Text/Li Bing
It's no secret that many RTX3060 Ti graphics cards with Hynix memory often fail. There are some details that indicate which memory is most often burned.
I've written about the RTX3060 TI graphics card before, which malfunctions due to the problematic Hynix memory. Now that the story has a sequel, many of these graphics cards have been fixed.
The head of the service center discovered some flagged memory failures. Hynix memory, in particular, labeled H56G32CS4DX and PARTNAMBER-005, has product status CS. Product Status CS refers to memory debugging in pre-release releases and is generally not installed on commercial graphics cards. However, it was not disposed of, but was sent to the supplier to produce the graphics card directly. Obviously, the shortage of chips during the pandemic has been affected, but it is still necessary for manufacturers to install at least one usable memory on the graphics card.
The debug memory already shows the status of the MP (Mass Production), which is the only thing on retail products that needs to be tagged. That is, this memory does not disappear immediately, it will exist for a while.
One of the reasons is that the chip crystal is unstable due to temperature, that is, the memory cannot operate for a long time under the declared factory parameters. This problem is also exacerbated by inadequate cooling system efficiency, especially for the RTX3060 Ti series. It will make their temperature rise above 90 degrees, or even 100 degrees. You can use the GPU-Z program to check for specific memory on the graphics card, but it doesn't show the memory tag. If you want to see this mark, you'll need to take it apart, but it can cost you after-sales service.