On April 10th, at the "6th China Electronics Information Expo" held in Shenzhen, the vivo X21 won the Innovation Award. It is even more noteworthy that during the exhibition, Vivo together with Qualcomm held 'from AIE to Jovi vivo | Qualcomm. Media communication meeting' for deep revealing of artificial intelligence appearing on vivo X21.
In the past few years, artificial intelligence has become a concept that sweeps across all technology industries. It also requires the cooperation of the cloud and the end-side. The cloud is responsible for the training of models. The end is mainly reasoning and computing. However, it is too dependent on the artificial intelligence in the cloud. It also gave birth to some questions: If every implementation of artificial intelligence must rely on the cloud, is the data upload secure enough? Second, does the cloud need a stable network, can the low latency and reliability be met? Finally, the entire system is completely dependent on the network. For the end of the use of computing resources is not economical.
We must consider such a problem. In the past, the dependence of the artificial intelligence on the cloud was mainly limited to the artificial intelligence of the terminal side, which was the so-called 'painful and happy'. Nowadays, chip manufacturers such as Qualcomm Xiaolong provide powerful chips and mobile phones. The manufacturer vivo completes the application's landing, so that the artificial intelligence can be rooted in the end.
What exactly is Qualcomm's AIE?
In the past, the introduction of mobile phone products referred to the processor will always be Qualcomm Xiaolong ***, but this time the suffix is also the highest in history --- Qualcomm Xiaolong multi-core neural network acceleration chip 660 AIE, the Chinese word can still be partially understood, What exactly is AIE? In this communication meeting, Qualcomm also gave the answer:
The so-called AIE artificial intelligence engine is a collection concept, which includes three major parts, one part is hardware, including CPU, GPU and DSP (Hexagon vector processor), is responsible for FP32, FP16, INT8 network composition and operation; software includes Xiaolong neural processing SDK, Android NN and Hexagon NN (the DSP part of Qualcomm SoC); and the framework of the third part, which is the design language of AI algorithm, including FaceBook's Caffee/Caffee2, Google's TensorFlow/TensorFlow Lite , ONNX.
It can be seen from the structure here that, unlike Huawei Unicorn processors that use separate NPUs to deal with artificial intelligence tasks, Qualcomm Snapdragon AIE releases AI from individual CPU calculations to provide powerful AI computing capabilities. Multiple AI models can be loaded into memory at the same time. Its compatibility is excellent, coverage system, image, sound service and other aspects. And CPU, GPU and DSP support different floating-point operations, using their respective advantages to meet different artificial intelligence processing Gathering the strengths of each family is superior to waiting for the NPU to complete when compared to all artificial intelligence queues.
On the other hand, we all know that this time Qualcomm itself is also emphasizing the AIE of its own chip. In fact, this is not Qualcomm's first-generation product. Even this time can be extended to 2 years ago. The key core of the Pao Nerve Processing SDK has been developed. The third generation, so the technical precipitation is ahead of the current industry level, which provides an excellent foundation for the development of artificial intelligence.
In addition, Qualcomm Snapdragon processors are widely used in the flagship products of major mobile phone manufacturers, from high-end coverage to low-end. At the same time, Qualcomm officials also said that the construction of the ecosystem partners and The in-depth cooperation of cloud service providers are all leading companies in the industry. A good foundation for cooperation achieves tremendous support in terms of services. This attitude of attracting firewood allows AI applications to be maximally satisfied. This is also the case of Huawei and Apple today. Home, despite having powerful artificial intelligence, because of their different degree of closure, the application of artificial intelligence is not as fast as expected. After the release of Huawei Mate 10, the application of artificial intelligence did not show a big step. Forward gesture.
Regarding the future, Qualcomm Xiaolong has its own technological advantages and built on an ecological basis. Artificial intelligence services will be further enhanced. Mobile phones are currently transitioning from smart phones to smart phones. Artificial intelligence built on machine learning has a certain degree of Limitations, but in the future under the guidance of mobile phone manufacturers such as Qualcomm, Vivo, etc., mobile phones will start to be able to read, understand, and feel the status of users. Providing the most suitable services for each scenario is undoubtedly the future direction.
Has laid a good foundation, Vivo how to artificial intelligence landing it?
Jovi= enjoy vivo's AI is the name of vivo's interpretation of the artificial intelligence on the X21. As the world’s top five smartphone manufacturers, Vivo has a large user base worldwide, and cooperation with Qualcomm can be described as It is worth noting that how to place artificial intelligence applications into the user's daily life. At the beginning of the conference, Vivian AI’s senior product manager Hong Jiehong also referred to vivo understanding. Artificial Intelligence Application:
Handle massive data, identify these data as real use scenarios for users, and provide the best services to consumers based on these usage scenarios.
And hope that through artificial intelligence, let the mobile phone learn to learn and understand the process of user habits, turn the mobile phone from the original small intelligence into great wisdom, break the consumer's habits of using mobile phones, bring them extraordinary experience.
So we are going to have to understand what exactly Jovi AI is doing.
On the one hand, Jovi artificial intelligence is taking advantage of the fact that vivo itself is taking pictures, and continues to improve. Backlight photography and intellectual beauty have been completely upgraded. With AI technology, further intelligence has been achieved. For example, the AI scene camera is implemented. , trying to achieve 'real-life point-and-shoot' cameras through artificial intelligence. All kinds of complex parameter adjustments are fully handed to vivo, users only need to tap the camera button to take a good photo. In addition, the AI self-timer lighting effect is added. , Artificial intelligence developed since AR sprouting and so on, further enriched the daily use.
On the other hand, Vivo provides a AI-wisdom scene at the negative screen, through the "fast application" structure to achieve services for everyday life. For example, intelligent reminders of ball games, advance travel plans, these are all artificial by Jovi. Part of the intelligence, it brings the guidance of all aspects of life. These processes do not need to be manually entered, the phone will silently record it all in the background.
The last aspect, namely the AI intelligence engine and AI game engine implemented by Jovi artificial intelligence, brings more 'hiddenness'. It may be difficult to visualize these services, but the mobile phone remains smooth and the game is switched between When we have absolutely no past reloading, we can understand the advantages brought by this kind of thing. The latter two can all be planned as part of a smart life. Of course, the function is not just that. It provides a full range of services. You can push it out automatically when you need it.
In a follow-up question and answer session, he talked about the concept of vivo for Jovi. Hong Jiehong also mentioned:
In fact, our goal for Jovi, so far we hope to create a one-stop scene experience with a mobile phone as the core, the mobile phone itself will have some sensor data, there will be some corresponding functions to consumers. At the same time we will also Obtaining cooperation with third-party software companies, such as FEI and other service providers, to provide accurate and useful data to consumers. These service providers also hope to provide information and services to consumers through mobile phone manufacturers, so that consumers do not need to search the Internet. information.
In addition, vivo is preparing to set up a data center to analyze the data and identify the scenes of these data to provide corresponding services. It is particularly worthwhile to affirm that Vivo is also trying to build its own account system to use artificial intelligence. The experience continues better on the next generation of mobile phones. This means that only one vivo account is required. Even if you change your mobile phone, you don’t have to worry about the original service being lost. The mobile phone is still in the position to “understand you most.” keeping it.
These layouts show that Vivo is trying to make the mobile phone more able to understand the user's habits, so as to understand the user's thoughts, until the understanding of the heart. From the original personal assistant, began to understand your friends, then to the indispensable 'lover So, from the current point of view, Vivi's Jovi artificial intelligence is still not saturated, and it is worth looking forward to.
to sum up
This media communication meeting has now come to a successful conclusion, but the revolution on artificial intelligence is still in the ascendant. In the display of the strengths of the various 'show muscles', Vivo and Qualcomm have joined forces and pioneered efforts in the current mobile phone market. It is a unique existence. In fact, the performance of Jovi artificial intelligence is also enough to be loved. The improvement in the level of user experience has become the biggest bright spot in the future.
On April 10th, at the "6th China Electronics Information Expo" held in Shenzhen, the vivo X21 won the Innovation Award. It is even more noteworthy that during the exhibition, Vivo together with Qualcomm held 'from AIE to Jovi vivo | Qualcomm. Media communication meeting' for the deep reveal of artificial intelligence appearing on vivo X21.
In the past few years, artificial intelligence has become a concept that sweeps across all technology industries. It also requires the cooperation of the cloud and the end-side. The cloud is responsible for the training of models. The end is mainly reasoning and computing. However, it is too dependent on the artificial intelligence in the cloud. It also gave birth to some questions: If every implementation of artificial intelligence must rely on the cloud, is the data upload secure enough? Second, does the cloud need a stable network, can the low latency and reliability be met? Finally, the entire system is completely dependent on the network. For the end of the use of computing resources is not economical.
We must consider such a problem. In the past, the dependence of the artificial intelligence on the cloud was mainly limited to the artificial intelligence of the terminal side, which was the so-called 'painful and happy'. Nowadays, chip manufacturers such as Qualcomm Xiaolong provide powerful chips and mobile phones. The manufacturer vivo completes the application's landing, so that the artificial intelligence can be rooted in the end.
What exactly is Qualcomm's AIE?
In the past, the introduction of mobile phone products referred to the processor will always be Qualcomm Xiaolong ***, but this time the suffix is also reached the highest in history --- Qualcomm Xiaolong multi-core neural network acceleration chip 660 AIE, the Chinese word can still be partially understood, What exactly is AIE? In this communication meeting, Qualcomm also gave the answer:
The so-called AIE artificial intelligence engine is a collection concept, which includes three major parts, one part is hardware, including CPU, GPU and DSP (Hexagon vector processor), is responsible for FP32, FP16, INT8 network composition and operation; software includes Xiaolong neural processing SDK, Android NN and Hexagon NN (the DSP part of Qualcomm SoC); and the framework of the third part, which is the design language of AI algorithm, including FaceBook's Caffee/Caffee2, Google's TensorFlow/TensorFlow Lite , ONNX.
From this structure, we can also see that, unlike Huawei unicorn processor using a separate NPU to deal with artificial intelligence tasks, Qualcomm Snapdragon AIE releases AI from separate CPU calculations, providing powerful AI computing capabilities, and Multiple AI models can be loaded into memory at the same time, with excellent compatibility, coverage system, image, and sound services. In addition, CPUs, GPUs, and DSPs support different floating-point operations and use their respective advantages to satisfy different artificial intelligence processes. Gathering the strengths of each family is superior to waiting for the NPU to complete when compared to all artificial intelligence queues.
On the other hand, we all know that this time Qualcomm itself is also emphasizing the AIE of its own chip. In fact, this is not Qualcomm's first-generation product. Even this time can be extended to 2 years ago. The key core of the Xiaolong neural processing SDK has been developed to The third generation, so the technical precipitation is ahead of the current industry level, which provides an excellent foundation for the development of artificial intelligence.
In addition, Qualcomm Snapdragon processors are widely used in the flagship products of major mobile phone manufacturers, from high-end coverage to low-end. At the same time, Qualcomm officials also stated that the established ecosystem partners and The in-depth cooperation of cloud service providers are all leading companies in the industry. A good foundation for cooperation achieves tremendous support in terms of services. This attitude of attracting firewood enables AI applications to be maximally satisfied. This is also the case for Huawei and Apple today. Home, despite having powerful artificial intelligence, because of their different degree of closure, the application of artificial intelligence is not as fast as expected. After the release of Huawei Mate 10, the application of artificial intelligence did not show a big step. Forward gesture.
Regarding the future, Qualcomm Xiaolong has its own technological advantages and built on an ecological basis. Artificial intelligence services will be further enhanced. Mobile phones are currently transitioning from smart phones to smart phones. Artificial intelligence built on machine learning has a certain degree of Limitations, but in the future under the guidance of mobile phone manufacturers such as Qualcomm, Vivo, etc., mobile phones will begin to be able to read, understand, and feel the status of users. Providing the most suitable services for each scenario is undoubtedly the future direction.
Has laid a good foundation, Vivo how to artificial intelligence landing it?
Jovi= enjoy vivo's AI, is the vivo's naming interpretation of the artificial intelligence carried on the X21. As the world's top five smartphone manufacturers, Vivo has a huge user base worldwide, and cooperation with Qualcomm can be described as It is worth noting that how to place artificial intelligence applications into the user's daily life. At the beginning of the conference, Vivian AI’s Senior Product Manager Hong Jiehong also referred to vivo understanding. Artificial Intelligence Application:
Handle massive data, identify these data as real use scenarios for users, and provide the best services to consumers based on these usage scenarios.
And hope that through artificial intelligence, let the mobile phone learn to learn and understand the process of user habits, turn the mobile phone from the original small intelligence into great wisdom, break the consumer's habits of using mobile phones, bring them extraordinary experience.
So we are going to have to understand what exactly Jovi AI is doing.
On the one hand, Jovi artificial intelligence is taking advantage of the fact that vivo itself is taking pictures to continue to improve. Backlight photography and intellectual beauty have been newly upgraded, and AI technology has completed further intelligentization. For example, the AI scene camera implemented , trying to achieve 'real-life point-and-shoot' cameras through artificial intelligence. All kinds of complex parameter adjustments are fully handed to vivo, users only need to tap the camera button to take a good photo. In addition, the AI self-timer photo effect is added. , Artificial intelligence developed since AR sprouting and so on, further enriched daily use.
On the other hand, Vivo provides a AI-wisdom scene at the negative screen, through the "fast application" structure to achieve services for everyday life. For example, intelligent reminders of ball games, advance travel plans, these are all artificial by Jovi. Part of the intelligence, it brings the guidance of all aspects of life. These processes do not require manual input, the phone will silently record all this in the background.
The last aspect, namely the AI intelligence engine and the AI game engine implemented by Jovi artificial intelligence, brings more 'concealedness'. It may be difficult to visualize these services, but the mobile phone remains smooth and the game is switched between When we have absolutely no past reloading, we can understand the advantages brought by this kind of thing. The latter two can all be planned as part of a smart life. Of course, the function is not just that. It provides a full range of services. You can push it out automatically when you need it.
In a follow-up question and answer session, he talked about the concept of vivo for Jovi. Hong Jiehong also mentioned:
In fact, our goal for Jovi, so far we hope to create a one-stop scene experience with a mobile phone as the core, the mobile phone itself will have some sensor data, there will be some corresponding functions to consumers. At the same time we will also Obtaining cooperation with third-party software companies, such as FEI and other service providers, to provide accurate and useful data to consumers. These service providers also hope to provide information and services to consumers via mobile phone manufacturers, so that consumers do not need to search the Internet. information.
In addition, vivo is preparing to set up a data center to analyze the data and identify the scenes of these data to provide corresponding services. It is particularly worthwhile to affirm that Vivo is also trying to build its own account system to use artificial intelligence. The experience continues better on the next generation of mobile phones. This means that only one vivo account is required. Even if you change your mobile phone, you don’t have to worry about the original service being lost. The mobile phone is still in the “Know Your Way” attitude. keeping it.
These layouts show that Vivo is trying to make the mobile phone more able to understand the user's usage habits, so as to understand the user's thoughts, until they understand the heart. From the original personal assistant, began to understand your friends, then to the indispensable 'lover Therefore, from the current point of view, vivo's Jovi artificial intelligence is still not saturated, and it is worth looking forward to.
to sum up
This media communication meeting has now come to a successful conclusion, but the revolution on artificial intelligence is just in the ascendant. In the display of the strengths of the various 'show muscles', Vivo and Qualcomm have joined forces and pioneered efforts in the current mobile phone market. It is a unique existence. In fact, the performance of Jovi artificial intelligence is also enough to be loved. The improvement in the level of user experience has become the biggest bright spot in the future.
