NVIDIA's unreleased GeForce RTX 2080 Ti SUPER has emerged through a rare engineering sample, providing the first concrete evidence of a graphics card that was rumored but never officially announced. The prototype recently appeared in an online auction listing, revealing specifications that place it significantly closer to the Titan RTX than the retail GeForce RTX 2080 Ti. At the heart of the card sits a fully enabled TU102 graphics processor featuring 4,608 CUDA cores. For comparison, the retail RTX 2080 Ti shipped with 4,352 CUDA cores, while the Titan RTX utilized the same fully unlocked TU102 configuration found on this prototype. The engineering sample also upgrades the memory subsystem, pairing 12 GB of GDDR6 memory with a full 384-bit memory interface. The retail RTX 2080 Ti, by comparison, featured 11 GB of GDDR6 memory connected through a 352-bit bus. The hardware effectively positioned the card between the GeForce RTX 2080 Ti and Titan RTX. Internally, the project was reportedly viewed as a product intended to close the performance and pricing gap between those two models. Had it reached market, it would likely have delivered performance approaching Titan RTX levels while carrying a significantly lower price tag.
Photographs of the prototype show several unique design elements. Unlike retail RTX 20-series Founders Edition models, the engineering sample combines a glossy front shroud with an all-black rear backplate.
Interestingly, the cooler assembly carries RTX 2080 SUPER branding despite the card's Ti-class specifications. This has led to speculation that NVIDIA reused existing SUPER-series cooler components during the development process. The engineering sample appears to have progressed beyond an early prototype stage. Reports indicate the card carries a valid NVIDIA product serial number, suggesting it was relatively close to final validation.
However, because the product was ultimately cancelled, no official driver support was ever developed. Users attempting to run the hardware reportedly need to enable Windows test mode and manually modify NVIDIA driver INF files to add support for the device's hardware identifier. Industry observers have long speculated about why NVIDIA never launched an RTX 2080 Ti SUPER despite releasing SUPER refreshes for other Turing-based GeForce models. The specifications of this newly discovered prototype provide a possible explanation.
A fully enabled TU102 GPU with a wider memory interface would have narrowed the performance gap to the Titan RTX considerably, potentially creating internal competition for NVIDIA's flagship enthusiast product. Only a small number of engineering samples are believed to exist. As a result, the appearance of this prototype offers a rare look at an alternate path NVIDIA considered for its Turing product stack before deciding to keep the Titan RTX isolated at the top of the lineup.
| Specification | RTX 2080 Ti | RTX 2080 Ti SUPER Prototype | Titan RTX |
|---|---|---|---|
| GPU | TU102 | TU102 | TU102 |
| CUDA Cores | 4,352 | 4,608 | 4,608 |
| Memory Capacity | 11 GB GDDR6 | 12 GB GDDR6 | 24 GB GDDR6 |
| Memory Bus | 352-bit | 384-bit | 384-bit |
| Status | Released | Cancelled Prototype | Released |
Source: ithome