InstaForex

Nvidia Will Release Old But New GPU Series

 


Nvidia's newest GPU series is currently the RTX 3000, but they will instead release a new GPU from the RTX 2000 series, the RTX 2050.

The RTX 2050 will be released after Nvidia released the RTX 3050 and 3050 Ti in early 2021. According to Nvidia, the RTX 2050 will be available in a line of laptops released in the spring of 2022.


Despite coming from the RTX 2000 series, the RTX 2050 is different from other GPUs in the series. Because the chip architecture uses GA107, the Ampere architecture. Not the Turing architecture as used in the RTX 2000 series.



This amperage is the chip architecture used by the RTX 3050 and other RTX 3000 series. The hope is that the RTX 2050 can provide performance that is close to the other RTX 3000 series but at a more affordable price.


The number of CUDA RTX 2050 cores is 2,048, the same as the RTX 3050, as well as the GDDR6 RAM capacity which is both 4GB. The difference is in the memory bus which is only 64 bits, not 128 bits like the RTX 3050.


The reasons for the release of the RTX 2050 may vary, from the scarcity of chips that impacted the RTX 3000 series, or it could be that Nvidia is trying to fill the gap between the GTX 1650 and RTX 3050 series. Namely GPUs with higher performance than the GTX 1650 but at a more affordable price than the RTX. 3050.


The RTX 2050 will be released with the MX550 and MX570 laptop GPUs. However, Nvidia has not revealed the full specifications for the two GPU chips. They only say that the GPU will have more CUDA cores and higher memory speeds.


Previously, Nvidia had just released a new version of its old GPU, the RTX 2060, which now uses a 12GB VRAM chip, up considerably from the first version released in 2019, which was 6GB.


For mid-2022 Nvidia is expected to release the RTX 4000, which is made using the 5nm process from TSMC. This GPU is said to have 50% higher gaming performance than the RTX 3000.


The use of 5nm fabrication is reportedly much more expensive than the 7nm fabrication used by AMD in the RX 6000 series GPU. This rumor is also reinforced by statements from sources in the component supply industry.


Digitimes in its report said Nvidia plans to use TSMC's chip-on-wafer-on-substrate (CoWoS) packaging technology in the H100 GPU series. This model will be based on Nvidia Hopper's compute architecture, and is expected to be Nvidia's first multi-chip GPU module.

Previous Post Next Post