At the moment all this is just hearing, so treat them like that. However, they are well-supported rumors because NVIDIA’s A100 GPUs have a place to accommodate a large number of Shaders, so we can say that this product has room for development or launch of the RTX 3090. Ti or RTX 3000 TITAN using i -GPU same and without much effort.
AMD War with NVIDIA, as well as the RTX 3000 TITAN
From what we’ve seen so far, AMD’s upcoming GPUs won’t have a “gun” against NVIDIA’s high end, as its top Radeon RX 6000 GPU is expected to be similar or slightly out of the RTX 3080
The image that is thought to be in the bedroom of AMD is known as “Biggest Navi” (as this generation is known as Big Navi), and is said to be entering the market in the first quarter of 2021. Obviously, NVIDIA (who have room for improvement in their current GPUs as we mentioned earlier) will already be preparing their response in the form of the RTX 3000 TITAN.
Rumorally (and we also say that they are nothing but rumors) the Greatest Naviest GPU can fit up to 80 computer units, much more than the standard Radeon RX 6000 series, and may have HBM2 graphics memory instead of GDDR6 / X. However, like the Radeon VII, they will have very limited availability and will be targeted in the professional or enthusiastic market because of their high price.
If this turns out to be the case, what is the possible assumption that the RTX 3000 TITAN should have? Basically, GDDR6X graphics memory is essential, either at 21 Gbps already achieved or with a fast variation of 23 Gbps, scheduled for direct launch in 2021. In terms of memory capacity, on the PCB They have a lot of storage space up to 48 GB, like the Quadro 6000 GPUs experts have already filtered, but in the case of running at 23 Gbps they can equip “only” 44 GB to have a 352-bit bus.
In terms of its TDP / TGP, it is estimated to be about 370W in total, and since the new 12-pin connector used by Ampere has the capacity to provide 600 watts of power (according to NVIDIA, though they have not yet explained how) you will need such a connector to work.
In the end, the price will be very difficult to estimate (you know what NVIDIA looks like in this case); Officially it can cost less than a Quadro 6000 with a price of almost $ 3,500, but it’s all up in the air and more if there are rumors. We will see what happens, but what seems clear is that this will be NVIDIA waiting for AMD to take the initiative.