It is hard to imagine that 'smoothness' has become the keyword for flagship mobile phone systems this year.

With the end of the Realme GT7 Pro launch event, this batch of mainstream Android manufacturers' flagship series have basically completed their iterations. Reviewing these new product launch events, we can notice that they are different from previous years, such as AI deeply integrated into every part of each system, with smoothness becoming the main topic.

(Image source: OPPO)

A few years ago, users' understanding of smoothness was still Stay in the level of 'the phone does not lag', but now 'smoothness' has become a key indicator affecting user experience.

But what exactly does 'smoothness' refer to? Were mobile phone systems not smooth in the previous two years? The intricacies here are actually more complex than we imagine.

What are we discussing when we talk about 'smoothness'?

Most users' impression of Android is often that it is not as smooth as iOS, but this is actually related to historical issues.

As an open-source system, Android often needs to be compatible with more hardware devices of different configurations, whether it is a model equipped with the contemporary Snapdragon 8 flagship chip or an entry-level thousand-yuan phone, they must be able to run the latest version of Android as the qualification standard, which is different from iOS. Google does not strictly restrict hardware and cannot have enough say in customized hardware, which has made Android progress slowly for a long time to be compatible with more devices.

Higher compatibility has led Android to choose the Hardware Abstraction Layer (HAL) for the adaptation of different devices, but this type of standardized abstract design is not as efficient as Apple's in system call efficiency, leading to inefficient resource scheduling, especially evident in scenarios with high performance requirements.

(Image source: Google)

In terms of multi-task management mechanisms, Android's preemptive multi-tasking based on the Linux kernel can support parallel running of background tasks, but this mechanism requires extremely high precision in resource allocation. When the device's hardware resources are insufficient or certain high-resource-consuming applications continue to run in the background, the smoothness of foreground applications will be affected.

For example, Android's OOM (Out of Memory) mechanism automatically closes processes with lower background priority when system memory is insufficient, releasing resources to ensure smooth operation of foreground tasks. When certain applications frequently restart in the background, frequent triggering of the OOM mechanism not only fails to truly solve the problem of insufficient memory but also brings additional burden to system resource management, causing more severe lagging phenomena.

When it comes to memory management mechanisms, it is impossible not to mention Android's most controversial garbage collection mechanism. This mechanism increases CPU load while releasing memory. When running large applications, frequent garbage collection can cause obvious lagging, especially in scenarios with high real-time requirements. This garbage collection mechanism reduces system burden but also brings certain performance loss, becoming a major bottleneck in improving smoothness.

In addition, Android's memory management also shows fragmentation issues when facing large memory applications. The allocated memory cannot be efficiently utilized, resulting in a reduction in actual available memory space, further affecting the system's smoothness. For users, the decline in the smoothness of Android devices after using them for a period of time is often directly related to this memory fragmentation phenomenon.

(Image source: Qualcomm)

Although Android as a whole has many problematic issues, with the continuous improvement of hardware performance, many manufacturers can fully utilize the high-performance advantages of SoC, memory, and storage chips to compensate for these defects. For example, Android phones in recent years have basically started with 12GB of memory, and storage specifications have also entered the UFS 4.0 era early, which has brought optimizations for improving the system's multi-task management and fragmentation performance loss.

Relying solely on hardware stacking, phones can indeed ensure 'no lagging', but there is still an insurmountable gap between no lagging and smoothness. For example, OPPO ColorOS14 and vivo OriginOS 4 are considered by many users as not being 'natural' enough, or Xiaomi's Pengpai OS only provides 'advanced material' functions that support real-time animations and frosted glass effects for models with powerful performance. These are all clever tricks to solve the problem of lagging but are not the fundamental way to make it 'smooth'.

How to make Android smoother? A new battle for domestic mobile phones

Earlier, we mentioned the problems caused by Android due to its excessive openness and high compatibility, but it is precisely because of its openness that a group of customized Android systems have the space to rewrite the underlying code.

The first is to solve the problem of performance scheduling. Currently, Xiaomi's Pengpai OS, vivo's OriginOS, OPPO's ColorOS, and Honor's MagicOS all have self-developed performance optimization architectures to optimize the performance scheduling of the underlying system. Pengpai OS 2's brand-new self-developed micro-scheduling architecture can reduce CPU idle time by 19%, increase IPC by 16% in high-load scenarios, and reduce the maximum scheduling delay of key threads by 46%; OriginOS 5 has been upgraded to Unfair Scheduling 3.0, focusing on identifying the performance requirements of CPUs and GPUs in various scenarios to improve instantaneous response speed and follow-up rate through resource tilt . OPPO's ColorOS 15 and Honor's MagicOS 9.0 also have the Tidal Architecture and Turbo X, respectively, for reasonable allocation of performance resources.

(Image source: vivo)

OriginOS can be considered as one of the earliest customized Androids to reorganize resource allocation from the system bottom layer. Before vivo introduced this architecture, most manufacturers were thinking about how to improve instantaneous performance to solve high-load scenarios such as multi-task switching, small windows, and opening applications. Unfair Scheduling essentially tilts performance resources, which can not only increase the utilization rate of CPU resources but also reduce high power consumption caused by excessive calls and empty power consumption.

Reconstructing the bottom layer of the system is actually only part of this major surgery. How to make these reasonably allocated idle resources do more work beneficial to the user experience is the focus of customized systems. This leads to the mention of: animations.

Animation is one of the souls of UI interaction. A key reason why iOS is sought after is its smooth animations. In this year's launch events, various manufacturers basically took out animations as upgrade highlights. For example, OPPO first launched the Aurora Engine on ColorOS 15, using this engine to achieve multiple interrupt animations, parallel animations, and real-time blur effects.

OriginOS 5 and MagicOS 9.0 also brought Atom Animation 5.0 and a brand-new animation engine, respectively, also bringing parallel animations and naturally connected animation processes. In addition to smooth animations, Honor's first Vulkan engine on MagicOS 9.0 is responsible for smooth loading. Loading speed has always been a long-standing problem on Android systems, such as system albums, files, etc. This scattered and information-rich content often cannot be previewed instantaneously, and the addition of the Vulkan engine improves the overall loading speed, allowing the Magic7 series to quickly complete the fast preview of tens of thousands of images, which was previously only possible on iOS.

(Image source: Lei Technology live footage)

Although customized systems can reorganize the underlying architecture and optimize animations, Android's high openness still poses some challenges. To give a simple example, the openness of the Android system gives users the freedom to install third-party applications, but it also brings the problem of uneven application quality. A large number of applications fail to follow system optimization standards and are not optimized for resource usage during development. When these applications continue to run in the background, they will occupy a large amount of system resources, exacerbating resource contention issues. Some developers have designed continuously running processes or background services to increase application retention rates, which, although increasing application activity, have a negative impact on system smoothness.

Therefore, in recent years, we have also seen that almost all manufacturers have begun to prevent some risky, unknown third-party apps from being installed by users, and there will also be clear risk notifications before installation. Xiaomi's Pengpai OS is relatively rigorous in this regard. It requires users to go through multiple verifications and repeatedly confirm risks before installing risky applications, which brings advantages to system security and resource usage optimization.

Currently, although the customized systems of various mobile phone manufacturers have different tuning directions, the general trend is the same. One is to add more details of performance allocation at the bottom layer to broaden the scenarios suitable for performance scheduling; the other is to add more animations to let users feel smooth in vision and touch.

AI and deep customization, the only two magic weapons for mobile phone OS to be 'smoother'

The reason why iOS can become a benchmark for smoothness is inseparable from Apple's full deployment in ecology, hardware, and software. For example, the iPhone can customize the SoC according to Apple's needs, such as cache, bandwidth, etc. Therefore, Apple can almost grasp every link in the smooth operation conditions of iOS.

Earlier, we mentioned that mobile phone manufacturers have already conducted in-depth 'modifications' to customized Android, so is there a way to make the SoC work for themselves? Of course.

At the X200 series launch event, vivo emphasized the in-depth cooperation with MediaTek. On the latest Dimensity 9400 mobile platform, vivo customized this chip according to its own needs, enabling it to perform stronger with OriginOS 5 while consuming less power than the previous generation. Not only that, vivo even added the imaging capabilities of the V2 chip to the customization of the Dimensity 9400, allowing the two new models, X200 and X200 Pro mini, to still exert the same computing power as the V2 chip without an independent imaging chip.

(Image source: Lei Technology live footage)

In addition to cooperating with MediaTek, vivo also announced the establishment of a laboratory with Arm to delve into the hardware bottom layer for SoC tuning and optimization. Of course, the official has not yet disclosed some commercial-grade results of the cooperation between the two parties, and there may be some unexpected surprises on the X300 series.

Of course, cooperating with chip manufacturers is indeed an important link, but considering the many models of different positions and directions in the Android camp, if only one chip is customized, the optimization effect for all models is still quite limited. Honor's solution is much more direct: AI.

MagicOS 9.0 first launched the AI system scheduling engine, converting users' daily usage scenarios into key optimization scenarios through intention recognition and proactive human-computer interaction. It can provide different performance scheduling to meet users' actual needs in different demands, usage locations, and times.

(Image source: Lei Technology live footage)

The advantage of this solution is that its cost is lower, and the benefited models can cover from flagship phones to thousand-yuan phones, and mobile phones can also perform performance scheduling more in line with users' actual needs, ensuring smoothness without affecting power consumption. It has to be said that in the era of AI fully supporting mobile systems, using AI to predict user behavior is already the general trend. Even Apple has quietly added an automatic game mode to iOS 18, showing the role of artificial intelligence in system bottom-layer optimization.

In the past few years, the huge advantage of the iPhone has been its smoothness, which does not only stay in the release of extreme performance but also maintains a silky experience in daily operations. Nowadays, Android manufacturers have already surpassed the iPhone in terms of performance, imaging, and battery life, and the only part left to optimize is system smoothness. And the new versions of various manufacturers this year have indeed achieved layer-by-layer optimization from the bottom layer to vision, with a qualitative change in experience compared to the past. It can be foreseen that as Apple's last fig leaf is torn off, the mobile phone market landscape will also usher in more variables.

Source: Lei Technology