NVIDIA sent out a short letter this morning announcing the release date for their next GeForce desktop video card, the RTX 3060. The mainstream (ish) video card, previously released at CES 2021 with a late February release date, is now slated for a February 25 launchde, with prices starting at $ 329.
As a quick summary, the RTX 3060 is the next card in NVIDIA’s Ampere Architecture video card stack. Using the new GA106 GPU – already shipped in RTX 3060 laptops – the RTX 3060 follows the traditional price / performance cadence for video card launches, with NVIDIA releasing a cheaper and lower performing video card for the mainstream enthusiast video card market. NVIDIA’s 60-level cards have long been the workhorse parts for 1080p games – as well as some of their highest-volume parts in North America – and the RTX 3060 is expected to play the same role in the Ampere / 30 series.
| NVIDIA GeForce Specification Comparison | ||||||
| RTX 3060 | RTX 3060 Ti | RTX 2060 | GTX 1060 | |||
| CUDA core | 3584 | 4864 | 1920 | 1280 | ||
| ROP’s | 64? | 80 | 48 | 48 | ||
| Boost Clock | 1.78 GHz | 1.665GHz | 1.68 GHz | 1.709 GHz | ||
| Memory Clock | 14Gbps? GDDR6 | 14 Gbps GDDR6 | 14 Gbps GDDR6 | 8 Gbps GDDR5 | ||
| Memory bus width | 192-bis | 256-bis | 192-bis | 192-bis | ||
| VRAM | 12GB | 8GB | 6GB | 6GB | ||
| Single precision perf. | 12.8 TFLOPS | 16.2 TFLOPS | 6.5 TFLOPS | 4.4 TFLOPS | ||
| Tensor Perf. (FP16) | 51.2 TFLOPS | 64.8 TFLOPS | 51.6 TFLOPS | Nvt | ||
| Tensor Perf. (FP16-yl) | 102.4 TFLOPS | 129.6 TFLOPS | 51.6 TFLOPS | Nvt | ||
| TDP | 170W | 200W | 160W | 120W | ||
| GPU | GA106 | GA104 | TU106 | GP106 | ||
| Transistor count | ? B | 17.4B | 10.8B | 4.4B | ||
| Architecture | Ampere | Ampere | Turing | Pascal | ||
| Manufacturing process | Samsung 8nm? | Samsung 8nm | TSMC 12nm “FFN” | TSMC 16nm | ||
| Start date | 21/02/2021 | 12/02/2020 | 15/01/2019 | 19/07/2016 | ||
| Introductory price | MSRP: $ 329 | MSRP: $ 399 | MSRP: $ 349 | MSRP: $ 249 Founders $ 299 |
||
NVIDIA already published the most specifications for the card in January. Including the fact that it offers 28 SMs (3584 CUDA cores), and 12 GB GDDR6 running on a 192-bit memory bus. As with previous 60-level cards, the memory bus without power-of-two means that NVIDIA sends out a somewhat odd amount of memory, in this case 12 GB, which is actually more than is even available on the RTX 3080. The only other option is an anemia-for-2021 6GB, and NVIDIA prefers to make sure the card does not require VRAM capacity.
Meanwhile, the RTX 3060 is for better or worse guaranteed to fly off the shelves quickly. Since each video card is more powerful than a GTX 1050 Ti apparently climbed into Ethereum’s mining, desperate gamblers will fight with hungry miners for their supplies. Even with the 192-bit memory bus, I would be shocked if the RTX 3060 is not profitable, especially with Ethereum reaching record heights. So for anyone who wants to grab the card, be well prepared to get to your favorite retailer or e-retailer that Thursday morning.
Finally, unlike the other RTX 30 series cards introduced so far, NVIDIA will not be producing any Founders Edition cards for the RTX 3060 series. All the cards that are released are therefore AIB cards with their own design. And if the tradition holds true, you should not be surprised to see that the AIBs are equipping their cards with premium features and raising their prices accordingly.
