With less than a month to go until the grand unveiling of GeForce RTX 50 Series, questions surrounding graphics card specifications only continue to grow in their intensity and number. Sadly, this brand of curiosity may come with a tinge of disappointment. It appears that Nvidia plans to largely mirror its current generation offerings in terms of memory capacity albeit with a large boost to bandwidth.
According to WCCFTech, there’s good news and possible cause for concern. On the positive front, GeForce RTX 5090 will feature 32GB of VRAM up from the 24GB buffer on GeForce RTX 4090. Better still, its bandwidth will skyrocket from 1TB/s to an enormous 1.79TB/s thanks to the jump to GDDR7 memory modules. It’s difficult to imagine what use-case would saturate the graphics card barring some 8K extremities but they’re impressive specifications nonetheless.
Elsewhere in GeForce RTX 50 Series, bandwidth is similarly greater but buffers remain somewhat stagnant. To Nvidia’s credit, it will apparently boost GeForce RTX 5070 Ti and 5060 Ti capacities to match GeForce RTX 4070 Ti Super and GeForce RTX 4060 Ti 16GB, respectively. However, non-Ti graphics cards aren’t as lucky.
| GPU | VRAM | Bandwidth |
|---|---|---|
| GeForce RTX 5090 | 32GB GDDR7 | 1,792GB/s |
| GeForce RTX 5080 | 16GB GDDR7 | 1,024GB/s |
| GeForce RTX 5070 Ti | 16GB GDDR7 | 896GB/s |
| GeForce RTX 5070 | 12GB GDDR7 | 672GB/s |
| GeForce RTX 5060 Ti | 16GB GDDR7 | 448GB/s |
| GeForce RTX 5060 | 8GB GDDR7 | TBC |
Perplexingly, GeForce RTX 5070 will only feature 12GB of VRAM, 4GB than the less-powerful and presumably cheaper GeForce RTX 5060 Ti. The biggest concern on the board, though, is undoubtedly the prospect of GeForce RTX 5060 sporting a small 8GB buffer. Even with bandwidth increases bolstering its capabilities, it’s difficult shake the feeling history could repeat itself ala GeForce RTX 4060.
Indiana Jones and the Great Circle highlights the constraints of an 8GB buffer in the modern gaming landscape, but the weaknesses inherent to this configuration are nothing new. While it’s true that upscaling technologies like DLSS Super Resolution alleviate this burden, not forgetting whatever Nvidia could have planned for GeForce RTX 50 Series, these are performance enhancers and not fixers. Speaking from experience, these features cannot overcome hard limits of technological limitations in some scenarios.
Naturally, all of this hinges on the aforementioned specifications manifesting into an actual graphics card and we should all take the above with a pinch of salt in the meantime. Should this technical makeup manifest, though, I’m intrigued to see how Arc B580 will hold its own against GeForce RTX 5060 given the former’s 12GB VRAM buffer.
Nvidia should reveal the first GeForce RTX 50 Series cards during its CES 2025 press conference, on January 7, 2025. The company is throwing a big bash in the lead-up to launch dubbed ‘GeForce Greats‘, complete with prize draws. Come the day Jensen Huang takes the stage in Las Vegas, you’ll find Club386 on the showfloor ready to report on all the announcements.
Give Club386 a follow on Google News to make sure you don’t miss any of our daily news coverage, as well as reviews and other thoughts on all things tech.
