A screenshot of an alleged spec sheet of the NVIDIA RTX 3080 Ti has leaked out via Twitter (@CyberPunkCat). I will say upfront that the leak has some oddities, technical inconsistencies and is not verified and is therefore in solid rumor territory.
A screenshot of an alleged spec sheet of the NVIDIA RTX 3080 Ti has leaked out via Twitter (@CyberPunkCat). I will say upfront that the leak has some oddities, technical inconsistencies and is not verified and is therefore in solid rumor territory. As with all rumors – a grain of salt is highly recommended by the doctor. If unverified rumors aren’t your peace of cake – stop reading now.
Rumored spec sheet of an NVIDIA RTX 3080 Ti leaks out
Before we begin, keep in mind that the nomenclature for the upcoming cards has not been confirmed yet. Previous leaks have pointed out that they may be called RTX 3080, RTX 3090 and a 2nd generation RTX TITAN as well as Ti and SUPER variants. In other words, we know next to nothing about which chip may be called what. The only thing we know for sure is that there are currently three variants in the works – called the SKU 10, 20 and 30. SKU 10 is the GA102-200 GPU, SKU 20 is the GA102-300 GPU and SKU30 is the GA102-400 GPU.
🧐🧐🧐🧐🧐 GA102-400-A1……. pic.twitter.com/q6ZAu9XDBa
— CyberPunkCat (@CyberCatPunk) June 23, 2020
According to previous, highly reliable leaks, there will be an RTX 3090 in the lineup – which this particular rumor goes against. Speeds for the memory has also been leaked, which this rumor goes against as well. This rumor seems to be based on information that we would now consider to be invalid so either it is completely not true or most of the leaks from reliable leakers have been wrong. [opinion] Personally, I would trust previous leaks a lot more considering they have an established history of reliable information [/opinion].
Twitter user KARMAA summed up the inconsistencies with prior leaks:
TDP vs TGP.
CUDA cores vs CUDA Cores.
TFLOPS are wrong.
7nm (presumably TSMC) vs (Samsung – Dunno what process, probably 8nm)
GDDR6 vs GDDR6X.
Honestly, none of this lines up with other supposed leaks. Only thing that is constant is the word ‘Ampere’.
— KARMAAA 🇺🇸 🇦🇺 (@VOLTECYT) June 23, 2020
If this is true, however [opinion] which I consider unlikely [/opinion] then you are looking at a 22 TFLOPs RTX 3080 Ti instead of an RTX 3090 or a TITAN. Since this is called the GA102-400 chip, this is purported to be the SKU30 which is the flagship. We can also derive the rated boost clock considering the discrepancy between the TFLOPs and shown boost clocks. At 21 rated TFLOPs the “official” boost clock would be 1953 MHz. This is roughly a 100-200 MHz increase from the rated boost clocks of last generation and is pretty plausible in itself (if you ignore all the other technical inconsistencies).