Nvidia Tips GeForce RTX 4060 Ti for Fast 1080p Play, With RTX 4060 Close Behind
Nvidia’s top-to-bottom rollout of its newest family of graphics cards, the GeForce RTX 40 Series, looks to be nearing completion, with two fresh midrange cards imminent. The GPU giant will be launching the GeForce RTX 4060 Ti on May 23 at a starting price of $399 for the 8GB version, which will be followed in July by an Nvidia GeForce RTX 4060 (without the "Ti") and a 16GB RTX 4060 Ti.
At the moment, Nvidia is keeping some fundamental details about these new GPUs secret, among them details about the core. We don’t know which GPU core these new cards will use, nor do we know how many CUDA cores, RT cores, or Tensor cores they will have. Clock speeds are also a mystery at this point, though Nvidia did toss out a few numbers on the computational power of the GPUs.
The compute numbers suggest that the GeForce RTX 4060 Ti will have significantly more compute horsepower than the RTX 3060 Ti that it replaces. (So should the RTX 4060, compared to its direct predecessor, the RTX 3060.) These numbers need to be taken with a pinch of salt, as compute performance does not directly correlate to real-world performance.
One key detail we do know about the new RTX 4060 and RTX 4060 Ti is that these cards will have far less bandwidth than their direct predecessors. Nvidia limited both of these graphics cards to a 128-bit memory interface, whereas the RTX 3060 Ti had a 256-bit memory interface, and the RTX 3060 12GB featured a 192-bit interface.
Memory bandwidth is vitally important to GPU performance, and this could hurt the new RTX 4060 and RTX 4060 Ti under certain circumstances. In the initial briefing we attended, Nvidia attempted to assuage concerns about the reduced bandwidth by highlighting the vastly increased pool of L2 cache on
Read more on pcmag.com
