NVIDIA and its partners are preparing to launch the new RTX Super lineup next month with a reveal very soon (next week?). From what we know, the new GPUs will feature improved clocks, faster memory, increased shader counts and even a wider bus in some cases. However, it seems like the upgrades won’t be limited to just those. The GeForce RTX Super graphics cards might even feature better ray-tracing capabilities, thanks to a higher number of RTCores and Tensor cores.

Source: WikiPedia

Other than the RTX 2080, all the RTX cards are getting upgraded to the next tier die, with part of the chip disabled. This is where the additional CUDA cores will come from, and in case of the RTX 2060 more memory and a fatter bus as well.

TU104 die that powers the RTX 2080

However, each successively higher Turing card not only features more shaders, but also extra fancy RTCores and Tensors that power ray-tracing in Turing hardware. This will especially be more prominent in case of the RTX 2060 Super and the RTX 2070 Super as they’ll be getter much bigger dies compared to earlier, resulting in a higher RTCore as well as Tensor core count.

Related:

AMD Preparing X590 Platform for Ryzen 3000 CPUs, Meant for Enthusiasts

NVIDIA’s SM Engine

So while the RTX 2080 Super will be getting 2 additional RTCores, the RTX 2060 and 2070 Supers will be getting up to 10 more compared to the vanilla versions. The additional bandwidth and memory in case of the RTX 2060 Super should make the card more suitable for 1440p and in some cases even 4K.

Despite that though, the RTX 2080 Super won’t be a minor upgrade. We are looking at almost 200 more shaders, faster memory as well as an increased ROP and TMU figure. The RTX Super cards will challenge AMD’s recently announced Navi GPUs which feature the new RDNA architecture, but still, lack H/W level ray-tracing support. Expect an official announcement in the coming days.

Read more:

NVIDIA GeForce RTX Super Graphics Cards Launching in Mid-July

Image Credits: TechPowerUp

Leave a Reply