What just happened? The GeForce RTX 2080 Ti has remained on top as the most powerful gaming GPU since its launch in September 2018. However someone recently spotted a new unnamed AMD Radeon ‘graphics device’ claiming the top spot in the OpenVR benchmark leaderboards with a framerate of 103.3 fps. The average RTX 2080 Ti gets ~80 fps in the same test, making the new GPU a whopping 30% faster. Even the very best RTX 2080 Ti result falls 15% short of the new unnamed card.
But what if we told you, the mysterious GPU may not be a Radeon, but a next-gen Nvidia part?
Before you start cheering for AMD, lets consider a few things: the graphics card was paired with an engineering sample of the Ryzen 7 4800H mobile processor, which has its own GPU built-in. A GPU that software registers as AMD Radeon Graphics. This is hardly an explicit title, so the software might have registered the correct name and we’re looking at a new AMD GPU.
On the other hand, all unreleased discrete GPUs from AMD have registered as AMD Radeon RX in the past. The most likely scenario is that the benchmark is registering the integrated GPU and not the discrete GPU that performed the benchmark.
Intuition might suggest that if the GPU is being tested with an unreleased AMD CPU, then it must be AMD doing the testing. But why would AMD test such a powerful GPU with a laptop CPU that it would never be paired with? Why would AMD double their chances of error by testing two engineering sample products together?
It’s worth noting that before this not a single benchmark had ever surfaced for a “big Navi” GPU. Nvidia on the other hand… their GPUs appear in online benchmarks ahead of release all the time. Yeah, I’ve got my tinfoil hat on here but bear with me.
Nvidia is significantly more likely to have a GPU this powerful in the late development stages (Ampere, anyone?). AMD has also sent out a lot of 4800H samples to their partners. We’ve already seen Asus testing one with an unknown Nvidia GPU. If Nvidia or an OEM wanted to test the new GPU while concealing its existence, then grabbing an AMD APU is one way to do it.
Of course, that’s a lot of effort to go to for one benchmark result. There are other ways to go about it, like just giving the GPU a new identification code.
To be honest, none of the theories really seem to fit (could also be Intel Xe… what?), but needless to be said, we seem to have an awesome new GPU to look forward to in 2020.