NVIDIA has officially registered the name Turing and the brands GeForce RTX, and Quadro RTX, a declaration of intent directly related to the next generation of graphics cards and accompanied by a new specification filtration.
From the very beginning, we start to provide an appropriate contextualization of this new information to better understand it. NVIDIA has always been committed to developing proprietary technologies to enhance the graphics of games in a special way, a reality that we can now perfectly represent in GameWorks.
Since the beginning of the year, the green giant has publicly demonstrated its interest in ray tracing for video games, a technology called NVIDIA RTX, which will be included in the PC version of Metro Exodus, one of the most anticipated next-generation games.
In previous articles, we told you that the next generation of NVIDIA graphics cards will be able to maintain Volta’s core sensors and dedicate them to a ray tracing system. This would significantly improve performance over the Pascal-based GeForce GTX 10 while giving the company the ability to run raytracing.
A way to justify the generation change
Let’s face it, developments focused on PS4 and Xbox One have put a heavy strain on the PC gaming industry. Compatibility adjustments are associated with minor improvements and do not really exploit the potential of the next generation hardware, a situation that has extended the update cycles to such an extent that a high-end graphics card can still be played in 1080p from 2012 to 2018.
Against this background, we can understand why NVIDIA wants to introduce concrete advances that demonstrate the superiority of the PC and justify a generation change. At the moment, anyone with a GTX 970, GTX 1060 3GB or higher has no compelling reasons to update their graphics card, neither are they going to be with the advent of the next generation of graphics cards, as console-centric developments and hardware-level requirements are maintained.
With the emergence of raytracing and its introduction as a “new” technology unique to GeForce RTX, it could be the revival NVIDIA needs to “sell” its new generation of graphics and convince users that they need to update their graphics card, which would be a particularly good approach for players who have a larger budget and always want ” state of the art “.
The average user would be less interested in raytracing and will probably choose to keep their current graphics card and wait directly for the arrival of Xbox Two and PS5, two next-generation consoles that will lead the way in game development and determine their requirements.
GeForce RTX – Underperforming
Those of you who read us every day know that Turing is the name of the architecture that will use the new generation GeForce and that it is a simple update of the Volta architecture. The latest information reinforces this idea and leaves us with some unfavorable data, as this new generation would be far below expectations in terms of gross profit.
For better understanding, we provide you with a summary of the comparison performance we expect and filtered in the latest information:
- TITAN RTX: We expected it to be 30% more powerful than the GTX TITAN V, but it will only exceed it by 15%. It will cost 50% more than the GTX 1080 IT and about $3,000.
- GeForce RTX 2080: Based on the TU104 kernel, we expected up to 40% more performance than the GTX 1080 TI, but it will only exceed it by 8%. It will be 50% more powerful than the GTX 1080 and cost between $500 and $700.
- GeForce RTX 2070: based on the TU106 kernel, we thought it would run at the level of a GTX 1080 IT, but it will be below. It will be 17% better than a GTX 1080 and 40% better than a GTX 1070. It’s between $300 and $500.
- GeForce GTX 2060: based on the TU116 kernel. We expected a similar performance as a GTX 1080, but it will be 7% slower than a GTX 1070 and 27% more powerful than a 6GB GTX 1060. It’s between $200 and $300.
GeForce GTX 2050: based on the TU118, we thought it would work like a 6GB GTX 1060, but it will be 17% slower than that. It will outperform the GTX 1050 IT by 50% and cost about $150-200.
Please note that GeForce GTX 2000 models (theoretically) have no pulling cores and are not optimally suited for working with NVIDIAs ray tracing. The attached table also shows the date of availability of each model and its assumed graphics memory configurations.
Conclusion: Which one is the best?
This information has had a great impact by changing the nomenclature, but the truth is that it should not surprise us. The issue of beam tracking and train cores is not new, and their arrival on the PC at the level of general consumption has not taken place overnight either.
In general, everything we said makes sense. The introduction of a part of the next generation graphics system, such as GeForce RTX, distinguishes it more clearly from the current generation, with which it will theoretically coexist indefinitely.
On the other hand, the smaller difference in gross performance between the GeForce RTX-GTX 20 and GeForce GTX 10, which saves performance in ray tracing, will help maintain interest in Pascal and facilitate inventory streamlining in the months following the introduction of the new generation of NVIDIA.
We’re looking for the first performance tests so you know what the new Turing-based GeForce has to offer. Before I finish, here’s a summary of how the GeForce GTX 900 would look like compared to the GTX 20 series:
- GeForce GTX 980 TI: It would be a little more than a GTX 2060.
- GeForce GTX 980: It surpasses a GTX 2050.
- GeForce GTX 970: reaches approximately the level of a GTX 2050.
- GeForce GTX 960: They are well below a GTX 2050.