Forum » Strojna oprema » NVIDIA Rubin (3nm GPU)
NVIDIA Rubin (3nm GPU)

oo7 ::
2026 že nova serija NVIDIA grafičnih kartic 
NVIDIA "Blackwell" Successor Codenamed "Rubin,"
Codenamed "Rubin," after Vera Rubin, the new architecture will power NVIDIA's future AI GPUs with generational jumps in performance, but more importantly, a design focus on lowering the power draw.
NVIDIA's first AI GPU based on "Rubin," the R100 (not to be confused with an ATI GPU from many moons ago); is expected to enter mass-production in Q4-2025
A Q4-2025 mass-production roadmap target would mean that customers will start receiving the chips by early 2026.
https://www.techpowerup.com/322282/nvid...

NVIDIA "Blackwell" Successor Codenamed "Rubin,"
Codenamed "Rubin," after Vera Rubin, the new architecture will power NVIDIA's future AI GPUs with generational jumps in performance, but more importantly, a design focus on lowering the power draw.
NVIDIA's first AI GPU based on "Rubin," the R100 (not to be confused with an ATI GPU from many moons ago); is expected to enter mass-production in Q4-2025
A Q4-2025 mass-production roadmap target would mean that customers will start receiving the chips by early 2026.
https://www.techpowerup.com/322282/nvid...
- spremenil: oo7 ()

oo7 ::
Well, I can announce that after Blackwell, there's another chip. And we are on a one-year rhythm. And so and you can also count that.
New CPUs, new GPUs, new networking NICs, new switches, a mound of chips that are coming. And all of it, the beautiful thing is all of it runs CUDA.
https://wccftech.com/nvidia-accelerates...
New CPUs, new GPUs, new networking NICs, new switches, a mound of chips that are coming. And all of it, the beautiful thing is all of it runs CUDA.
https://wccftech.com/nvidia-accelerates...
Zgodovina sprememb…
- spremenil: oo7 ()

oo7 ::
NVIDIA Unveils Vera Rubin as next-gen GPU architecture patron: Rubin Ultra to Sport 12 HBM4 stacks
https://videocardz.com/newz/nvidia-unve...
Q4 2025 ali Q1 2026.
https://videocardz.com/newz/nvidia-unve...
Q4 2025 ali Q1 2026.

errhec ::
--the Rubin platform will succeed the upcoming Blackwell platform, featuring new GPUs, a new Arm-based CPU — Vera — .
V TEJ novici je omenjeno.
V TEJ novici je omenjeno.
dd

oo7 ::
TSMC 3nm Set To Witness Massive Adoption From AI Tech Giants; NVIDIA Rubin, AMD Instinct MI355X & Intel Falcon Shores
https://wccftech.com/tsmc-3nm-witness-m...
https://wccftech.com/tsmc-3nm-witness-m...

oo7 ::
GeForce RTX 50 and RTX 60 SKU names spotted in first EEC filing
https://videocardz.com/pixel/geforce-rt...
https://videocardz.com/pixel/geforce-rt...

oo7 ::
Rubin naj bi izšel 6 mesecev prej kot je bilo načrtovano ?
NVIDIA's Next-Gen "Rubin" Architecture Is Now Rumored To Be Released By "Six Months" Ahead of Schedule
https://wccftech.com/nvidias-next-gen-r...
NVIDIA's Next-Gen "Rubin" Architecture Is Now Rumored To Be Released By "Six Months" Ahead of Schedule
https://wccftech.com/nvidias-next-gen-r...

FlyingBee ::
AI se jim splača, dokler imajo takšno prednost, da izkoristijo moment.
P200 MMX, 32mb ram, 2gb HDD, s3 virge 2mb, 14" CRT 640x480
New Sphincter Kvartet:
Roko Spestner, Namaž Zlevčar, Daje Heading, Maraje Spetan
New Sphincter Kvartet:
Roko Spestner, Namaž Zlevčar, Daje Heading, Maraje Spetan

oo7 ::
NVIDIA's Next-Gen Rubin AI Architecture Now Rumored To Launch By H2 2025; Disrupting Market Competition
NVIDIA's Rubin Architecture To Now Debut Six Months Earlier Than Schedule, Credits To SK Hynix's "Speedy" HBM4 Development
https://wccftech.com/nvidia-next-gen-ru...
Se splača potem sploh kupiti Blackwell če bo že letos Rubin ?
NVIDIA's Rubin Architecture To Now Debut Six Months Earlier Than Schedule, Credits To SK Hynix's "Speedy" HBM4 Development
https://wccftech.com/nvidia-next-gen-ru...
Se splača potem sploh kupiti Blackwell če bo že letos Rubin ?


FlyingBee ::
To ne bo client a.k.a geforce.
P200 MMX, 32mb ram, 2gb HDD, s3 virge 2mb, 14" CRT 640x480
New Sphincter Kvartet:
Roko Spestner, Namaž Zlevčar, Daje Heading, Maraje Spetan
New Sphincter Kvartet:
Roko Spestner, Namaž Zlevčar, Daje Heading, Maraje Spetan

oo7 ::

oo7 ::
NVIDIA Gears Up for Next-Gen Vera Rubin AI Servers While Rolling Out Blackwell Ultra GB300 Units; A Relentless Standard Only Jensen Dares to Set
https://wccftech.com/nvidia-gears-up-fo...
https://wccftech.com/nvidia-gears-up-fo...

LeQuack ::
Na žalost jim sedaj AI piše driverje, tako da za gaming so kartice zelo neoptimizirane in veliko crashev.
Quack !

FlyingBee ::
Verjetno Geforce Rubin konec 2026 ali januar 2027, glede na to, da bodo Superce konec 2025.
P200 MMX, 32mb ram, 2gb HDD, s3 virge 2mb, 14" CRT 640x480
New Sphincter Kvartet:
Roko Spestner, Namaž Zlevčar, Daje Heading, Maraje Spetan
New Sphincter Kvartet:
Roko Spestner, Namaž Zlevčar, Daje Heading, Maraje Spetan

oo7 ::
Some quick assumptions about @Nvidia's new Rubin CPX. Quiet similar to GB202, with some obvious changes to the GPCs. Looks like it has full raster units (full GPU) with up to 256 ROPs. Could this be the RTX 6090?
https://x.com/highyieldYT/status/196577...
https://x.com/highyieldYT/status/196577...

FlyingBee ::
Ob pomanjkanju konkurence, 94% dGPU market share, vprašanje koliko jim je sploh še mar za Geforce consumer grafične.
P200 MMX, 32mb ram, 2gb HDD, s3 virge 2mb, 14" CRT 640x480
New Sphincter Kvartet:
Roko Spestner, Namaž Zlevčar, Daje Heading, Maraje Spetan
New Sphincter Kvartet:
Roko Spestner, Namaž Zlevčar, Daje Heading, Maraje Spetan

FlyingBee ::
Povsem verjetno bo tak 6090 monster.
If we assume NVIDIA disables two full GPCs, we're still looking at a GPU with 28,672 CUDAs and 224 ROPs, absolutely dwarfing the GeForce RTX 5090 with its 21,760 CUDAs and 176 ROPs.
If we assume NVIDIA disables two full GPCs, we're still looking at a GPU with 28,672 CUDAs and 224 ROPs, absolutely dwarfing the GeForce RTX 5090 with its 21,760 CUDAs and 176 ROPs.
P200 MMX, 32mb ram, 2gb HDD, s3 virge 2mb, 14" CRT 640x480
New Sphincter Kvartet:
Roko Spestner, Namaž Zlevčar, Daje Heading, Maraje Spetan
New Sphincter Kvartet:
Roko Spestner, Namaž Zlevčar, Daje Heading, Maraje Spetan

Magic1 ::
Zanimivo, da se sploh ukvarjajo s x090 serijo, saj je silicij primerljiv s AI čipi, prodajajo pa ga za vsaj 10x manj denarja. Na B200 sta dva rahlo šibkejša čipa, kot na 5090, s več spomina (192Gb). Cena pa 40k$+. Eh ja, kaj dela monopol...
Nvidia naredi nov čip za 6090, nato se ga malo opimizira, da dva v sendvič in ima nov AI čip za 60k$. Ali pa obratno, naredi Ai čip in ga pol samo navije za 5090. Nah, prva opcija ima več smisla, saj v tem primeru ni tolk štala, če je kaj narobe. Gamerji so super testerji. Učijo se na x090 in pol preverjeno uporabijo na Bx0.
Nvidia naredi nov čip za 6090, nato se ga malo opimizira, da dva v sendvič in ima nov AI čip za 60k$. Ali pa obratno, naredi Ai čip in ga pol samo navije za 5090. Nah, prva opcija ima več smisla, saj v tem primeru ni tolk štala, če je kaj narobe. Gamerji so super testerji. Učijo se na x090 in pol preverjeno uporabijo na Bx0.
Magic
Zgodovina sprememb…
- spremenil: Magic1 ()

FlyingBee ::
.
@OpenAI
and NVIDIA just announced a landmark AI infrastructure partnership:
? 10 gigawatts of compute
? Millions of NVIDIA Vera Rubin GPUs across gigascale AI factories.
The next era of AI
@OpenAI
and NVIDIA just announced a landmark AI infrastructure partnership:
? 10 gigawatts of compute
? Millions of NVIDIA Vera Rubin GPUs across gigascale AI factories.
The next era of AI
P200 MMX, 32mb ram, 2gb HDD, s3 virge 2mb, 14" CRT 640x480
New Sphincter Kvartet:
Roko Spestner, Namaž Zlevčar, Daje Heading, Maraje Spetan
New Sphincter Kvartet:
Roko Spestner, Namaž Zlevčar, Daje Heading, Maraje Spetan

Jarno ::
Nova arhitektura je lahko hitrejša in varčnejša, več enot pomeni več resoursov za določen task in več taskov naenkrat - za uporabnike.
Težava osnove za trening in internega AI SW-ja je nekaj drugega.
Težava osnove za trening in internega AI SW-ja je nekaj drugega.
Chuck Norris je med števili 0.999... in 1 uspel vriniti konstanto imenovano CN.
#65W!
#65W!
Vredno ogleda ...
Tema | Ogledi | Zadnje sporočilo | |
---|---|---|---|
Tema | Ogledi | Zadnje sporočilo | |
» | AMD UDNA - GPUOddelek: Strojna oprema | 865 (226) | oo7 |
» | Nvidia Feynman - GPUOddelek: Strojna oprema | 887 (444) | GenZNPC |
» | Nvidia privzema strategijo tik-takOddelek: Novice / Procesorji | 3942 (159) | Ghost7 |