Upcoming additions to NVIDIA's GeForce RTX 40 series, the Super variants, have had their specifications outlined, including also, a 16GB model of the GeForce RTX 4070 Ti.
Reports suggest that the new lineup will be finalized by January 2024. According to information from a Videocardz report, which matches other sources, NVIDIA has provided key details to its AIC partners for the initiation of case designs. The final product names are yet to be confirmed, but chip codenames and CUDA Core counts, as stated by Kopite7kimi, are consistent.
The GeForce RTX 4080 Super is expected to feature 16GB of GDDR6X memory and will be based on the AD103-400 chip with 10,240 CUDA Cores. The TGP (Total Graphics Power) is anticipated to remain at 320W, despite the increased core count.
The AD103-275 chip is set to be used in the GeForce RTX 4070 Ti Super, which will have 8,448 CUDA Cores and a 16GB GDDR6X memory configuration, with a TGP of 285W, consistent with the non-Super RTX 4070 Ti model.
An increase in TGP to 220W is expected for the GeForce RTX 4070 Super, which will necessitate a move to the 12VHPWR connector from the PCIe 8 PIN connector. This graphics card will use the AD104-350 chip with 7,168 CUDA Cores and 12GB GDDR6X memory.
Performance estimates for the AD102 chip indicate a 4800 TPP (Total Processing Performance). However, due to US export restrictions, the GeForce RTX 4090 will be excluded from the Chinese market after November 16th, US time. This has led to the suggestion that the 20GB GeForce RTX 4080, potentially using an AD102 chip, will not continue in production.
The three GeForce RTX 40 Super series graphics cards are expected to be released in January 2024, with pricing details to be announced nearer to their release.
