The year 2019 was an eventful one for PC gamers. NVIDIA launched the Super variants of its RTX 20 and GTX 16 series lineups and AMD finally moved on from the ancient GCN design to the more gamer-centric RDNA architecture. The Radeon RX 5700 series kept the budget and midrange graphics card quite affordable while the older Polaris GPUs continued to sell in decent volume.
NVIDIA was forced to launch the GTX 1650 Super and 1660 Super when rumors of the Radeon RX 5500 XT and 5600 XT started popping up. While the 4GB variant of the 5500 XT stacked well against the GTX 1650 Super, the 8GB model didn’t offer any notable advantage, all the while costing as much as the 1660, a much faster GPU. This hampered the overall sales of the GPU, forcing AMD to reveal the RX 5600 XT. The card is set to launch on 21st January, with an aim to take down the GTX 1660 Ti while also trading blows with the RTX 2060.
Like usual, it seems like NVIDIA isn’t sitting still. Leaks suggest that team green in working on yet another GTX 16 series card, the GTX 1650 Ti tp keep AMD’s RX 5500 XT and 5600 XT in check. The info has surfaced on AIDA64 benchmark update-notes. It includes initial support for a GeForce GTX 1650 Ti (TU117) GPU.
Considering that the GTX 1650 uses the entire TU117 die, it’s highly likely that the 1650 Ti swaps out the GDDR5 memory for the faster GDDR6 standard. We’ve already seen with the GTX 1660 Super how of an impact faster memory can have on performance. That impact will be even more pronounced in the case of the GTX 1650 as it has a smaller 128-bit bus (the 1660 has a 192-bit bus).
Keep in mind that this could be an error but I think it’s highly plausible. Considering that the 5500 XT beats the 1650 Super and the 5600 XT levels with the GTX 1660 Super and the 1660 Ti, it won’t be surprising that NVIDIA responds with a Ti variant of the cheapest Turing GPU (at least it’s not Super Duper!).