The rumor mill has started grinding about NVIDIA’s next-generation gaming GPU, and it looks like NVIDIA does not want to allow the current market environment to get in the way of implementing its product roadmap, with a roughly 2-year GeForce generation product launch cadence. The next-generation GeForce RTX 60-series will be powered by the “Rubin” graphics architecture. “Rubin” already debuted on NVIDIA’s bread-winning AI GPU series, and is making its way to GeForce, reports RedGaming Tech.
The first slice of rumors about GeForce RTX 60-series predicts that NVIDIA will stick to a more conservative approach with foundry nodes, and not go with a sub-2 nm nanosheet-based node. GeForce “Rubin” will be built on some variant of the current TSMC 3 nm FinFET node. They need not be the same N3 node that’s in use by Apple, Intel, and others; and NVIDIA might collaborate with TSMC on an exclusive variant just the way it created the NVIDIA 4N node, derived from TSMC N5. Chips in the series will follow the numbering scheme “GR20x,” with examples being “GR202” for the biggest part powering the flagship product. The 3 nm node will allow NVIDIA to maintain GPU clock speeds ranging between high 2 GHz and low 3 GHz, which is a minor increase over the current “Blackwell.”
GeForce “Rubin” will introduce the new 6th Gen Tensor cores, and 5th Gen RT cores. We already have some clue about the design goals of the two. The new 6th Gen Tensor cores, along with the generational increase in the main compute muscle of the GPU, should better accelerate neural rendering and DLSS 5, the revolutionary new technology that augments game assets with AI-generated effects in real time. NVIDIA’s early demo of DLSS 5 required a pair of RTX 5090 graphics cards, so the goal should be to bring DLSS 5 within the performance envelope of a single GPU, including mid-range and performance segment ones, not just the high-end. Next up, is the 5th Gen RT core, and RedGaming Tech says that NVIDIA is chasing a 100% gen-over-gen increase in real-time Path Tracing performance over the current RTX 50-series “Blackwell” series.
As for the traditional pure raster 3D graphics performance, it’s predicted that “Rubin” will post modest generational gains, ranging between 30% to 35% SKU-to-SKU versus the current RTX 50-series. This would be banking on the IPC, clock speed, and energy-efficiency advances brought about by 3 nm and the new “Rubin” SM (streaming multiprocessor).
Memory size and bandwidth could increase SKU-to-SKU over the current “Blackwell” generation, with the exception of the top RTX 6090. NVIDIA will stick to GDDR7 as the memory standard, however, the company could increase memory bus width in pursuit of memory size and bandwidth gains. The top-spec “GE202” silicon powering the RTX 6090 will retain a 512-bit GDDR7 memory bus of its predecessor, the RTX 5090. NVIDIA might tap into faster GDDR7 speeds for modest bandwidth increase. The silicon has 192 SM, and the RTX 6090 could be cut down from this. 32 GB might remain the memory size for the RTX 6090.
The RTX 6080 could be based on the “GR203” silicon, which physically features a 320-bit wide GDDR7 memory interface for 20 GB of memory, along with a memory bandwidth increase of at least 25% over the current RTX 5080.
The RTX 6070 will see a major memory size and bandwidth increase. NVIDIA will build the RTX 6070 on the “GR205” silicon, which comes with a 256-bit GDDR7 memory interface, and the RTX 6070 will get 16 GB of it, for an at least 33% memory bandwidth increase
