Skip to main content

Nvidia’s GPUs could smash AMD in the next-gen performance war

Nvidia and AMD are each set to release next-gen graphics cards this year, and according to the latest leaks, we could witness quite a performance war between the two giants.

New rumors imply that Nvidia’s next flagship, the RTX 4090, could be a powerful GPU with 100 teraflops. However, the news for AMD’s flagship Navi 31 graphics card is not quite as good.

100t for ada

— Greymon55 (@greymon55) April 30, 2022

This latest batch of Nvidia RTX 4000 and AMD RDNA 3 information comes from numerous Twitter leakers, and it’s safe to say that it’s been a bit of a rollercoaster ride. While it’s good news for Nvidia through and through, AMD’s next flagship seems to have received a cut compared to the specifications we expected to see just a few days ago.

Let’s start with Nvidia. We’ve already gathered a lot of information as to the approximate specifications of the card. The current consensus is that it will have monstrous power requirements, but that power draw is likely to come with outstanding performance to match.

The flagship RTX 4090 is going to be based on the AD102 GPU, and according to the latest round of leaks shared by VideoCardz, it might deliver up to 100 teraflops (TFLOPs) of computing power. This is 2.5 times more than the current best graphics card, the RTX 3090 Ti, which comes with 40 TFLOPs.

To hit such numbers, the rumored RTX 4090 would have to have a staggering clock speed of at least 2.7GHz. It’s also said to come with 18,432 CUDA cores. This is, once again, a huge increase over the RTX 3090 Ti (10,752 CUDA cores), but the clock speed is rather close (2.6GHz vs 2.7GHz).

Seeing as both Team Green and Team Red are releasing new graphics cards this year, it’s hard not to compare Nvidia’s alleged flagship to that of AMD. In AMD’s case, the RX 7900 XT is going to be a Navi 31-based graphics card. Although we’re only a few months away from the release, the card has recently seen a huge shift in specifications, although all of this is still just in speculation territory.

Navi 31: 12288 SPs, 48 WGPs, 12 SAs, 6 SEs
Navi 32: 8192 SPs, 32 WGPs, 8 SAs, 4 SEs
Navi 33: 4096 SPs, 16 WGPs, 4 SAs, 2 SEs

— Redfire (@Redfire75369) May 2, 2022

Just a few days ago, AMD leaker Greymon55 reported that the new AMD RDNA 3 flagship will feature 92 TFLOPs of FP32 performance, putting it within reach of the specs we now expect from the RTX 4090. If proven true, this would mean we’re going to witness a real TFLOP war, which should be exciting for potential customers, with more top-notch GPUs to choose from. However, the news changed almost as soon as it appeared, and a new set of specs has been shared for the AMD RX 7900 XT.

As per the new specs, once again shared by Greymon55 and Redfire75369, the Navi 31 GPU will arrive with 12,288 stream processors (SPs) as opposed to the 15,360 they previously predicted. This implies that the 92 TFLOP rumor could be mistaken or it could be referring to a different, not cut-down version of the GPU. As VideoCardz guesses, the updated specs would give us just 73 TFLOPs instead of 92, putting the RX 7900 XT miles behind Nvidia in that regard.

With months to go until the release of either the Nvidia Ada Lovelace or AMD RDNA 3 graphics cards, it’s hard to say anything with real certainty. Based on the numbers we have now, it seems that Nvidia’s flagship may win in terms of sheer performance over AMD, but when it comes to the balance of performance and power draw, things could be looking up for AMD. Until either tech giant says more, all we can do is speculate.

Editors' Recommendations