Xiaomi’s XRING 01 Die Shot Analysis Reveals That The SoC Lacks Any SLC Cache, Sacrificing Some Performance, But Eliminating Any Unnecessary Increased Power Draw During Light Workloads

A myriad of decisions were made regarding the final design of Xiaomi’s XRING 01, with the company’s team of talented engineers intricately developing a solution that would compete with the flagship silicon of this generation without compromising on power efficiency. As you would expect, one way to directly meet the competition head-on was to leverage TSMC’s second-generation 3nm process, but according to some detailed analysis, it is more than simply choosing a foundry partner’s advanced manufacturing process. A closer look at the chipset reveals that the XRING 01 lacks any form of SLC cache, which has a downside and an upside, which we have discussed below.

To compensate for the performance loss as a result of removing the SLC cache, Xiaomi increased the CPU, GPU, and NPU cache of the XRING 01

The XRING 01 features a 10-core CPU cluster, with Xiaomi dedicating 16MB of L3 cache to this particular space on the die. Bear in mind that the silicon features the smallest die size compared to the Snapdragon 8 Elite, Dimensity 9400, and the A18 Pro, as increasing these dimensions leads to a price bump. In short, Xiaomi had to ensure that every bit of space is utilized efficiently, with each Cortex-X925 sporting 2MB of L2 cache, with 1MB being used by every Cortex-A725, and 512KB given to the Cortex-A520.

The 16-core ARM Immortalis-G925 GPU has 4MB of L2 cache, with Xiaomi’s in-house 6-core NPU sporting 10MB. The analysis performed by YouTube channel Geekerwan discovered that the XRING 01 lacks SLC cache, and mentions that this decision can result in some performance being left on the table. On the flip side, Xiaomi may have intentionally chosen this route because employing SLC cache on the die can potentially increase wattage when the XRING 01 is not doing any heavy lifting and is focused on handling low-power workloads.

Likely realizing that adding SLC cache to the custom chipset might adversely affect its efficiency, Xiaomi chose to omit this specification, but gave its CPU, GPU, and NPU a cache bump as compensation. As we think about the decisions Xiaomi made when designing the XRING 01, we also realize that instead of opting for the typical 8-core CPU, the company chose a 10-core configuration, possibly for the same reason. Well, it turns out that this choice worked in the company’s favor, as one Geekbench 6 result shows the SoC is barely slower than the Snapdragon 8 Elite in the multi-core test.

However, during the official announcement, Xiaomi claimed that the XRING 01 could obtain a score of 3 million in AnTuTu, but an independent test revealed that the in-house SoC achieves a 13 percent lower figure than advertised. It is not concluded whether this is due to the lack of an SLC cache. Compared to Google’s Tensor and Huawei’s Kirin, the XRING 01 is miles ahead when directly comparing custom solutions, so in the end, it is an exceptional effort from the company.

News Source: Geekerwan

Follow Wccftech on Google to get more of our news coverage in your feeds.