The next Computex event kicks off in just a few more days and we're expecting a whole bunch of announcements, which could include new graphics cards by all three major players—NVIDIA, AMD (
Radeon RX 9060 XT?), and Intel (
Arc B770?). Pre-event teasers are in full effect and so are a bevy of leaks and rumors. Regarding the latter, it's said that NVIDIA might be readying another addition to its GeForce RTX 50 series lineup.
While nothing has been confirmed, there is chatter that NVIDIA could soon release a GeForce RTX 5080 Super or GeForce RTX 5080 Ti. Before we get into the rumor, let's break down the existing GeForce RTX 50 series lineup, in terms of specs (CUDA cores, VRAM, memory bus, memory bandwidth) and MSRPs...
- GeForce RTX 5090: 21,760 CUDA cores, 32GB GDDR7, 512-bit bus, 1,792GB/s, $1,999
- GeForce RTX 5080: 10,725 CUDA cores, 16GB GDDR7, 256-bit bus, 960GB/s, $999
- GeForce RTX 5070 Ti: 8,960 CUDA cores, 16GB GDDR7, 256-bit, 896GB/s, $749
- GeForce RTX 5070: 6,144 CUDA cores, 12GB GDDR7, 192-bit, 672GB/s, $549
- GeForce RTX 5060 Ti: 4,608 CUDA cores, 8GB/16GB GDDR7, 128-bit, 448GB/s, $379 / $429
- GeForce RTX 5060: 3,840 CUDA cores, 8GB GDDR7, 128-bit, 448GB/s, $299
We've linked to our reviews of each model, save for the vanilla (read: non-Ti) GeForce RTX 5060, in case you want to take a deeper dive into the specifications, features, performance, and our overall analysis.
As we've seen in the past, NVIDIA is prone to refreshing its lineups with Ti and Super models. There already exist two Ti models in the GeForce RTX 50 series, and if a
post on Baidu (as spotted by
@harukaze5719, a reliable leaker on X) is any indication, there could soon be a third, or the first Super model in the Blackwell family.
According to the post, NVIDIA will release either a GeForce RTX 5080 Ti or GeForce RTX 5080 Super by the end of the year. Not much is known about the specs, save for it being "basically certain" (via Google Translate) that the upcoming card will feature 24GB of GDDR7 memory.
In terms of VRAM, the Ti/Super model would slot neatly between the GeForce RTX 5080 (16GB) and GeForce RTX 5090 (32GB). That kind of VRAM configuration could be possible by employing 3GB GDDR7 memory chips, as NVIDIA has done on the mobile version of the GeForce RTX 5090. All of the desktop variants employ 2GB memory chips.
We're speculating here, but if that's the case, it could eventually open the door to a more broad Super refresh. One of the biggest criticisms NVIDIA has faced over the years is the perception that it
skimps on VRAM, which is something AMD's marketing team has occasionally
pounced on.
Gamers would certainly welcome additional SKUs with more VRAM, though what the gaming community would really like to see is more widespread availability at baseline MSRPs.