NVIDIA’s Blackwell-powered GeForce RTX 50 Series Founders Edition GPUs may feature a multi-part PCB design, if rumours are to be believed. According to the popular leakster Kopite7kimi, they found the rumour posted on the Chinese-language forums, Chiphell, stating that the board design would be split into three parts.
Kopite themself stated that these multi-part PCB would comprise the main board, the IO rigid board, and a separate PCIe Slot. However, they pulled back slightly with the last bit, saying that it should not be considered as “the third PCB”.
I love your sketch. You won’t be disappointed. I think everyone would prefer a compact cooler. https://t.co/kcSINpFzva pic.twitter.com/XCsMFKe5va
— kopite7kimi (@kopite7kimi) May 23, 2024
It wouldn’t be the first time NVIDIA has employed a multi-part design on its products. The Hopper H100 board, for example, comprises multiple elements in its construction, with a small rigid board being the anchor to everything else.
There isn’t a whole lot of information about this claim but that being said, it would be easy to understand why NVIDIA may do something like this with the RTX 50 Series. The upcoming Blackwell -powered cards are expected to feature up to 16 GDDR7 modules beneath all that metal. Technically speaking, it should consist of two GPU clusters, connected in a manner similar to the way NVIDIA connects its data centre accelerators.
NVIDIA is expected to launch the GeForce RTX 50 Series later this year, starting with the RTX 5080 first, followed by the RTX 5090, in a pattern similar to when it launched the RTX 30 Series all those years ago.
(Source: Chiphell, X)
