Yesterday, we saw the possible design of an Nvidia GeForce RTX 3080 in prototype phase or engineering sample, now there are rumors about its performance, indicating that it would come with the GA102-200 silicon, which would offer a total of 4352 CUDA Cores under the Nvidia Ampere @ 7nm architecture that could be accompanied by 10 GB of GDDR6 memory.

You Might Like: AMD Radeon RDNA2 and Nvidia Ampere Graphics cards would launch in September

It is estimated that if this silicon reaches a factory frequency of 1.75 GHz, we would be talking about a performance of at least 15 TFLOPs, or in other words, it would be almost 40 percent more powerful than the current GeForce RTX 2080 with a performance around the 11 TFLOPs, although of course, now all the doubts arise about the price, if Nvidia will maintain them, thus having 40 percent more performance at the same price as the RTX 2080, or if it will position itself as a more expensive GPU.

These data are similar to those leaked at the end of April, where it was revealed that a GeForce RTX 3080 would feature 4068 CUDA Cores. Rumors also revealed a possible GeForce RTX 3090 with 5248 CUDA Cores and GDDR6X @ 21 Gbps memory to take advantage of the PCI-Express 4.0 interface. If the leaked images of graphics card are real then it is to be expected that we will begin to have more leaks in the coming week.