According to popular rumour monger and reliable leakster, kopite7kimi (@kopite7kimi), they have suggested that the RTX 4090 will be more than 66% faster than the current Ampere king-of-the-hill, the RTX 3090. As for the basis of this claim, they say that this is based on UL’s 3DMark benchmark and more specifically, the Time Spy Extreme test. Specifically, Kopite says that the GPU allegedly gained a score of more than 19000 points in the test. For context, our RTX 3090 FE has, on paper, an average score of 9200 points out of the box on the same test, and the most we managed to push it to was 9745 points overclocked. By comparison, the alleged score produced by the RTX 4090 is more than double that number, a point that really beggars belief and pose the question of just how much NVIDIA has juiced up Lovelace. On that note, it also makes one think about just how much more power the Ti version of the card will be churning out. That being said, this is just an alleged score lifted from a synthetic benchmark, and isn’t necessarily a good point of reference. After all, AMD’s own Radeon RX 6800XT and 6900XT graphics cards were shown to be able to keep up with their respective counterparts in the same benchmarks.
— kopite7kimi (@kopite7kimi) July 18, 2022 Beyond that, Kopite’s tweet is devoid of any further details of the RTX 4090. But again, initial reports have already suggested that the GPU could feature up to 16384 CUDA cores, 24GB GDDR6X graphics memory running at a frequency of 21Gbps, a boost clock going beyond 2750MHz, and a TDP of 450W. In any case, we’ll just have to wait until NVIDIA’s official announcement to learn more. (Source: Videocardz)
