It seems Nvidia is still in the process of finalising specs for its upcoming RTX 40 series GPUs. According to reports this week, Nvidia is now considering two specs for the RTX 4070. One would have the card drawing more power, paving the way for additional performance, while the other spec seen in rumours is a bit more tame on power draw, resulting in less performance.
The two specs shared by kopite7kimi put the RTX 4070 with either 7,680 CUDA cores and 12GB of GDDR6X memory at 21Gbps or 7,168 CUDA cores and 10GB of GDDR6X memory. In the first configuration, the estimated Time Spy Extreme score is just below 11,000 points, and the total power draw is about 285W. As for the 10GB config, the Time Spy Extreme score drops to less than 10,000 points and the power draw to 250W.
Review the two possible specs of RTX 4070:
PG141-SKU340/341, 7680FP32, 12G 21Gbps GDDR6X, total card power 285W, TSE <11000.
PG141-SKU336/337, 7168FP32, 10G 21Gbps GDDR6X, total card power 250W, TSE <10000.
I found that I had made a typo before, mistaken wrote 341 as 331. ?— kopite7kimi (@kopite7kimi) August 29, 2022
Considering the rumoured specs and performance of the RTX 4080, the performance gap between this and the RTX 4070 will be bigger if Nvidia decides to go for the 10GB variant. Launching an RTX 4070 Ti later down the line can close that gap. However, even the 12GB variant would still offer a big enough difference in performance compared to the RTX 4080 to launch a new SKU between the two, assuming recently leaked RTX 4080 specs and performance estimates are accurate.
Nvidia is expected to officially reveal the RTX 40 series soon, likely at its upcoming GTC September conference, followed by a Q4 rollout for the first GPUs.
Discuss on our Facebook page, HERE.
KitGuru says: Do you think we'll end up seeing a 10GB RTX 4070 or a 12GB version instead?