The graphics card manufacturer Nvidia introduced yesterday their upcoming range of new generation RTX 4000 series cards, starting with the top performance RTX 4090 due for release on next October 12th. It will have 24 Gb of memory, announced at USD 1599. In late November, there will be two RTX 3080.. one with 16 Gb memory (USD 1199) and the other with 12 Gb memory (USD 899), and also less cudacores (less performance) but without difference in the models name (no Ti suffix here to distinguish beween those 2).
If simFlight is, for once, interested in sharing these hardware news, it’s because the Nvidia event specifically targeted Microsoft Flight Simulator among other PC video games to demonstrate the significant boost to expect in these games thanks to their new technology.
Before we explain the video embedded here, we have to insert first a quick reminder about the technical feature exclusive to Nvidia : DLSS. This is a computation generated by the processur of the graphics card, GPU, to produce a higher number of framerates in high resolution, but without actually processing each image at high resolution. These frames are calculated with a reduced resolution, and then, they are upscaled to display high definition animations to the gamer.
In difference with DLSS 1.0 and 2.0 where the graphics processor was adding pixels where they were missing in the HD frame. the new DLSS 3.0, exclusively introduced by the upcoming RTX 4000, enables to add more HD frames between 2 rendered frames. This new process explains easily why we can expect much higher framerates with such technology.
This means that DLSS 3.0 works only with the GPU and doesn’t add more work to the main CPU. That’s particularly efficient for games limited by CPU, and Flight Simulator is one of them. And this also explains why the boost may be as strong as the preview video suggests : because you can get a wave of more frames provided by your powerful GPU without need to make work the CPU, which would limit the positive effect of DLSS if the CPU was required in its process.
Basically, future owners of a RTX 4000 graphics card, instead of “just” getting upscaled pictures with enhanced resolution and the missing pixels provided by the GPU, the new cards will send more upscaled frames to the screen. And, their technology also uses “Deep Learning”, which consists of computing the intermediate frames by guessing what it should display with high accuracy according to Nvidia. In the end, the upcaled images are more frequent, with enhanced visuals.
AMD also has a similar technology, FidelityFX, engineered slighter after Nvidia, and it works well too. But here, Nvidia is showing that, on selected games, the boost of performance is really significant according to their internal tests and communication. Further tests will be available among the specialist medias in PC gaming and hardware when the cards will be available, so they can confirm this expectation, hopefully.