Of the technology and specifications of the new NVIDIA GeForce RTX based on the 12-nanometer manufacturing process and equipped with memory GDDR6 we have already spoken after the presentation of Cologne, pointing the spotlight on news advertised to the full extent by the NVIDIA itself as the hardware Ray tracing, capable of generating physically accurate reflections and global illumination, and a new extremely effective supersampling, guaranteed by the Tensor Core inherited from the Volta architecture. But during the presentation segment reserved for operators we have obtained a lot of other information on the possibilities of the new GeForce that we know are in release on the market on 20 September in the versions RTX 2080 and RTX 2080 Ti , with 2070 coming instead in October. Combining these new details with those already known we try to outline a complete profile of the new NVIDIA Turing GPUs.

    
        

High prices but concrete innovations in the GPU field

    
    

The revolution of Ray tracing has been in preparation for a long time and it is likely that AMD is working on it too, but NVIDIA has played ahead of time and, strong of an already dominant position, has imposed its prices asking 650 euros for the NVIDIA RTX 2070, 869 euros for the NVIDIA RTX 2080 and a good 1279 euros for the NVIDIA RTX 2080 Ti. Figures important in an absolute sense, but that have caused some bad mood more than expected given the emergence of a leap lower, in terms of brute power, compared to that made by the previous series. In fact, we are talking about 35 percentage points that do not depend only on the CUDA Core or the new GDDR6 memories at 14Gbps, but also from the highest boost of the Founders editions, this time superior to the design reference. But technology transcends the technical specifications that are bare and crude, both from a hardware and a software point of view. In fact, there are also 12-bit HDR 8K with dedicated HEVC decoder, optimized audio, Virtual Link for virtual reality and NVLink the new interconnect technology that we are really curious to discover where it will take us the promise to overturn the bad situation of the SLI support even if the exclusion of the RTX 2070, with the offer therefore bound only to the top models, has caused several alarm bells to sound. But it could be a simple question of prices as the NVLink is used in supercomputers and enjoys professional support that, translated into video games, could bring back in vogue multiGPU configurations capable of handling a surround of 4K screens . 144Hz screens obviously, but we humans are forced to put aside because of the currently too high prices. But then we already know that it will be difficult to exploit them with the new GPU unless you give up the RTX which, summing up, is a new technology dedicated to Ray Tracing in real time and guaranteed by the new RT Core that reach a maximum of 72 for the RTX 2080 Ti.

        
            

        
             005 "title =" 005 "class =" img-responsive "/> </a><br />
        <br />
        <br />
    </div>
</figure>
<p> Not many, it seems, but able to drastically raise the bar of Ray tracing in real time, for a global illumination that can now count on a myriad of refractions and tons of reflections thanks to the passage from the 1.2 billion light rays manageable by the GTX 1080 Ti to the 6, 8 and 10 billion of the three new cards. The most powerful of the three is obviously the RTX 2080 Ti, an extreme solution that can count on 576 Tensor Core for 110TFLOP half precision dedicated to deep learning and on 4608 CUDA Core, against the 3584 of the GTX 1080 Ti, for a power of <strong> 14TFLOP, </strong> against 11.3 of the flagship of the previous series. And all this, combined, ends up under the heading RTX-OPS, a resonant unit of measurement designed by NVIDIA to enhance and make understandable the capabilities of the new GeForce. According to the NVIDIA slides, the RTX 2070 is content with 45T RTX-OPS, the RTX 2080 goes to 60T RTX-OPS and the GeForce RTX 2080 Ti reaches 78T RTX-OPS, taking the field, at least from the specific point of view of Ray Tracing, a power of <strong> six times higher </strong> than previous GPUs, also because of a new and more effective algorithm for the management of BVH. All accompanied by another technology that promises, with RTX deactivated, given the weight of the latter on performance, to increase performance to almost double that of the use of Temporal Anti Aliasing. The novelty in question, called <strong> Deep Learning Super Sampling </strong> (DLSS), is based on the Tensor Core which, implemented in the Volta GPUs to enhance their capabilities in the field of Deep Learning, allow the new GeForce to literally create pixels , producing an image with a comparable overall quality but with clearer contours than existing technologies. Not bad, therefore, that enjoy, as the RTX, a good support, at least on paper, with fifteen compatible games and full support from Unreal Engine and Unity.</p>
<figure class=
        
            

        
             Turingvspascal Editorsday Aug22 2 "title =" Turingvspascal Editorsday Aug22 2 "class =" img-responsive "/> </a><br />
        <br />
        <br />
    </div>
</figure>
</div>
<div id=     
        

Beyond the cores, silicon and frequencies

    
    

The new GeForce RTX are based on a production process destined to become obsolete rather soon, with the 12 nanometers that will soon be outclassed by 7 of the AMD chips. Precisely for this there are those who hypothesize the arrival in 2019 of a refresh at 7 nanometers and the immediate launch of the RTX 2080 Ti could be a clue. But this step allows the top 10 series boards, which will remain on the market for some time, to maintain a discrete value and moreover, as anticipated, the value of the new NVIDIA GPUs transcends the progress of the production process and the increase in CUDA Core. The new GeForce are the springboard for technologies most likely destined to change the market of graphics in real time and this also in terms of gameplay with the reflection of elements external to the visual that can revolutionize the world of shooters. Moreover, the usefulness of artificial intelligence is not limited to the possibility of improving definition, rendering of slow-motion and overall image quality. The Tensor Core, made easily accessible through the creation of an NGX Framework, are able to autonomously simulate the performance of the materials, can be used to improve the intelligence of the opponents, are useful for detecting cheats, increase the ability of the hardware to interpret voice commands and can even help in the field of character expressiveness. It is therefore not difficult to imagine alternative uses to the DLSS and this only magnifies the importance of the Tensor Core less publicized but perhaps as important as the RT cores to determine the nature of the GeForce RTX. And the potential of the new NVIDIA GPUs goes further thanks to the development of variable quality shading that allows to concentrate the computing power on the relevant elements of the image, while at the same time saving on the less visible ones like the peripheral areas of the visual in VR games. like the sides of the road in racing or like the dark areas of a poorly lit environment.

Create Account



Log In Your Account