VideoCardz has collected all the known insider information about the next generation of NVIDIA RTX graphics cards. To date, with a high degree of certainty, we know about the RTX 4090 with 24 GB of video memory and two variants of the RTX 4080 – with 16 GB and 12 GB. Official data on some of them may appear at the end of September.
GeForce RTX 4090 24 GB
This model is equipped with an AD102-300 GPU with 16384 CUDA cores, clocked at 2520MHz. The chip will receive more than 75 billion transistors. L2 cache up to 96 MB is expected.
The card has 24 GB of GDDR6X memory with a possible maximum bandwidth of 1 TB / s over a 384-bit bus. The default power consumption will be 450W, the maximum configurable TGP is 660W. This is the upper limit of the possible setting of this indicator in the BIOS, available to manufacturers.
GeForce RTX 4080 16 GB
RTX 4080 16 GB will get AD103-300 and 9728 CUDA cores. Its boost clock is 2505MHz, not much less than the RTX 4090. The TGP is set to 340W, peaking at 516W.
GeForce RTX 4080 12 GB
Previously, this card was referred to as the RTX 4070 Ti or RTX 4070. According to rumors, NVIDIA recently changed the name of this model. It is equipped with an AD104-400 GPU with 7680 CUDA cores capable of boosting the clock speed up to 2610 MHz. Uses GDDR6X memory. The power consumption of the RTX 4080 12GB is 285W and can go up to 366W.
The NVIDIA RTX 4090 is expected to launch in the first half of October, and the RTX 4080 in the first half of November. So far, there is an official embargo on the release of specifications and prices, and PCIe 5.0 support has not been officially announced. There are a few days left before the announcement of video accelerators (at least the older model) – most likely, it will take place on September 20 at the GeForce Beyond event as part of GTC 2022.
A production sample of the NVIDIA RTX 4090 graphics card poses for a photo – a four-slot design and a power consumption of up to 600W