that Nvidia is making with RTX, in an period the place Moore’s legislation is over. It’s not simply how a lot quicker a CPU or GPU is anymore that issues, nor the variety of cores it has. It’s now the large architectural adjustments that can make a distinction for each picture high quality and efficiency for years to return. Nvidia, Intel, AMD, Samsung, Apple, and lots of others will more and more must do extra with the present transistors on chips as an alternative of constant to shrink their dimension. Nvidia has clearly realized this inevitability, and it’s time for a change of tempo.

Nvidia is offloading a few of the work that may usually be dealt with by the corporate’s CUDA cores to those separate Tensor and RT cores. Nvidia can be introducing new rendering and texture shading methods that can enable recreation builders to keep away from having to repeatedly shade each a part of a scene when the participant strikes round. This can assist enhance efficiency, however Nvidia goes a step additional with the assistance of supercomputers.


Nvidia Deep Studying Tremendous-Sampling (DLSS) might be crucial a part of the corporate’s efficiency enhancements. DLSS is a technique that makes use of Nvidia’s supercomputers and a game-scanning impartial community to work out essentially the most environment friendly approach to carry out AI-powered antialiasing. The supercomputers will work this out utilizing early entry copies of the sport, with these directions then utilized by Nvidia’s GPUs. Consider it just like the supercomputer understanding one of the simplest ways to render graphics, then passing that hard-won information onto customers’ PCs. It’s a fancy course of, however the finish outcome must be improved picture high quality and efficiency whether or not you’re taking part in on-line or offline.

The actual take a look at for DLSS shall be how nicely it improves efficiency for 4K video games. Whereas the GeForce GTX 1080 Ti struggled to hit 60 fps in 4K for contemporary video games, Nvidia is promising some spectacular efficiency positive factors at 4K with DLSS enabled on the RTX 2080. We solely lately obtained playing cards for testing, so we’ll be checking to see simply how nicely a few of these DLSS video games carry out within the coming weeks.


Nvidia is basically overhauling its whole structure right here, and the way in which it thinks about graphics playing cards with Turing and RTX. Prior to now, Nvidia could be completely satisfied to promote you a graphics card as a result of it’s quicker, extra highly effective, and has extra reminiscence. That also holds true immediately, however the dialog is shifting towards new capabilities and optimizations for graphics playing cards over the uncooked specs. Whether or not it’s ray tracing, DLSS, or Nvidia’s many different adjustments and enhancements, you’re shopping for into this complicated new structure with a promise of what’s to return.

Solely 11 video games help full ray tracing proper now, and 21 video games in complete (for now) will implement both ray tracing, DLSS, or different RTX platform enhancements. Some will clearly implement all three, offering the most effective efficiency boosts. Nvidia can be working with Microsoft to convey a lot of its RTX parts to DirectX, however its essential rival AMD may even want to assist push this new period of ray tracing. AMD provides the graphics playing cards for each the PlayStation four and Xbox One, and a brand new era of consoles is anticipated in 2020. The {hardware} inside them will actually form the upcoming way forward for video games as builders more and more look to make sure titles can be found throughout consoles, PCs, and even cell gadgets. If the following consoles can’t help real-time ray tracing, then it’s unlikely to be broadly adopted throughout video games within the close to future.

That’s not going to cease Nvidia, although. The corporate is taking an enormous wager on the way forward for the graphics card. As we enter an period the place uncooked chip speeds don’t inform the entire story, it will likely be the large overhauls of structure and intelligent use of {hardware} and software program that can push the video games business ahead. Nvidia has made its gamble, and the cube are actually rolling.

Supply hyperlink

LEAVE A REPLY

Please enter your comment!
Please enter your name here