One of the biggest players on the market of processors for smartphones are considered Huawei, Samsung and Qualcomm. Qualcomm is the supplier of processors for most smartphone manufacturers, Samsung and Huawei make their own chipsets. Is it possible to call among all the products of these brands one processor, which of course is the best?
To begin to understand what processors?
Processors smartphones are always delivered in the form of single-chip systems (SoC). This means that several types of processors are laser the mold in one piece of silicon, and the technology on which this is done significantly affects the operating speed of the smartphone. When the description of a chipset we see that it is made, for example, for 12-nm technology, this means that the thickness of the wirings is 12Нм. The smaller this number, the thinner the track, the less resistance and faster signal. Less, respectively, and heat, and hence wasteful energy consumption. So matter how surprising or ridiculous it may sound, but on the clock speed and number of cores of the Central processor (CPU) you can not judge how fast the smartphone will work when solving specific problems.
But back to SoC: it consists of CPU, graphics coprocessor, LTE modem, a processor for multimedia, security, digital signal processor, accelerator AI and additional elements to ensure data exchange via Bluetooth and WiFi. Such a variety and diversity of functions greatly complicates the problem of choice and does not allow us to determine
Best processor for Android smartphones, released in the first half of 2018, according to most market analysts:
Almost all smartphones use ARM architecture for their processors. ARM provides the CPU with different performance, which are based on various combinations of single crystals. The trick is in the layout and synchronization of components. The situation is similar with GPUs. Qualcomm uses Adreno developing their own accelerators. Samsung and Huawei, in turn, prefer units from ARM Mali. And those and others involved in Android smartphones through a platform independent interface to OpenGL in different versions. The latest models of smartphones is implemented and support the technology Vulkan API. From the point of view of game developers, it facilitates communication with a graphics processor that makes the chips largely interchangeable, and their playing characteristics shown in the testing process, comparable. The difference between the slow and the fast graphics processor in Android most noticeable when you start playing a newly released game on a really outdated smartphone. Unlike a Windows PC, the game is not retarding and goes in spurts. All quite smooth, only the parts are shown on the display is smaller – the picture is noticeably poorer.
Qualcomm Snapdragon 845
New Snapdragon 845, released in 2018, has demonstrated a major upgrade of the graphics, and here, indeed, were an important virtual reality. Optimized resolution and speed of imaging. There are six degrees of freedom motion tracking without any additional external hardware. The chipset can be used not only in smartphones, but in a standalone VR headsets. Snapdragon 845 does not need to clean the chip AI. Unlike Huawei, Qualcomm still does not rely on a dedicated processor AI for their chipsets. Digital signal processor Hexagon DSP 685 performs specific tasks for the AI, but at the same time works as a Sensor Hub-and collecting and processing sensor data without intervention from the main processor.
Samsung Exynos 9810
New Exynos 9810, which is set in the Galaxy S9, intended for the market in European countries (in the US the device is equipped with Snapdragon 845), the received biometric update, and optimized 3D face recognition. so far, neither Apple nor Samsung are unable to fully satisfy their customers. In the 9810 instead of the Exynos GPU Mali-MP20 applied G71 graphics processor Mali-G72 MP18, which has two cores less. But the clock speed of the new adapter to 30 MHz higher than last year’s model, which allows to compensate for loss of performance. Samsung takes a different approach to design than Huawei or Qualcomm, and does not require any additional module to handle the task AI. Due to the lack of benchmarks for AI and deep learning machine it is difficult to say how justified this solution.
Huawei / HiSilicon Kirin 970
Simultaneously with the flagship smartphone Mate 10 Pro, Huawei unveiled its first chipset coprocessor AI accelerator. If you look at the advertising company of this smartphone, it is immediately clear that the brand made the main rate on artificial intelligence. Chipset almost nothing has improved. However, this task was not intended as Kirin 960 used in Mate 9, has left competitors behind with its excellent graphics performance and speed. Now with the new module neural processing, Huawei should definitely raise the bar to a new level. “Huawei-NPU” is still rarely used. People are betting on the fact that Android is adaptable, and not optimized apps will work with NPU, not the CPU. And now, when working in tandem with neural coprocessor can be expected to increase approximately 25 times the performance and up to 50 times — for energy efficiency. At the very least, these results were obtained when comparing heterogeneous complex Kirin 970 with a normal Quad core CPU Cortex-A73.
We have reviewed the best processors for smartphones in 2018, but did not answer the question: is there among them one is the fastest? All these chipsets have strengths in different areas, so it’s hard to say which CPU is best for a smartphone. Qualcomm innovates, especially in graphics, and preparing for independent VR. At the same time, the brand develops the segment of AI assistants that can work offline, i.e. without an Internet connection. Samsung has focused on more efficient and hopefully fans of the brand, more accurate 3D face detection. Huawei is not going to launch a new chipset for six months after Kirin 970 and hopes that his investment in AI was not in vain. Meanwhile, the user still faces a problem of choice. And what processor is your smartphone? How quick is he? Is there enough productive? Share your opinion in the comments, and perhaps your information will help someone make the right decision.