New speculation about Nvidia's future generation Ada includes information about the clock speeds of the flagship GeForce RTX 4090. According to this, Ada would clock significantly higher than the current Ampere series.
First of all, it has to be mentioned that the previously reliable kopite7kimi in particular has been reporting on changing specifications very often lately. And even if these may not be final, they will probably not change fundamentally. In other words, the new rumors should be treated with a great deal of caution.
Clock rates are initially only available for the expected GeForce RTX 4090. According to this, the base clock should be 2,235 MHz and the boost clock 2,520 MHz. Assuming that Nvidia does not suddenly calculate the clock rates differently, the frequencies are significantly higher than those of the GeForce RTX 3090 Ti (test), which is specified at 1,560 MHz and 1,860 MHz.
3 GHz is within reach with Ada
As is well known, GeForce graphics cards can also clock even higher than the boost specification. And that should also be the case with the GeForce RTX 4090. More than 2,750 MHz should be possible, but it is unclear whether this also applies in the factory state. Anyhow, Nvidia will make a big step up in clock speeds with Ada, which is not surprising given the significantly higher frequencies on AMD's competing RDNA 2 models – and RDNA 3 will clock even higher.
Apart from that, there is little new to report or the rumor mill is going in circles. The GeForce RTX 4090 is said to be the flagship model with 16,384 FP32 ALUs, 24 GB of GDDR6X memory on a 384-bit interface with 21 Gbps data throughput and a TDP of 450 watts. The GeForce RTX 4080 should rely on 10,240 ALUs, 256 bit, 16 GB GDDR6X with 21 Gbps at 420 watts. The significantly reduced number of shader units in combination with the barely lower power consumption indicate that the GeForce RTX 4080 will probably clock a good bit higher than the GeForce RTX 4090.
RTX 4070 supposedly with similar performance to the RTX 3090 Ti
The GeForce RTX 4070 is said to have 7,168 ALUs and a 160-bit memory interface with 18 Gbps fast GDDR6 memory with a total capacity of 10 GB. The TDP is given as 300 watts. Meanwhile, the leaker Kepler claims to have found out that the GeForce RTX 4070 is said to have comparable performance to the GeForce RTX 3090 Ti. However, more detailed information is missing, making it difficult to classify.
The GeForce RTX 4090 is said to start in October, followed by the GeForce RTX 4080 in November and the GeForce RTX 4070 in December. According to current speculation, the competing RDNA 3 generation may also be launched in October.
Continue to update.
RTX 4090, AD102-300-A1, 16384FP32, 384bit 21Gbps 24G GDDR6X, 450W, base 2235 boost 2520 actual max >2750;
RTX 4080, AD103-300-A1, 10240FP32, 256bit 21Gbps 16G GDDR6X, 420W,
RTX 4070, AD104-275-Kx(x is a number)-A1, 7168FP32, 160bit 18Gbps 10G GDDR6, 300W.— kopite7kimi (@kopite7kimi) July 4, 2022