Samsung’s chip-making arm is seemingly close to turning its fortune around. There was a time not too long ago when the South Korean company’s Exynos chips were frowned upon and apparently that’s why the more lucrative markets of North America and China got Qualcomm’s Snapdragon SoCs. That perception appears to have changed, thanks to Samsung’s adoption of Arm cores and the upcoming AMD GPU.
Samsung taking advantage of the situation by negotiating lower price for the Snapdragon 898
Per an earlier report, it’s possible that only a few markets will get the Exynos 2200 and that’s because of low yield and production issues.
That explains why Verizon is eager to get the chip. The latest report also says that Europe and Southeast Asia will continue to stick with Exynos chips but Korea may switch to Snapdragon. Samsung also plans to equip Chinese models with Snapdragon chips but China Telecom is allegedly also trying to procure the Exynos 2200.
The yield issues have purportedly also led to some internal disagreements between different units of Samsung. The high demand for the Exynos 2200 is apparently also being used as leverage to negotiate price cuts for the Snapdragon 898 with Qualcomm, so it looks like a win-win situation for Samsung.
Next year’s Galaxy A series midrangers could employ an AMD-powered Exynos chip
In late June, Samsung was said to be testing four configurations for AMD-based SoCs, including one with Arm’s latest X2 processing core, two with Cortex X1 CPUs, and one with the A78.
Apparently, new A-series phones will employ an A78-based Exynos chip – possibly the Exynos 1080 successor – with two compute units (CU). Its performance is said to be sufficient enough for mid-tier phones.
Latest AMD Exynos 2200 benchmark results
The Exynos 2200 is likely to use 6 CUs. According to previously available information, it will be built on Samsung’s 4LPP (4nm Low Power Plus) process, which succeeds the 5nm FinFET node that was used for Galaxy S21’s Exynos 2100.
The 4nm tech is based on the MBCFETTM structure (Multi Bridge Channel FET) that doesn’t suffer from the same performance and physical scaling limitations as the FinFET architecture.
- Manhattan 3.1: 127.5 Fps
- Aztec Normal: 90.7 Fps
- Aztec High: 39.65 Fps (Exynos 2100 cannot get past 30)