Recently, there is news that Nvidia has once again adjusted the GeForce RTX 4070, reducing the power consumption of the graphics card from 300W to 285W. It has 60 SMs, that is, 7680 CUDA cores, equipped with 12GB of GDDR6X memory, a bit width of 192 bits, a speed of 21 Gbps, and a memory bandwidth of 504 GB/s. The overall specifications are also improved than the original rumors.
Recently, Twitter user @kopite7kimi further disclosed the information of GeForce RTX 4070, saying that its base clock is 2310 MHz, its boost clock is 2610 MHz, and its max clock can reach more than 2800 MH. For comparison, the current GeForce RTX 3070 has a base frequency of 1500 MHz and a boost frequency of 1725 MHz. The foundry of the new generation of GPUs has changed from Samsung to TSMC, and the advancement of technology has made the GeForce RTX 40 series a leap in frequency.
According to estimates, the theoretical performance of the GeForce RTX 4070 can be doubled on the basis of the GeForce RTX 3070, from 20 TFLOPs to 40 TFLOPs, which is comparable to the GeForce RTX 3090 Ti. However, in terms of graphics card power consumption, the GeForce RTX 4070 still has an increase compared to the GeForce RTX 3070, from 220W to 285W, which is similar to the GeForce RTX 3070 Ti.