Rubin (microarchitecture)

{{Infobox GPU microarchitecture

| name = Rubin

| image =

| caption =

| alt =

| launching = 2026

| discontinued =

| soldby = Nvidia

| designfirm = Nvidia

| manuf1 = TSMC

| process = 3NP or 3PN

| codename =

| products-desktop1 =

| products-hedt1 =

| products-server1 =

| directx-version =

| direct3d-version =

| shadermodel-version =

| opencl-version =

| opengl-version =

| opengles-version =

| cuda-version =

| optix-version =

| mantle-api =

| vulkan-api =

| opengl-compute-version =

| cuda-compute-version =

| directcompute-version =

| compute =

| slowest =

| slow-unit =

| fastest =

| fast-unit =

| shader-clock =

| l0-cache =

| l1-cache =

| l2-cache =

| l3-cache =

| memory-support = HBM4

| memory-clock =

| pcie-support =

| encode-codec =

| decode-codec =

| color-depth =

| encoders =

| display-outputs =

| predecessor = Blackwell

| variant =

| successor = Feynman

}}

Rubin is a microarchitecture for GPUs by Nvidia announced at Computex in Taipei in 2024 by CEO Jensen Huang. It is named after astrophysicist Vera Rubin and will consist of a GPU named Rubin and a CPU named Vera. The chips will be manufactured by TSMC using a 3 nm process and will use HBM4 memory. It is scheduled for mass production in late 2025 and will be available for purchase in early 2026.{{Cite web|url=https://www.tomshardware.com/pc-components/gpus/nvidia-rubin-revealed-as-blackwell-successor-powerful-vera-cpu-coming-too|title=Nvidia Rubin revealed as Blackwell successor, powerful Vera CPU coming too|author1=Mark Tyson with contributions from Paul Alcorn|date=June 2, 2024|website=Tom's Hardware}}{{Cite web|url=https://www.zdnet.com/article/nvidia-teases-rubin-gpus-and-cpus-to-succeed-blackwell-in-2026/|title=Nvidia teases Rubin GPUs and CPUs to succeed Blackwell in 2026|website=ZDNET}} Nvidia is using its own Blackwell GPUs to accelerate the design of Vera and Rubin, as well as Rubin's successor, Feynman.{{cite news|url=https://www.eenewseurope.com/en/nvidia-accelerates-feynman-chip-design-manufacture-on-blackwell-gpu/|title=Nvidia accelerates Feynman chip design, manufacture on Blackwell GPU|author=Nick Flaherty|work=eenews|date=March 18, 2025}}

Rubin Ultra

At Nvidia GTC 2025 it was announced that Rubin will be followed by an improved Rubin Ultra architecture in 2027.{{cite news|url=https://www.tomshardware.com/pc-components/gpus/nvidia-shows-off-rubin-ultra-with-600-000-watt-kyber-racks-and-infrastructure-coming-in-2027|author=Jarred Walton|title=Nvidia shows off Rubin Ultra with 600,000-Watt Kyber racks and infrastructure, coming in 2027|date=March 20, 2025|work=Tom's Hardware}} It would be in effect two of the Rubin cores connected together.

Performance

Rubin is stated to have 50 petaflops performance in FP4 (4-bit floating point math, often used for AI), increased from 20 petaflops in Blackwell, while Rubin Ultra will double the performance of Rubin with 100 petaflops.{{cite news|url=https://www.theverge.com/news/631835/nvidia-blackwell-ultra-ai-chip-gb300|title=Nvidia announces Blackwell Ultra GB300 and Vera Rubin, its next AI ‘superchips’|author=Sean Hollister|date=March 18, 2025|work=The Verge}}

References

Category:Nvidia microarchitectures

Category:Graphics microarchitectures