The memory granule is upgraded from DDR6 to DDR6X, and the frequency of graphics card BOSS reaches 1.670 MHz, which is 5 MHz higher than the old version of the existing version, and the cache frequency also reaches 1.9 GBBS, which is the same as rtx3070K.

In terms of parameters, the main difference between the new model and the old model is that the new model is far ahead of the old model in the number of cuda cores, optical drive cores, bit width and bandwidth. However, the old model has not been completely crushed by the new model, and the memory size and memory speed of the new model have surpassed those of the old model.

Therefore, when the new 3060ti 8GB memory is fully loaded, the old model may narrow the gap with the new model.

What is the performance of the new 3060ti?

From the naming point of view, RTX 3060 Ti is a new generation dessert graphics card that replaces RTX 2060 SUPER. Thanks to the new NVIDIA Ampere architecture, RTX 30 series graphics cards have more CUDA stream processor units than the previous generation. RTX 3060 Ti has 4864 CUDAs, which is more than twice that of RTX 2060 SUPER.

Compared with RTX 3070, both of them use GA 104 medium-sized cores, but RTX 3070 retains 46 groups of SM units, while RTX 3060 Ti expands to 38 groups of SM units, CUDA? The quantity is 2 1% less, and the specifications are quite different.

In terms of memory, RTX 3060 Ti and RTX 3070, the previous generation RTX 2060 SUPER and official rival RTX 2080 SUPER all have 8GB GDDR6 memory, and the memory frequency is only slightly higher than that of RTX 2080 SUPER, so the memory bandwidth of the four graphics cards is similar.