Almost a year of talking about the rumors surrounding the NVIDIA RTX 40 graphics cards, direct successors of the current RTX 30 and now, just now, new rumors are coming in that make us think we were wrong. And it is more than likely that those of Huang will eventually change the nomenclature of their graphics cards to RTX50, leaving on paper a name that seemed plausible and partly following the product leap from GTX to RTX. Therefore, the supposed RTX 5090 would be a beast according to what has been leaked.
Not that there are really any surprises in the performance of the RTX 4090/ RTX 5090 (it’s hard to give veracity to BOM change rumors, but they’re not implausible as such) since the filtering isn’t that it reveals unseen relevant information, but it focuses much more on the lens of the lens with which we see the graphics cards of the new batch.
The increase in the number of SMs as well as Shaders will go to a series of PCBs with very interesting characteristics, starting with the most high-end graphics card from NVIDIA, which will consume nothing less than 600 watts.
NVIDIA RTX 4090 / RTX 5090, two PCB versions and a single chip
Previous rumors certified up to three different PCBs that would encompass a launch of three graphics cards, where now a small change could give NVIDIA a lot more output. What has been revealed claims that the company’s supposed highest-end graphics card with the name RTX xx90 (it was curious that the full BOM was not given) will have two possible PCB layouts: a PG137 and a PG139 which focus on a single SKU-330, which clashes head-on with what we saw two weeks ago.
**90 PG137/139-SKU330, AD102-300, 24G 21Gbps, 600W
– kopite7kimi (@kopite7kimi) April 11, 2022
The leak would leave the RTX 4080 / RTX 5080 with the same SKU330 (at least in theory) and so NVIDIA would unify the criteria in this aspect by setting a clear guideline to only change the number of VRAM chips to thus go from 24 GB to 20 GB.
In other words, there would be two different PCBs and possibly a repeat of the RTX 30 game where one was the NVIDIA designed PCB and the other was the manufacturer’s reference PCB. But why then would NVIDIA change the name of its GPUs from RTX 40 to RTX50 if there are no substantial changes to chips, PCB or functionality?
RTX 50 and the Bad Omen of Number 4
What is whispered and spinning like wildfire is precisely that NVIDIA wants to avoid number 4 in its reference products. Why would you do this? It seems 4 in China is associated with death and it seems that those of Huang have become superstitious, but this clashes head-on with what was experienced more than a decade ago with the GTX 480.
Well, not that this series has been a sweet time for the greens, and since the Chinese market is getting stronger, NVIDIA may want to make this change leaving this bad luck behind.
The other theory is the evolutionary leap they’re going to take with the new graphics, which would leave room to leave one generation in the foreground and jump straight to the next in the purest style of traditional marketing and incidentally they’re approach AMD with their RX 7000 series in this aspect.
Perhaps both theories are true, though it’s more than likely we’ll never know the real reasons if, indeed, NVIDIA ends up calling its Ada Lovelace gaming GPUs the RTX 50.
