Home » NVIDIA RTX 40 graphics cards will break speed records
Technology

NVIDIA RTX 40 graphics cards will break speed records

That NVIDIA is increasing the speed of its graphics cards is a fact that almost everyone takes for granted. Although there is no official statement on this yet, the rumors continue to run and the latest leaves a two-way panorama for the same fact: the RTX 40 will have a much higher frequency than the 2.8 GHz previously namedbreaking the speed record.

Well yeah, it’s hard to believe, but apparently NVIDIA is going to be tightening the nuts more than expected. If everything is true and as we will see, the leaker is trustworthy and everything suggests that yes, Huang’s men are not going to laugh and can surprise in terms of performance.

NVIDIA RTX 40: the GPU that will break the serial speed record

The accounts have already come out with the figures provided at the time by this same leaker, as seen in another article. And although we will recalculate these figures, it will be necessary to redo them if, indeed, and as they say, the frequencies of the RTX 40 will be not a little, but much higher. But before we continue, let’s recap with this data:

The AD102-400 chip that the RTX 4090 would carry would have 16,384 Shaders at a frequency of 3 GHz, which would leave us -> 16,384 x 2 x 3 GHz -> 98,304 TFLOPS. Instead, the full AD102-300 chip of the RTX 4090 Ti or the alleged TITAN RTX 40 would have -> 18,432×2×2.8GHz -> 103,219 TFLOPS.

Of course, let’s remember that TFLOPS are not a reliable measure of display performance, because, for example, they don’t take bandwidth or GPU bus into account, and since VRAM is the main bottleneck of display performance. throttling, the differences are usually quite large. compared to that shown by TFLOPS.

Similarly, if the frequencies are much higher than those quoted in 2.8GHz and 3GHzWhat difference are we talking about specifically? It’s really complicated, but it looks like the RTX 40 will break the existing stock speed record.

TSMC’s 4N Node for NVIDIA and the Frequency Controversy

RTX-4090-Shaders

By parts and slowly everything becomes more complicated. First of all, we know very little about the TSMC exclusive node for the green team. There is no specific data from the company indicating the differences between N5 and this 4N, so we must take a second point for granted, and that is consumption.

But yes 800 watt, 600 watt, 450 watt… There is only one constant in the face of such a hubbub: the graphics cards will consume more and this can be key and the result of much higher frequencies, as confirmed by the leaker. If 3 GHz was already a huge leap forward and everyone was “happy” with this justification, which is to be expected, if what is written is true, these are figures close to 3.2GHz.

That said, now comes the reasons for said increase and rectification by the leakers and obviously by NVIDIA. this seems to be Navi 31 from AMD, and yes, we are aware that it was first said that it might not be as competitive as expected, but the rumors and the winds now are that it would be really very close to NVIDIA with better consumption and better efficiency.

Therefore, the only trick that would have left Huang is to push to the limit to go over the top at the cost of higher consumption, since it seems that in IPC AMD will again be in the lead in this section, although this unfortunately doesn’t translate to more FPS than its rival, which ultimately coupled with the price of the GPUs, is what matters to end users.

About the author

admin

Add Comment

Click here to post a comment

Your email address will not be published. Required fields are marked *