Nvidia RTX3060 12GB delivers a new level of performance to mainstream markets. The RTX 3060 officially launches today at $329. It’s possible to be realistic? Realistically? Nvidia’s Ampere architecture powers many of the top graphics cards and is attracting huge amounts of interest from gamers as well as cryptocurrency miners. Nvidia has updated firmware and driver code to detect Ethereum mining. However, when people are willing and able to pay extremely high scalper prices on eBay for cards such as the GTX 1660 Super or RTX 2060 GPU benchmarks hierarchy, all of our GPU benchmarks hierarchy is almost sold out. Nvidia has even partnered with partners to bring back old generation Turing cards and Pascal cards.
This does not make it a bad GPU. However, we expect the RTX 3060 will be as hard to obtain as any modern GPU. The current Ethereum mining boom will eventually fade, but it may take several years before we see an end to the shortage of chips. This shouldn’t be surprising, but it is a sad reality for those who have been waiting for a reasonable-priced gaming PC upgrade.
Nvidia will not offer the RTX 3060 Founders edition, unlike previous Ampere GPUs. We’re now looking at a third-party card. Nvidia sent us the EVGA GeForce RTX3060 XC to review for this launch. It is a compact card that’s not too noticeable. The card has no backplate (or plastic), no RGB lighting, and two custom-sized 87mm fans to cool the system with a 2.0 slot form factor. This card measures 202x110x38mm in size and weighs 653g. It is quite a change from the other third-party Ampere cards that we have reviewed.
Of course, there are good reasons. It costs money to create a mainstream card with all the bells, whistles, and features. We believe that gamers looking for good value will be better served by simple designs and good performance. The RTX 3060 will have some extreme versions, some of which will cost more than the RTX 3060T options. Let’s not forget that even the fastest RTX3060 won’t be able to beat a 3060T in most cases — yes, even with 12GB VRAM. Memory capacity doesn’t matter if you exceed 8GB. The 3060 Ti has a significant advantage due to its larger memory bus and greater memory bandwidth. The 3060 Ti also has 35% more GPU cores.
Here are the details: comparing the RTX 3060 to its closest Ampere sibling, and Turing predecessors. The RTX 2060 Super and 2060 Super demonstrate how things have changed between Turing, Ampere for the -60 suffix cards. Ampere offers a lot more shader CPUs which can lead to a much better computational performance and a slight increase in memory bandwidth for the 12GB cards. The VRAM capacity is doubled (at least until Nvidia’s RTX3060 6GB comes out), and there are improvements in the RT, Tensor, and memory subsystems. All of these factors lead to improved performance. The power consumption is similar to the RTX3060 Ti’s 220W, but it has a 170W total graphics power (Total GraphicsPower).
Nvidia is using 15Gbps GDDR6 for the first time. Except for the RTX 2080 Super, which came with 15.5Gbps VRAM, all RTX 20-series cards used 14Gbps memory. This narrows the bandwidth gap between 3060 Ti and 3060 Ti slightly, but the GA104 cards still have a clear advantage due to the 64-bits more interface width. GA106 has the same 48 ROPs, Render outputs as the RTX2060, so it doesn’t have any advantage.
The GA106 chip contains 12 billion transistors, compared to 17.4 billion in GA104. This reduces the die size to 393mm square from 17.4 billion in GA104. It also helps to lower the cost and increases the number Nvidia can obtain from a single wafer. In case you were wondering, GA106 is half the size of GA102 which measures 628.4mm sq and contains 28.3 million transistors. Nvidia can obtain around 130 dies per wafer using GA104 (some of these are defective and most end up as partially disabled chips), while GA106 is smaller and allows for about 200 dies per Wafer. Higher yields equal more graphics cards. This is the goal.
