With the launch of the Microsoft Surface Pro X yesterday, the Redmond-giant also introduced a new ARM processor called the SQ1. The SQ1 is a 7nm 7W part built in collaboration with Qualcomm and is really a Snapdragon 8cx with an incremental GPU update. While the Snapdragon 8cx features the Adreno 680 GPU, the SQ1 features a more powerful Adreno 685.
The SQ1 processor is an octa-core chip that can reach a peak clock speed up to 3 GHz. The Adreno 685 GPU can offer 2.1 TFLOPS of compute performance (possibly, FP16)— 2x more than the integrated GPU power found in 8th gen Intel Core processors or the Qualcomm Snapdragon 850, which is an impressive feat considering the dimensions of the Surface Pro X. The SQ1 processor also boasts of 3x more performance per watt compared to the 15W quad-core Intel Coffee Lake CPU found in the Surface Pro 6. How much of this becomes evident in real-world performance needs to be tested. While the SQ1 performs most tasks within a 7W TDP window, it can boost for a short period up to 15W if needed.
Qualcomm says that the graphics capabilities are exclusive to the SQ1 and won't be seen on the company's mainstream offerings. Another feature of the SQ1 is that it boasts of 9 TFLOPS of AI performance on a similar level as that of the Snapdragon 855. The AI can be used for a variety of tasks such as face/object recognition, optimizing battery performance, and more. It remains to be seen how day-to-day apps would make use of this, though.
While the SQ1 is a great collaborative effort between Microsoft and Qualcomm, there path to the Surface Pro X's success is not thorn-free. Windows 10 on ARM (WoA) had several lacunae that hindered its growth even with support for apps recompiled for ARM64 and other PC OEMs such as Asus, Lenovo, and HP releasing WoA devices in the recent past.
Moreover, the Adreno 685 supports only the DirectX 12 API so the vast library of OpenGL games that are already available for ARM devices is out of the equation. However, Microsoft is looking at the future of emerging web experiences and the ability to run two 4K displays simultaneously as potential use cases for this kind of GPU horsepower.