If there is a type of application that requires increasingly powerful hardware and that manufacturers do not stay behind, it is games. Even the most powerful graphics card is not enough to move the games that will be released in the medium term with sufficient image quality and speed. That is why future GPU from NVIDIA Y amd exceed the 90 TFLOPS of power should not surprise us.
There is no doubt that there is still much to do and that we have not reached the visual limit. The next challenge? Being able to move the most advanced games to a 4K resolution, at a frame rate of more than 60 and with Ray Tracing activated. Believe us that this requires a very high power. How many? Well, as much as becoming almost ten times larger than SONY’s PS5. Of course, we will see the impressive figure of more than 90 TFLOPS with the RTX 4090 and the RX 7970 XT, or whatever they are called in the end.
We must clarify that the TFLOPS only serve as a comparative measure within the same architecture. And it is that each instruction can be composed of different basic operations and differ in execution time from one architecture to another. In the world of graphics cards, its importance has been growing over time, especially due to the fact that there are more and more elements that depend less on fixed function units and more on the processing capacity of the different GPU cores. or shader units.
NVIDIA and AMD will exceed 90 TFLOPS of power in their future GPUs
In the last few days we have received information about the ability to do 32-bit floating point calculations of the next top-of-the-line GPUs from NVIDIA and AMD. Of course, it is a figure that refers to leaving both processors performing the FMA instruction, which is made up of 2 operations per cycle. Which is the fastest to run on this type of chip.
In the first place, and according to rumours, the RTX 4090 will be able to reach 100 TFLOPS of power. Which translates to 2.5 times the power of the RTX 3090 Ti at 40. And while it’s true that NVIDIA’s next top graphics card will consume around 600W, its power per watt will be much higher. With this we can deduce the approximate clock speed that the top-of-the-range GPU of the RTX 40 would have. Around 2.7 GHz. Although at the moment we do not know if it is a Boost speed and therefore temporary or a more stable one in time.
In the case of amd and your future RX7900XT, that we do not know its consumption. There are rumors that it will be smaller than its NVIDIA rivals. What does our intuition tell us? Those of Lisa Su are not going to want to lose even in a single benchmark. So we are very clear that we are going to see at least one version of its RX 7000 using the 600 W connector. The RX 7900 XT is very likely to hit 92 TFLOPS. All this thanks to its 3 GHz clock speed.