NVIDIA GeForce RTX 3050: New entry-level RTX card is here

NVIDIA officially released the entry-level GeForce RTX 3050 8GB graphics card, using the new “GA106-150” graphics core, with 2,560 CUDA Cores, 20 RT Cores and 80 Tensor Cores, the memory interface is reduced to 128bit, 8GB GDDR6 capacity, Replacing the old entry-level GeForce GTX 1650 / 1660 series, the suggested price is US$249, mainly targeting the 1080p gamer market. The editorial department found COLORFUL GeForce RTX 3050 NB Duo 8G-V for performance testing.

Entry-level GeForce RTX 3050 debuts

NVIDIA further lowered the threshold for RTX and DLSS technology, officially released the new entry-level GeForce RTX 3050 graphics card on the 26th, replacing the old GeForce GTX 1650 / 1660 market positioning, equipped with more CUDA Cores, faster second-generation RT Cores Ray tracing core, the faster third-generation Tensor core, the official suggested price is US$249, RTX and DLSS technology are no longer the patent of mid-to-high-end players.

According to the Steam hardware ranking data, the three graphics cards GTX 1050, GTX 1050 Ti and GTX 1650 occupy the top 5 positions. These products provide 1080p @ 60fps performance in the 3A masterpieces of the year, but by 2021 these entry-level cards have Powerful, GeForce RTX 3050 brings a good choice for these players to upgrade. NVIDIA official data points out that the performance of RTX 3050 is 30~50% higher than that of GTX 1650.

More importantly, GeForce RTX 3050 can support RTX On ray tracing technology and DLSS deep learning super sampling. According to NVIDIA data, RTX 3050 can provide 60+ fps game performance in large 1080p RTX on + DLSS Performance mode, which is GTX 1050 / GTX 1050 Ti and GTX 1650 can’t do it at all, because they don’t support it at all.

8nm process, GA106-150 graphics core

Many media reported earlier that the GeForce RTX 3050 will use the GA107 graphics core with lower scale and lower cost. This chip has actually appeared in the RTX 3050 Mobile version and the RTX 3050 Ti Mobile version. Mobile is the same, but the result uses the larger GA106 graphics core. From the Block Diagram, it can be seen that the GA106-150 has nearly 1/3 of the operation unit.

According to industry insiders, the original GA107 graphics core capacity is not enough to cope with the Notebook and Desktop markets at the same time, but NVIDIA does not want to delay the launch schedule of the RTX 3050, so the GA106 graphics core will be used for the time being, and it is expected to be introduced in Q2 2022. GA107 graphics core, GA106-150 will stop delivery immediately, there will be a Transition Period in the middle, depending on the actual SAMSUNG delivery status, it can be long or short.

“GA106” graphics core adopts 8nm NVIDIA Custom process, SAMSUNG foundry, has 13.25 billion transistors, Die Size is about 276mm², a complete GA106 chip has 3 GPC units, 15 TPC texture processing clusters and 30 built-in SM Streaming Multiprocessor with a total of 3,840 CUDA Cores, 30 RT Cores and 120 Tensor Cores.

However, the “GeForce RTX 3050” computing unit is masked by 1/3, reduced to 10 TPC texture processing clusters, with only 20 SM streaming multiprocessors, with 2,560 CUDA Cores, 20 RT Cores and 80 Compared with the RTX 3060, the scale is reduced by 1/3.

In terms of core clock, the default clock of GeForce RTX 3050 is 1,552MHz Base Clock, 1,777 MHz Boost Clock, and the highest TDP is 130W. In addition, the GeForce RTX 3050 memory interface is reduced to 128bit, 8GB GDDR6 memory capacity, the memory clock is 1,750MHz, the transfer speed is as high as 14Gbps, and the total bandwidth is 224GB/s.