A few days ago, a YouTuber specializing in leaking information and speculating on future hardware dared to state publicly that an alleged NVIDIA Titan ADA, which is said to be his highest performance graphics card, and which supposedly uses a double 16-pin connector, in the tests melted the power supplies. Which we don’t believe since advanced simulators are usually used for this. However, this is not the only nonsense we have seen, however, all of this needs an explanation
Will we see an RTX 4090 Ti from NVIDIA or is it cancelled?
In the midst of the rumor before the presentation of the new NVIDIA RTX 4000, we already knew that graphics cards with the high-end GPU, AD102, would use the new connector, mainly due to the fact that we had already seen the 12VHPWR in the RTX 3090 You a few months ago. The little difference in launch time between the RTX 3090 Ti and the RTX 4090 and the fact that the GA102 and AD102 chips have similar sizes already made us logically think that the PCB of the two cards was going to be the same or almost.
And so, not in vain, the name of the part on which the components of one are mounted receives the code name of PG136 for that of the RTX 3090 Ti and that of the RTX 4090 has been dubbed PG136D. And at first glance they look the same, except for a number of small details. So does the lack of an NVLink connector and the use of chokes that allow for higher power supply. Which would mark us for a potential higher-end model, using the full power of the new connector, 600W, instead of the 450W currently used by the RTX 4090.
An additional clue could also be found in the information of the maximum TGP supported by the RTX 4090 BIOS, which everything indicates would be 660 W, which would require the base power of PCI Express plus that of the new connector. However, the biggest clue is that the AD102 chip can have up to 144 SMs in its configuration, but the RTX 4090 has a configuration of 128.
So why didn’t NVIDIA release it?
Although the PCB of the two graphics cards, RTX 3090 Ti and RTX 4090, are almost identical, NVIDIA surprised us with a taller heatsink and fan. How is this possible if they supposedly evolve in the same range of consumption? Well, the simplest explanation is that the whole design of the RTX 4090 could be thinner, but the engineers decided to kill two birds with one stone and design a possible RTX 4090 Ti. The good thing is that it’s going to be great for overclockers and we’ll see some really cool stuff in the future.
Anyway, as for its absence at launch, that’s not surprising either, the same happened with its direct predecessor which came out 18 months after the launch of the RTX 30 and we weren’t torn our clothes. In any case, it is clear that NVIDIA has all the elements to launch an RTX 4090 Ti at some point, but they do not need it at the moment. Additionally, this board is not Titan ADA and will not require the use of a dual connector to function as many claim.
And speaking of TITAN Ada, what about her?
NVIDIA ditched the TITAN brand in the previous generation, but what has more numbers to come out is a professional graphics card, with full NVLink support and 48GB of GDDR6X. More than anything due to the fact that when faced with AI professionally, 24 GB of memory is insufficient. In any case, the PC gaming market is not ready to use a consumption beyond 700 W only on the graphics card. That doesn’t mean we don’t see custom boards with a dual connector, as happened with the 3090 Ti and some models.
In any case, given the size of the RTX 4090 for a consumption slightly higher than 600 W, it is clear that the eventual TITAN ADA would have needed more complex solutions such as the use of liquid cooling. In any case, we don’t think NVIDIA wants to go beyond a single 12+4 pin connector, at least in its own graphics cards designed to be installed in PCs for gaming. Is there a chance they’re getting ahead of themselves? Yes, if AMD with its RX 7900 XT gets its hands on the face of the RTX 4090.
