RTX 5080 Super rumored with 24GB of memory — Same 10,752 CUDA cores as the vanilla variant with a 400W+ TGP

New, possible details about a potential RTX 5080 Super have emerged, thanks to an X posting from well-reputed hardware leaker Kopite, who has a solid track record when it comes to all things Nvidia. The RTX 5080 Super is alleged to address GPU memory limitations with 24GB of fast GDDR7 memory, and a TGP (Total Graphics Power) that's rumored to exceed 400W. The leaker has not shared any details in regards to pricing and availability.

GDDR7 memory modules are currently available in 16Gb (2GB) and 24Gb (3GB) densities, with even higher capacities down the pipeline. The denser option directly translates to a 50% increase in VRAM capacity, even if the memory bus width remains the same.

Despite sharing the same 256-bit bus width as the RTX 5080, the RTX 5090 Mobile achieves its 24GB configuration with denser memory modules. The remainder of the Blackwell family sticks with standard 16Gb modules, similar to GDDR6X. However, this new leak suggests Nvidia might be eying transitioning to denser 24Gb options for the RTX 50 Super refresh.

The RTX 5080 Super reportedly employs the full-fat GB203 die, similar to the RTX 5080, with 10,752 CUDA cores or 84 Streaming Multiprocessors. The 256-bit interface enables eight memory modules, which have been populated with 24Gb (3GB) modules for 24GB of GDDR7 memory, rated at 32 Gbps. This puts the memory bandwidth at an impressive 1 TB/s, or 6.6% faster than the stock RTX 5080. Nvidia is also said to increase the power requirements for the RTX 5080 Super over 400W.

It's improbable Nvidia will release a Super/Ti model for the premier RTX 5090 with 24Gb modules, as that would likely impact the sales of its professional GPUs. The most affordable Blackwell workstation card, the RTX Pro 4000 Blackwell, which is suggested to retail near $1,500, is equipped with 24GB of memory. This 24GB capacity is likely the maximum Nvidia will offer with its Super refresh for consumers, as going any further would risk encroaching on its distinct market segmentation, where higher VRAM capacities are key selling points for AI/ML developers and enthusiasts.

The RTX Pro 4000 also serves as a reasonable indicator for the price tag, where the RTX 5080 Super should cost somewhere between $1,000 to $1,500, assuming the MSRP is honored. The best possible outcome would be to pull an RTX 4080 Super and decrease the base MSRP, but that's wishful thinking, especially in a market where MSRPs are no more than a suggestion.

At the other end of the spectrum, all 128-bit models, which constitute the entire RTX 5060 family, have the potential to be loaded with up to 12GB of memory. But the final decision rests on Nvidia, which has recently been in hot water for restricting preview drivers from reviewers and limiting pre-release tests to very specific benchmarks, unduly focused on Multi Frame Generation.

Follow Tom's Hardware on Google News to get our up-to-date news, analysis, and reviews in your feeds. Make sure to click the Follow button.