It is rumored that Nvidia will not change the price of the new GeForce RTX 4060 Ti compared to its predecessor, and it will thus reach 399 USD. If that were to happen, the price of new graphics cards could approach the 10,000 mark in crowns. We probably shouldn’t expect it below that, but it could be somewhere between 10.5-11 thousand in that case. Personally, I wouldn’t rule out some middle option, i.e. that Nvidia will stop somewhere halfway and the cards won’t even be $399 like the predecessor, but also won’t keep the price jump between the xx70 and xx60 Ti in the form of only $100 (ie the $499 price). For example, a price of 449 USD is offered, but also various other amounts. But for now, let’s stick with the fact that the latest behind-the-scenes information is talking about $399. There are still probably a few weeks left before the cards launch, and we probably wouldn’t be surprised if there were a few more leaks around prices (and it’s still a bit early to be sure about the leaks in this regard).
| GeForce RTX 4060 Ti | GeForce RTX 4060 | GeForce RTX 3060 Ti | GeForce RTX 3060 | |
| Number of CUDA cores | 4352 | 3072 | 4864 | 3584 |
| Basic beat | 2310 MHz | – | 1410 MHz | 1320 MHz |
| Boost rate | 2535 MHz | – | 1665 MHz | 1777 MHz |
| FP32 performance | 22 TFLOPS | – | 16,2 TFLOPS | 12,74 TFLOPS |
| Memoirs | 8 GB GDDR6 | 8 GB GDDR6 | 8 GB GDDR6 | 12GB GDDR6 |
| Memory speed | 18 Gbps | 18 Gbps | 14 Gbps | 15 Gbps |
| Memory bus | 128-bit | 128-bit | 256-bit | 192-bit |
| Memory throughput | 288 GB/s | 288 GB/s | 448 GB/s | 360 GB/s |
| TGP | 160 W | 115 W | 200 W | 170 W |
On the other hand, there is already a fairly clear idea regarding the specifications of the cards. And given that the RTX 4060 Ti isn’t going to make an extreme move against its predecessor, the $399 price tag would also make sense. After all, the number of CUDA cores will not only not increase, but even decrease by about 10%. The cards want to catch up with a much higher clock, the Boost is supposed to be 2535 MHz, while the previous generation was only 1665 MHz, which is more than half an increase. The gross performance of the chip should thus go from 16.2 to 22 TFLOPS (an increase of 36%).
Unfortunately, the memories are disappointing. Here, it seems to stay at 8 GB, which was not enough for the previous generation, let alone today. And to make matters worse, there is talk of halving the memory bus, i.e. from 256 to just 128 bits. Although faster memory chips are to be used, this should still mean a drop in memory throughput from 448 to 288 GB/s (-36%). The good news is that the TGP of the cards is to be reduced from 200 W to 160 W, so the mainstream could finally see a little more reasonable GPU consumption again.
2023-04-25 22:05:23
#Nvidia #coming #senses #GeForce #RTX #cost
