The current development of technologies and industries such as the Internet and the Internet of Things has brought us into a preliminary digital world. We believe everyone feels the same way. Every aspect of daily work, life, entertainment, etc. has been largely built on digital technology.
So, what exactly do these digital technologies include? In addition to well-known communication technologies such as 5G and Wi-Fi, there is actually a key technology that supports our digital entertainment life: GPU.
Especially in the context of the current wave of mobility sweeping everything, what kind of development will the mobile GPU usher in is also one of the most important topics in the technology industry.
Mobile GPU is not a choice, both performance and energy efficiency are required
As we all know, the GPU is different from the CPU. The CPU has a small number of cores but has enough digital and logic operation units, so it is good at general digital logic operations. The GPU has a large number of cores, and each core has a relatively small cache, which is very good at large-scale data-parallel computing. And graphics computing is this type of computing, so it is naturally suitable for GPU to complete.
Specifically, ray tracing can improve the realistic quality of rendered images from three aspects.
The first is soft shadows, that is, letting the object produce strong and weak shadow effects according to the intensity of the light and the distance of the light source so that the shadows are more in line with the real performance. There are also reflection effects, such as the effect of adding reflections on the surface of the water and smooth objects, which can also make the look and feel closer to reality. Finally, global illumination is added, that is, the combination and interaction of direct light sources and indirect light sources can be delicately displayed, thereby further enhancing the authenticity of the environment.
To do this, a huge amount of computation is required. On the desktop side, Nvidia successfully brought ray tracing to the desktop game market in the Turing GPU architecture in 2018, but at the same time, it was a larger cooling module. It's obviously not possible to do this on mobile.
So how should mobile light tracking be implemented? At present, there are two directions. One is to achieve a ray-tracing effect based on software solutions, and the other is to achieve real-time ray-tracing through hardware-level acceleration. Only hardware-based ray tracing can achieve the best balance between performance and power consumption.
In addition to improving performance, increasing efficiency for mobile GPUs is also a top priority, but this is also a major difficulty in the development and evolution of mobile GPUs.
Improving the overall energy efficiency of mobile GPUs, is a systematic task that requires full-link optimization from the bottom layer of the chip to application software and hardware to optimize and control the overall power consumption of the system.
Following this idea, we believe that the efficiency enhancement scheme of mobile GPU can be divided into "three axes", namely, the adaptation of the underlying software and hardware, the system-level in-depth optimization of AI and other technologies, and the extensive cooperation at the ecological level.
The so-called adaptation of the underlying software and hardware includes the evolution of the GPU chip level and the optimization of energy efficiency by software such as graphics computing interfaces and standards.
The system-level optimization scheme is to introduce the corresponding platform technology on top of this to further reduce the load for GPU operations. Among them, the introduction of AI technology is very important, especially for those with independent AI processing modules. In this case, the workload of the GPU can be greatly reduced through heterogeneous synchronous operations. This is also an important trend in the development of mobile GPUs.
Finally, there is cooperation optimization at the ecological level, that is, when the mobile GPU is specifically used in games and other scenarios, it is optimized jointly with terminal manufacturers and content providers. This is equivalent to the last step in increasing efficiency, and it is often the easiest for ordinary consumers to perceive. to the step.
In a word, with the continuous improvement of the refresh rate and resolution of mobile terminals, the complexity of content continues to increase, and even the main line of large-scale and high-load graphics computing requirements in the future metaverse era has proposed higher performance and energy efficiency of mobile GPUs. Requirements, for mobile GPUs, it is necessary to constantly pursue the ultimate balance of the two in order to meet the needs of the highly digital entertainment era in the future. We believe this is also the main tone of the next mobile GPU development.
MediaTek took the lead, and the mobile GPU efficiency solution came into being
After understanding the main theme of mobile GPU development in the future, I believe many friends can't help but ask, how is the industry progressing along the way?
As mentioned earlier, MediaTek talked about mobile GPUs at this year's MediaTek Dimensity flagship technical communication meeting. For a long time, MediaTek's exploration of the development direction of mobile GPUs has been very active and forward-looking, and to a certain extent can also represent the most cutting-edge research process in the industry.
For example, in terms of mobile ray tracing, MediaTek officially released the mobile ray tracing SDK solution in October last year. At present, MediaTek's top Dimensity 9000 mobile platform, based on the Mail-G710 GPU, realizes the mobile terminal light-chasing graphics rendering technology, which can effectively improve the quality of mobile games. Products like the OPPO Find X5 Pro Dimensity Edition have also made unique system-level applications like ray tracing wallpapers.
You know, the Vulkan interface standard organization only released the API that supports Vulkan Raytracing in January this year, and MediaTek has already laid out the Ray Query technical solution in December 2020, and its technological leadership and forward-looking can be seen.
Of course, the previous mobile light tracking was still based on software, but with the launch of Arm’s first mobile hardware light tracking GPU product, Immortalis-G715, in June this year, the real hardware-level mobile light tracking is close at hand. Immortalis-G715 has a 2x increase in floating point computing power and a 3x increase in output capability for triangular slices. At the same time, it also has an upgrade in bandwidth, which can fully handle complex and realistic light-chasing scene rendering.
MediaTek also cooperated with ARM for the first time. At this communication meeting, some media asked about the progress of the cooperation between MediaTek and Immortalis GPU. MediaTek's response was that "corresponding products will be launched soon, you can look forward to it for a while." This obviously makes people feel that MediaTek is equipped with hardware-level mobile light. The next-generation Dimensity flagship chip that is chasing has more expectations.
In the exploration of mobile GPU efficiency solutions, MediaTek is also at the forefront of the industry.
We analyzed earlier that the efficiency enhancement of mobile GPU is inseparable from the three directions of system-level optimization technologies such as underlying software and hardware, AI, and ecological cooperation. In fact, the specific layout of MediaTek is also closely related to these three directions.
For example, at the bottom of the GPU chip, MediaTek's Dimensity Ecological Laboratory and the Gold Label Alliance jointly promote the popularization of the 64-bit ecology, because in the 64-bit computing environment, mainstream chips can achieve 13.2% to 35.2% energy efficiency improvement, It can also increase the game frame rate by an average of 16%.
On the other hand, MediaTek always supports the latest Vulkan standard at the first time, because Vulkan can significantly improve the efficiency and performance of GPU compared with other graphics interface standards. Under MediaTek's deep optimization, the Vulkan performance of popular games can be improved by an average of 10%. Therefore, popularizing the Vulkan ecosystem is very meaningful for the efficiency of mobile GPUs.
Of course, what cannot be ignored is the annual progress iteration of the GPU chip's own process technology, which can also bring about 15% energy efficiency gains. MediaTek's next-generation Dimensity flagship will definitely adopt the most advanced manufacturing process.
On the basis of the underlying software and hardware optimization, MediaTek will also achieve further load reduction rendering through system tuning, GPU driver optimization, etc. This part will introduce AI-related technical capabilities.
However, what attracts more attention related to AI is the upper-level optimization, that is, some self-developed technologies in cooperation with ecological partners such as application content developers and terminal manufacturers.
For example, the Mobile Game AI-SR technology announced by MediaTek in October last year uses the heterogeneous solution of GPU and APU to ensure the same or even higher game picture quality. to reduce the workload of the GPU, thereby improving energy efficiency. At present, this technology has been applied to Vivo X80 series mobile phones in cooperation with terminal manufacturer Vivo.
In addition to game super-score, MediaTek also has adaptive scheduling technology, variable rate coloring, and other technologies, which have really improved the energy efficiency of mobile GPUs.
Speaking of cooperation at the ecological level, MediaTek also proposed to define a series of adaptive frameworks with content solution providers, such as Google's ADPF solution, ARM's Adaptive Performance framework, Tencent's TGPA, etc. With these adaptive frameworks, content Manufacturers can receive real-time information from systems and chips to perform real-time load adjustment, which has direct benefits in reducing GPU load.
It can be seen that with the joint efforts of mobile chip platform solution providers such as MediaTek and the participants of the entire mobile digital industry, mobile GPUs are making steady progress in the general direction of "speeding up and increasing efficiency". Every step of our improvement in the digital entertainment experience on the mobile terminal makes people look forward to the wonderful future that mobile GPUs can "render" for us.
Of course, in the future, if we want to realize the digital transformation of the whole society, it is far from enough to rely on the upgrading and evolution of GPUs on the mobile side. It is a world that requires the deep integration of various technologies to build. At this MediaTek Dimensity flagship technical communication meeting, MediaTek also shared topics including AI image semantic segmentation, 5G new dual-pass, Wi-Fi 7, high-fidelity Bluetooth audio, high-precision navigation and other topics to help mobile phone users fully enjoy the sky The full-scene experience upgrade brought by Ji flagship technology has also fully prepared for the future mobile digital life.
For world-class technology companies, the hard power of technology determines whether they can fly high, and the insight and prediction of market and user needs determine whether they can go far. Obviously, MediaTek has done a very good job in these two aspects. Years of deep cultivation in multiple fields and the original intention of improving user experience have built a solid barrier together. Today, MediaTek's next-generation Dimensity flagship chip, Dimensity 9200, is approaching. With the blessing of leading flagship technology, it may become a product that transforms the experience of a new generation of terminals, which is worth our common expectation.
0 Comments