Nvidia GeForce RTX 5090 and 5080 Specifications Leak

Leak Specifications reveal a HUGE GAP between the RTX 5080 and RTX 5090

The hardware leaker @kopite7kimi has revealed the specifications of Nvidia’s RTX 5090 and RTX 5080, the company’s highest-end “Blackwell” graphics cards. These two GPUs will likely be the first RTX 50 series GPUs to be released, with the RTX 5090 being much higher-end than the RTX 5080.

Nvidia’s RTX 5080 reportedly features 10752 CUDA cores, a 256-bit memory bus, and 16GB of GDDR7 memory. The GPU is said to feature a 400W TDP, and earlier rumours suggest that it will be 10% faster than an RTX 4090.

With these specifications, Nvidia’s RTX 5080 will consume more power and be much faster than today’s RTX 4080. The RTX 4080 has a TDP of 320 watts, and uses a 256-bit memory bus with slower GDDR6 memory.

Moving onto the RTX 5090, it looks like Nvidia’s new flagship is essentially a doubled-up RTX 5080. The GPU features 32GB of GDDR7 memory over a 512-bit memory bus and 21760 CUDA cores. Yes, it doesn’t have exactly two times as many CUs… it has more than that!

Given its alleged TDP of 600W, it is likely that this GPU is power-constrained. Perhaps that’s why this GPU reportedly uses two 12V-2×6 power connectors. If this is the case, the RTX 5090 may respond well to overclocking, assuming that users can adequately cool such a power-hungry GPU.

If these specifications are correct, then Nvidia’s RTX 5090 will be a beastly gaming graphics card. This GPU will demand a huge price tag. After all, this GPU has more than two times as many CUDA cores as the RTX 5080. It also has exactly two times as much RAM as its lower-end counterpart. Those are insane specifications.

You can join the discussion on Nvidia’s leaked RTX 5090/5080 Specifications on the OC3D Forums.