2022 will be an interesting year in terms of the launch of gaming graphics cards, not only because of the potential departure of RDNA 3 from AMD and Lovelace from NVIDIA, but also because of the fact that early next year we will see Intel ARCs for the first time. . It is clear that in this situation NVIDIA is not going to sit idly by and is going to launch a renewal of its current generation in the form of its RTX 30 SUPER.
These would be the specifications of the NVIDIA RTX 30 SUPER range
What really interests us is to know how the second generation of graphics cards based on the GeForce Ampere architecture will be, and once again the most reliable information filter from NVIDIA, kopite7kimi, has surprised us again with information that if we break it down gives us quite juicy additional information about the RTX 30 SUPER.
What kopite has done is to give the data as follows: first the model, then the amount of ALU FP32, thirdly the memory density and to finish the type of this. It must be said that everything indicates that the capacities available per GDDR6X chip will continue to be 1 GB and not 2 GB as expected, but at the same time to highlight that the RTX 3070 SUPER is going to release this type of memory and this is important due to the fact that a change in the memory interface translates into a new chip, in this case a GPU.
Regarding the information regarding the number of 32-bit floating point units, we only have to divide it by 128 to obtain the number of SMs in the GPU, so the data obtained from the specifications of the RTX 30 SUPER that has given Kopite are as follows:
- The NVIDIA RTX 3090 SUPER have 84 SM, which is the full configuration in SM quantity of the GA102 GPU.
- In the case of RTX 3080 SUPER, SM quantity is 70, 2 more than the current RTX 3080. Although it is not the only improvement, since the configuration of 12 GB GDDR6X indicates that it has gone from having a 320-bit bus to having one of 384 bits, thus increasing its bandwidth and performance.
- The amount of SM of the RTX 3070 SUPER is 46 and therefore its configuration is the same as that of the current version. Its main improvement appears to be memory usage GDDR6X under a 256-bit bus and therefore 8 GB of this VRAM, allowing you to increase performance in scenarios where the GA104 GPU is limited by bandwidth.
- To finish we have the RTX 3060 SUPER with 44 SM units, which is the biggest jump from the second generation GeForce Ampere from 30 to 44 SM. Which makes us doubt that this model is using a GA106 and it is very likely that if this data is confirmed that it will use a variant of the GA104.
However, the data is not complete and we are still missing details such as the clock speed and other improvements that may have been added in this new generation of NVIDIA gaming graphics cards. For example, it is said that we are finally going to have a video codec that supports AV1 to encode and not just to play.
When will the second generation GeForce Ampere be available?
Given that it is rumored that the RTX 40 under Ada Lovelace architecture will appear well into 2022 then we can deduce that the RTX 30 SUPER should not be very far in time. What’s more, it is possible that they are already in production and that some other computer builder has even included it in their pre-built computers. As is the case with Hewlett Packard, who has already put on its website the HP ENVY 34 AIO-in-One PC with the RTX 3080 SUPER.
Considering that AIOs are often built with laptop parts, this is a confirmation that the RTX 3080 SUPER range could have simultaneous launch on laptops and desktops. The problem? The GA102 chip on which the RTX 3080 SUPER is based no laptop version, so this could be a typo and refer to the previous generation, specifically an RTX 2080 SUPER.
In any case, we hope to be able to buy them individually without problems this time around and not have to take a glove on already assembled computers in order to get an NVIDIA Gaming GPU without having to pay the surcharges caused by the mining bubble.