This year, NVIDIA will release a new generation of gaming graphics cards, the Geforce RTX 50 series, based on the Blackwell architecture. The release is anticipated as soon as the fourth quarter of this year. Similar to the previous Geforce RTX 30 and 40 series, the initial focus will be on high-end models, with the Geforce RTX 5090 likely to debut first.

Recently, on the CHH forums, a user (@panzerlied) mentioned that the RTX 5090's base frequency is close to 2.9 GHz, significantly higher than the RTX 4090's 2.23 GHz, and it even surpasses the current flagship's boosted frequency of 2.52 GHz. However, this does not necessarily imply that the RTX 5090's boosted frequency is much higher, as NVIDIA may alter the method of frequency adjustment for its new GPUs. Regardless, we can expect the next-generation flagship cards to offer boosted frequencies of 3.0 GHz or more.

As previously stated, the RTX 5090 will feature 28GB of GDDR7 memory with a 448-bit bus width. This indicates that the GB202 chip onboard the RTX 5090 is not the full version, which usually has a maximum memory capacity of 32GB and a 512-bit bus width. The RTX 5090 is expected to have a Total Graphics Power (TGP) of 450W. Additionally, there is a version with a 350W TGP, featuring 8 memory chips corresponding to a 16GB capacity with a 256-bit bus width. The GB202 has a total of 192 SMs (Streaming Multiprocessors) equating to 24,576 CUDA cores, but the RTX 5090 likely won't enable all of them, using between 170 and 180 instead.

Moreover, the RTX 5090 Founders Edition is rumored to be only double-slot thick and equipped with dual fans, utilizing a triple PCB design with space for double-sided blow-through cooling. NVIDIA's partners have already started working on the thermal design for the new Geforce cards based on the Blackwell architecture.