The NVIDIA GeForce RTX 5090 GPU will feature a massive monolithic GB202 “Blackwell” die

NVIDIA’s flagship GeForce RTX 5090 GPU will feature a massive and monolithic GB202 “Blackwell” board, reports Kopite7kimi.

NVIDIA GeForce RTX 5090 flagship will feature monolithic GB202 “Blackwell” die, but there may be a twist

Based on what we know so far, the NVIDIA GB202 “Blackwell” GPU will power the flagship GeForce RTX 5090 graphics card. Preliminary specs have revealed up to 192 SMs, which will represent up to 24,567 CUDA cores, if the chip retains the 128 cores per SM design from of AD102 “Ada” chips. Now, based on a new tweet from an insider, Kopite7kimi, it has been revealed that the GPU will use a monolithic design.

See more

physically monolithic

— kopite7kimi (@kopite7kimi) May 23, 2024

While NVIDIA has moved to a chip design for its HPC/AI chips such as the B100 and B200, it would appear that the company still wants to maintain monolithic packages for its consumer-focused GPUs. The GB202 “Blackwell” GPU is rumored to be physically monolithic in design, and we know from earlier reports that it is expected to have double the number of SMs and cores of the GB203, a less stripped-down board for its ilk. GeForce RTX 5080. This will make up a huge disparity in performance between the two cards, but the RTX 5090 is shaping up to be an absolute beast.

We’ll get to the NVIDIA GeForce RTX 5090 in a moment, but first let’s talk a bit more about the GB202 “Blackwell” GPU itself. So it looks like we’re looking at a monolithic design, but we also can’t rule out the possibility of a chiplet design under the hood.

NVIDIA can essentially pack two GB203 dies into a monolithic package without it looking like a chip design. It will allow for better inter-die communication rather than off-die communication bottlenecks associated with proper chip implementation. While NVIDIA has solutions to overcome bottlenecks such as NVLINK and other interconnects, they can be a bit more expensive as they increase the complexity of the GPU.

See more

Four years ago, we split the GA100 into two halves that communicate via interconnects. It was a big step – and yet almost nobody noticed, thanks to the amazing work of CUDA and the GPU team.

Today, this work is fulfilled with the launch of Blackwell. Two die. One amazing GPU. https://t.co/XuaUQPskkM pic.twitter.com/svRKhwPYEn

— Bryan Catanzaro (@ctnzr) March 18, 2024

However, NVIDIA already has a solution already on the market in the form of the GA100 and GH100, which are essentially two halves of a smaller die connected together and communicating via a split L2 cache. NVIDIA’s Bryan Catanzaro explained that this implementation improves scalability and their initial transition to this design went smoothly. The chip is also expected to be based on TSMC’s 4NP (5nm) process node, which improves density by 30% (transistor), so it should bring some nice improvements in addition to the architectural upgrade.

Now NVIDIA is likely to do the same on the gaming side, which means if the whole thing pays off, then we may see a B100/B200 style chip offering in the future.

Now back to the NVIDIA GeForce RTX 5090, now there are several reports that we could get a 512-bit interface on the next-gen flagship, and there is already information about a completely new cooling solution and PCB being worked on for this monster. card.

With rumors suggesting that AMD will step back from the ultra-high-end graphics performance segment with its RDNA 4 series, it looks like NVIDIA could further push its lead in the gaming segment with Blackwell GPUs once they hit the road. The NVIDIA GeForce RTX 5090 is expected to launch a few weeks after the RTX 5080, which is rumored to be Blackwell’s first gaming GPU on shelves.

NVIDIA Blackwell ‘GB202’ ‘Preliminary’ GPU Specifications: