The leaks continue around the launch of the other models of the GeForce RTX series, now running into the NVIDIA GeForce RTX 4060 Ti. According to the well-known leaker kopite7kimiwe already have what would be the specifications de la GPU de gama media from the RTX 40 Series family.
Unfortunately, as seen in the rest of the model, the GPU It’s not like I’m going to make a big show of muscle in computing. Add to this that it might arrive with a memory interface inherited from the GeForce RTX 3050, so in some ways it looks more like a GeForce RTX 4050 than really the successor to the RTX 3060 Ti.
Possible specifications of the NVIDIA GeForce RTX 4060 Ti
RTX 4060 Ti has a very short reference board. The PG190 still uses CEM5 connector.
AD106-350-A1
4352FP32
8G 18Gbps GDDR6
32M L2
220W– kopite7kimi (@kopite7kimi) December 13, 2022
As indicated, this GPU will make use of the graphics chip NVIDIA AD106-350-A1. This would offer a configuration of 4,352 CUDA Colors, less compared to the 4,864 CUDA Cores than its predecessor. To this is added that the graphics chip will come accompanied by 8 GB of GDDR6 memory at 18 Gbps. Here we would have the same amount of memory as its predecessor, but faster (vs 14 Gbps).
Now, the problem is that this memory might get attached to one 128-bit memory interfacewhich would result in a bandwidth of 288 GB/s. This is much less than the 448 GB/s that reaches the GeForce RTX 3060 Ti. Although your memory is slower, it helps considerably the bandwidth of 256 bits. In this way, if all these data are fulfilled, in essence the name of an NVIDIA GeForce RTX 4060 Ti is too big for it.
Finally, it indicates a power consumption of 220Wjust 30W less compared to the GeForce RTX 4070. On the other hand, they are just 20W more compared to the GeForce RTX 3060 Ti.
Is the mid-range of NVIDIA less and less interesting?
It is well known that NVIDIA’s x60 series is always the most popular. These graphics cards have always offered the sweet spot between price and performance. With the GeForce RTX 3060 things got quite complicated. pulling memorythis graph has as 4 or 5 variants.
If we stop using memory, we review the information, and we remove the portable variants, NVIDIA has come to launch nothing less than six distinct GeForce RTX 3060 models. We have tres RTX 3060 (1x 8GB with 128 bits + 2x 12 GB with 192 bits), to which are added other three RTX 3060 Ti (2x con 8 GB GDDR6 y silicio GA104 y G103; 1x con 8 GB GDDR6X).
In this way, the comparisons are odious, since the GeForce RTX 4060 Ti feels even more unbalanced if we think that the GDDR6X variant sees how the 8 GB of memory reaches 19 Gbps that, with the 256-bit interface, the bandwidth skyrockets once more up to 608.3 GB/s (320 GB/s more than what the replacement presumably gives you).
It must be remembered that these specifications of the GeForce RTX 4060 Ti they are not official. But this graphics can not be very powerful, because the GeForce RTX 4070 would have 1,536 CUDA Core more. In this way, the entire mid and low range will be greatly throttled in performance.