NVIDIA

RTX 4060 Ti features listed at retailer

The GeForce RTX 4060 Ti is coming step by step. The latest emerging in Russian retailers listingshas confirmed some of the features of the upcoming Ada Lovelace architecture card. We knew that the 4060 Ti would have 8GB of memory capacity. The GPU will communicate with the GDDR6 memories over the 128-bit bus.

NVIDIA will likely use modules running at 18 Gbps. If the leaks are correct, the RTX 4060 Ti will offer much lower bandwidth than the RTX 3060 Ti or even the RTX 2060 SUPER, with a total bandwidth of just 288GB/s.

The good news is that with the Ada Lovelace architecture adopted in the RTX 4000 series cards, the amount of L2 cache has increased to 32 MB. In addition, the new card is said to have 36% more TFLOP computing performance compared to the RTX 3060. It can therefore be significantly faster in terms of pure computational power.

Returning to the gaming part, the lower bus, low memory bandwidth of 288 GB/s and 8 GB of memory can be a bit of a bottleneck. Next-gen games are constantly pushing objects and textures into GPU memory as players navigate the game world. For example, Unreal Engine 5 games use Nanite because open game maps have become too large to contain all the necessary textures and assets in video memory. As a result, bandwidth is very important. Low width can lower frame rates when the GPU has to wait for new textures to stream before rendering a finished frame.

In addition, it should be noted that Palit branded cards are listed with four different frequency speeds ranging from 2310 MHz to 2685 MHz. Frequency rates seem a little lower than expected, but this information may not be definitively accurate. We will cover everything in detail at the end of May.

Yazar Fatih Işık