NVIDIA RTX 40 series SUPER specifications revealed

Previous reports indicated that NVIDIA is gearing up to launch three new GeForce RTX 40 series SUPER models based on the Ada Lovelace architecture: namely, the RTX 4080 SUPER, RTX 4070 Ti SUPER, and RTX 4070 SUPER. The PCI ID of the RTX 4080 SUPER has already surfaced, revealing it to be equipped with the AD103 chip.

Recent revelations from enthusiasts suggest that the RTX 4080 SUPER specifically incorporates the AD103-400, boasting 10,240 CUDA cores. This maneuver by NVIDIA seems to be anticipated, as some discerning readers had already speculated the CUDA core count of the RTX 4080 SUPER in comments on prior news pieces.

Additionally, there’s been further buzz surrounding the less frequently mentioned RTX 4070 Ti SUPER and RTX 4070 SUPER. Presently, the most perplexing element seems to be the nomenclature of the RTX 4070 Ti SUPER, as the combination of “Ti” and “SUPER” in its title is somewhat enigmatic. There’s speculation that the final product name might undergo modifications.

It’s purported that the RTX 4070 Ti SUPER might feature either the AD103-275 or AD102-175, with an L2 cache of 48MB and 8,448 CUDA cores. This marks a 10% increase from the existing RTX 4070 Ti’s 7,680 CUDA cores. The RTX 4070 SUPER is anticipated to harness either the AD104-350 or AD103-175, elevating its L2 cache from the current 36MB of the RTX 4070 to 48MB, and CUDA core count from 5,888 to 7,168—a rise of approximately 21.7%.

Rumors suggest that NVIDIA might employ a mix of chips for the RTX 40 series SUPER. According to current discourse, the RTX 4070 Ti SUPER might utilize both AD102 and AD103, while the RTX 4070 SUPER could incorporate both AD103 and AD104. Given NVIDIA’s historical practices, such an approach would hardly be considered unorthodox.