After Nvidia officially launched the RTX 3090 Ti, the day of the release of the RTX 40 series is getting closer and closer, what performance will the new Ada Lovelace GPU perform, with the continuous emergence of breaking news, the GPU lineup and chip details of the new RTX 40 series have been quite comprehensive, such as the memory capacity and TBP power consumption of each SKU.
First of all, according to Twitter's well-known whistleblower @Kopite7kimi, the RTX 40 series starting lineup includes three SKUs of the RTX 4090 / 4080 / 4070.
Referring to the positioning of the RTX 30 Series, the high-end RTX 4090 / 4080 SKU will be paired with 24 / 16GB GDDR6X memory, while the mid-range RTX 4070 model will use 12GB GDDR6 memory.
In terms of core specifications, the latest rumors say that the RTX 4090 / 4090 Ti is based on the AD102 GPU, while the RTX 4080 / 4070 will use AD103 / AD104 chips.
Nvidia doesn't appear to have tested the latter two yet, but @Kopite7kimi speculate that the AD103 will have 64MB L2 cache / 350-450W of TGP power, while the AD104 chip will have 48MB of L2 cache / 300W TGP.
In terms of memory bandwidth, the current rumors have not yet disclosed much. One thing is for sure, though, that Ada Lovelace will make a generational leap similar to that of Maxwell to Pascal.
Finally, while many are expecting the RTX 40 series graphics cards to bring considerable surprises in 2022, even if market prices begin to gradually return to rationality, manufacturers who have not cleared the inventory of RTX 30 series GPUs are unlikely to make the starting player too proud.
(7916102)