Please note: Most, if not all, of the articles published at this website were completed by Chat GPT (chat.openai.com) and/or copied and possibly remixed from other websites or Feedzy or WPeMatico or RSS Aggregrator or WP RSS Aggregrator. No copyright infringement is intended. If there are any copyright issues, please contact: bicycledays@yahoo.com.
Nvidia CEO Jensen Huang disclosed in his keynote Sunday night, on the Computex annual commerce present in, Taipei, Taiwan, that the successor to the corporate’s “Blackwell” GPU and CPU household will likely be known as “Rubin” named for US astronomer Vera Rubin, and will likely be made out there within the first half of 2026.
The announcement reinforces Huang’s pledge for Nvidia to be on a one-year cadence of releasing new chip architectures.
Earlier than clicking the slide to point out the Rubin line, Huang quipped to his viewers, “That is the very first time this subsequent click on has been made, and I am undecided but whether or not I’ll remorse this or not.”
“All of those chips are in full improvement, a 100% of them, and the rhythm is one yr, on the limits of expertise, all 100% architecturally suitable,” mentioned Huang, “and all the richness of software program on high of it.”
The shock unveiling of Rubin comes sizzling on the heels of the formal announcement of Blackwell by Huang on the GTC convention in March.
In keeping with a analysis observe from Wall Avenue brokerage Jefferies & Co., citing information website Wccftech, the Rubin chips, together with a GPU chip household, and a CPU chip named Vera, will transfer from utilizing the present quickest pc DRAM reminiscence, “HBM3e,” to utilizing the following technology, “HBM 4.” The GPU chips will likely be designated as “R100” GPUs.
The chips will likely be fabricated by Taiwan Semiconductor Manufacturing in a 3-nanometer course of expertise, mentioned Jefferies analysts, and can start manufacturing within the fourth quarter of 2025.
The Rubin makes use of what’s known as a “4x reticle design,” the analysts observe, which implies that it takes up extra of the floor space of a 12-inch silicon wafer than does the Blackwell, which is simply 3.3x. The “reticle” is the world of a wafer {that a} typical photolithography system can tackle with one go when laying down the circuits of a chip.
Huang additionally disclosed a forthcoming successor to the Blackwell GPU, Blackwell Extremely. The Rubin structure, mentioned Huang, may also have its “Extremely” model, following in 2027.
Take a look at all of the Computex information from Nvidia on the corporate’s information website or this round-up weblog put up on the occasion from the corporate.
