The source published rather detailed characteristics of the GeForce RTX 4090 and RTX 4080 GPUs – this time the data is supplemented with frequencies and TGP values \u200b\u200b(Total Graphics Power, this parameter reflects the consumption of the entire video card).
The GeForce RTX 4090 will be built on the AD102-300 GPU with 16,384 CUDA cores and clock speeds up to 2520 MHz. The video card will receive 24 GB of GDDR6X memory with a frequency of 21 GHz. The memory bandwidth was 1 TB / s, like the GeForce RTX 3090 Ti. The system’s base TGP is 450W, but it will be highly configurable up to 660W. But this can only be expected from the best overclocked models.
GeForce RTX 4080 with 16 GB of memory will be built on the AD103-300 GPU with 9728 CUDA cores. The maximum frequency will be 2505 MHz – almost like the older model. This model will receive GDDR6X memory with a frequency of 23 GHz (it will be the only model in the line with such memory). TGP will be 340 watts, but manufacturers of finished devices will be able to increase it to 516 watts.
GeForce RTX 4080 with 12 GB of memory – could have come out as RTX 4070 Ti or RTX 4070, but, according to the source, Nvidia made a last-minute name change. This model will be based on the AD104-400 GPU with 7680 CUDA cores and clock speeds up to 2610 MHz. GDDR6X modules with a frequency of 21 GHz will be used as memory chips. The TGP of this model is 285W, the maximum TGP is 366W.
When the GeForce RTX 4080 with 12 GB of memory will be released is unknown. The GeForce RTX 4090 is expected to be available in the first half of October, the GeForce RTX 4080 in the first half of November.