Extremely anticipated: Rumors round Nvidia’s Rubin-based GeForce RTX 60 sequence are choosing up tempo, however the newest wave of supposed spec leaks is already going through pushback. Most critical shops will urge readers to take the newest wave of RTX 60 “leaks” with an enormous grain of salt, whilst early technical chatter round Nvidia’s subsequent gaming structure begins to floor.
In accordance with VideoCardz and different sources, detailed specs for Rubin-based gaming GPUs merely do not exist in finalized kind but, that is regardless of a flood of assured claims circulating throughout social media and YouTube final weekend.
As Nvidia’s massive focus stays on AI {hardware}, Nvidia remains to be working with inner board numbers reasonably than public-facing SKU names, which means any report confidently labeling elements as RTX 6090, RTX 6080, or RTX 6070 must be handled as speculative at greatest. Clock speeds stay undefined, and Rubin chips supposed for GeForce playing cards reportedly have not even taped out but.
On that foundation, VideoCardz is framing any detailed “spec sheet” at present making the rounds as educated guesswork reasonably than data grounded in firm steerage. Extra credible indicators – overlaying clocks, efficiency, and reminiscence – are anticipated to emerge later in growth, significantly as soon as established leakers like Kopite7kimi start publishing information that may be independently cross-checked.
Nonetheless, the rumor mill is already producing granular Rubin claims that fans are eagerly dissecting. Early chatter suggests Nvidia might keep on with its roughly two-year GeForce cadence, with the RTX 60-series constructed on Rubin GPUs derived from the identical structure already deployed within the firm’s AI {hardware} stack.
On the manufacturing aspect, the gaming chips are most certainly to usea TSMC’s 3 nm FinFET course of reasonably than a extra aggressive sub-2 nm nanosheet node, doubtlessly through a customized variant much like how “Nvidia 4N” was tuned from TSMC’s 5nm tech.
Below this state of affairs, Rubin elements would carry “GR20x” die identifiers – akin to “GR202” for the flagship – and goal GPU frequencies within the high-2 GHz to low-3 GHz vary. That might translate into comparatively modest clock good points over immediately’s Blackwell-based GeForce {hardware}.
Characteristic-wise, Rubin is predicted to introduce sixth technology Tensor cores and fifth technology RT cores. The up to date Tensor blocks, paired with greater total compute throughput, are anticipated to enhance neural rendering and will assist convey DLSS 5 (up to now demonstrated utilizing two RTX 5090 GPUs) inside attain of a single card, doubtlessly increasing assist past flagship fashions.
In the meantime, the next-gen RT cores are mentioned to focus on a doubling of real-time path tracing efficiency in comparison with the RTX 50 sequence.
Rasterization good points, against this, are anticipated to be extra incremental. Leaks level to SKU-to-SKU enhancements within the 30-35% vary over Blackwell, pushed by architectural refinements, greater clocks, and effectivity good points from the 3nm node reasonably than main will increase in die dimension.
Reminiscence configurations may see extra significant shifts. A rumored RTX 6090 based mostly on “GR202” would retain a 512-bit GDDR7 interface, function 192 streaming multiprocessors on the complete die (with some disabled on transport playing cards), and pack 32 GB of reminiscence, doubtlessly with sooner GDDR7 speeds to push bandwidth greater.
A supposed RTX 6080 constructed round “GR203” is described with a 320-bit GDDR7 bus, 20GB of reminiscence, and no less than 25% extra bandwidth than the RTX 5080. Additional down the stack, a “GR205” RTX 6070 is rumored to maneuver to a 256-bit GDDR7 interface with 16 GB of reminiscence, delivering a bandwidth improve of no less than one-third over its predecessor.
