For those who don’t know, the RTX 4090 Ti is NVIDIA’s next BFGPU and the flagship from the RTX 40 family. Currently, we know that it’s supposed to draw up to 600W of power and will offer a significant performance improvement over its predecessor; some cases put that number at 2x indicating that the RTX 4090 (Ti) will be twice as fast. Now, Twitter user AGF (@XpeaGPU) has shared benchmarks for the RTX 4090 (Ti), testing the card in the game Control. AGF played Control at 4K with Ray-Tracing and DLSS on—further details on the last two were not provided. Regardless, they claim the card managed to deliver over 160FPS on Ultra settings. If this is indeed true, we could finally be looking at 100FPS+ AAA 4K gaming.
— AGF (@XpeaGPU) July 19, 2022 That’s an unprecedented number given the current landscape of GPUs. The RTX 3090 only gets around 72FPS on average with largely the same settings, meaning that this alleged RTX 4090 (Ti) is producing 2.2x the performance of the RTX 3090. If we dive into specifics, the RTX 3090 delivers that frame count on the High preset, so if we actually use the Ultra Preset, the difference jumps to 2.5x between the 3090 and 4090 (Ti). You don’t need me to tell you that’s an insane number. More than double the amount of performance in gaming would mean that the RTX 4090 (Ti) is one of NVIDIA’s largest generational gains ever. Previously, it was expected that the 4090 (Ti) would produce these kind of results in synthetic benchmarks, but if gaming’s a similar story, gamers looking forward to the series are about to have a party. Infamous leaker, @kopite7kimi even mentions that aftermarket variants of the RTX 4090 (Ti) can break the 20,000 points barrier in 3DMark Time Spy Extreme, which is also a 2x improvement over current-gen, and about a 70% gain over the RTX 3090 Ti. Not only that, but kopite also says that the RTX 4090’s maximum power limit can go up to 800W in the most extreme cases, despite having a tiny PCB but monstrous cooling.
— kopite7kimi (@kopite7kimi) July 19, 2022 But before you get all excited, AGF mentions this GPU as AD102 on “high power draw“. By now, it’s confirmed that AD102 GPU will be used inside the RTX 4090, so the high power draw could allude to a potential factory overclock or boost. Most AIB variants, in fact, come with higher TBPs and better cooling to push the card further, as stated by Kopite above. However, the leaker clearly mentions that they tested the “Full AD102” which means this is likely the SKU above the RTX 4090, as the 4090 itself is supposed to use a cut-down version of AD102. The full AD102 die features 18,432 CUDA Cores, and the cut-down version features only 16128, which definitely will affect the performance. So, unless things have changed behind the scenes, this 160FPS benchmark won’t be representative of the retail RTX 4090, rather that the 4090 Ti or a TITAN-class Ada Lovelace GPU. Still, considering the fact that the 3090 is only about 20% slower than the 3090 Ti in gaming, we could expect the RTX 4090 to produce around 120FPS in demanding games at 4K resolution with maxed-out settings.
