An eBay buyer acquired an engineering sample of the NVIDIA GeForce RTX 2080 Ti Super and verified its specifications through community analysis. The card arrives with 4,608 CUDA cores that match the count found on the standard RTX Titan. It uses a 384-bit memory bus to deliver 768 GB/s of bandwidth, which exceeds both the RTX 2080 Ti and the RTX Titan by more than 100 GB/s.
Community analysis confirms the card features 4608 CUDA cores and a 384-bit memory bus.
The graphics card carries physical badging that reads 'RTX 2080 Super' instead of the expected Ti designation. This discrepancy likely stems from NVIDIA repurposing a standard RTX 2080 Super middle piece for early engineering units rather than creating new tooling. The sample arrived with a sticker label identifying it as an RTX 2080 Ti Super, though the PCB itself lacks that specific nomenclature.
NVIDIA drivers required manual intervention to recognize the hardware during initial testing. Users successfully installed the software by modifying a configuration file in the driver package to acknowledge the new device ID. The card contains 12GB of GDDR6 memory, which cuts the VRAM capacity in half compared to the RTX Titan while maintaining identical core counts.
Community analysis suggests the engineering sample will deliver performance levels comparable to both the vanilla RTX 2080 Ti and the RTX Titan. Higher resolution gaming may see better results due to the increased memory bandwidth available on this variant. The discovery confirms NVIDIA was testing a faster memory configuration alongside standard core counts for its high-end Pascal lineup.
Source: TweakTown
