1. MediaTek Q2 shipments reached 100 million sets, which is better than the first half of the second half;
According to the micro-network news, since the second quarter, MediaTek has benefited from the shipment of mobile chip market. The operation is expected to grow significantly compared with the first quarter. The total mobile chip shipments will fall between 90 million and 100 million sets. The exact number will be It was announced during the second quarter of the law conference. CEO Cai Lixing had previously said that 'should be able to reach the market', and at the same time optimistic about MediaTek's weaker season effect this year is better than the past, the second half will be better than the first half.
It is reported that the mainland smartphone market continues to be in a downturn. Foreign investors pointed out that in the terminal market is not as expected, MediaTek P60's output in the second half of the year is expected to fall by 20~30%, which makes the average shipment target of smartphone chips overweight this year. Third-quarter sales may increase by only 10% quarter-on-quarter, below the market consensus of 20%.
2. MediaTek P60 drove Q2 revenue up 20%, Q3 growth and pressure;
European foreign investment in the latest research report issued by MediaTek (2454) pointed out that the third quarter mobile chip shipments due to OPPO will allocate some orders to Qualcomm, coupled with the higher base period in the second quarter, it is likely to suppress the growth of MediaTek's operation, but long-term Look, MediaTek's non-mobile business, its active layout of 5G and other related business opportunities, can play one of the important sports of the future, the target price from 400 yuan to 350 yuan, rating is better than the broader market.
European-funded foreign investment in the latest research report issued by MediaTek pointed out that MediaTek's second quarter benefited from the mainland brand smart machine for the P60, the revenue growth is expected to see 12% to 20%, the gross profit margin is also 38.8%, but In the second half of the year, the growth of mainland smartphones slowed down, coupled with the higher base period in the second quarter, and OPPO allocated some orders to Qualcomm, in order to achieve dual supply, further limiting MediaTek's growth space. Plus Qualcomm is also pricing In terms of maintaining a positive attitude, it will offset the growth of MediaTek's non-mobile phone products. It is expected that the growth rate of single-season revenue will fall to 11%.
In the non-mobile chip business, due to the growth of voice assistant, WiFi, ASIC (customized chip) and power manager, MediaTek is expected to remain stable in non-mobile services, and is expected to achieve at least 10~15% in 2018. Growth, MediaTek has said that the overall voice assistant market may double compared to last year, the overall market size of 60 million units, and MediaTek has a good appeal on the Google, Amazon and Alibaba platforms, a major development niche , the target price was reduced from 400 yuan to 350 yuan, the rating is better than the broader market.
In the long-term, European foreign investment said that MediaTek's mobile chip business should show moderate growth in 2019, but in view of the 5G business transfer in 2020, MediaTek is accelerating the development of 5G, in an attempt to achieve a more competitive cost structure. Entering the market will bring better opportunities in non-mobile businesses such as the Internet of Things and smart homes. Not only that, but the automotive sector and NB-IoT will continue to grow. Economic Daily
3. Why is the C-V2X better than the competition for car networking technology?
Set micro-network news (Reporter Ai Meng) In the competition for the Internet of Vehicles technology, C-V2X has a sense of coming up. Let's take a look at this 'rhythm': Not long ago, Ford, Panasonic, Qualcomm and Colorado Department of Transportation released The plan to deploy C-V2X on the road was selected at the CityNOW project headquarters in Panasonic, Denver. The "ConVeX" Alliance, which was established last year, is launching the first motorcycle, car and infrastructure in Europe. -V2X direct communication interoperability live demonstration, alliance members including Audi, Ducati, Ericsson, SWARCO, Kaiserslautern University of Technology and Qualcomm, etc. These major initiatives reflect the upstream and downstream of the industry chain in accelerating C-V2X applications Cooperation in cooperation.
Full of stamina
To achieve the interconnection of the car and everything, there must be a unified standard, which shows the importance of the V2V standard. The V2X industry is divided into DSRC (dedicated short-range wireless communication standard) and C-V2X two camps. DSRC to IEEE 802.11 Based on p, V2V is its main application method. After more than ten years of development history, it has formed a traditional automotive electronics industry chain based on NXP, ST, Renesas, and automobile companies. C-V2X is based on LTE-R14 technology. Through the two major technologies of LTE-VD and LTE-V-Cell, the company has formed a communication industry chain enterprise mainly based on Huawei and Qualcomm, and an industry camp dominated by telecom operators and automobile companies.
The advantage of DSRC is 'early birth'. DSRC has been confirmed by the US Department of Transportation as the V2V standard, the EU's collaborative intelligent transportation system and Japan's V2X are also based on DSRC. Although DSRC has the support of the United States and the European Union, the strength of C-V2X is also Not to be underestimated, C-V2X has gained support from a wide range of automotive ecosystems, including the fast-growing global industry organization, the 5G Automotive Alliance. In addition, Huawei, Qualcomm, Ericsson, Intel, Nokia, etc. are also actively promoting C-V2X chips. And equipment industrialization, Audi, Toyota, SAIC, Changan, Dongfeng and other car companies have jointly conducted C-V2X technology testing. C-V2X continues to accelerate in the industrialization process.
From the technical point of view, C-V2X has obvious advantages, is suitable for more complex security application scenarios, meets low latency, high reliability and meets bandwidth requirements, is more secure, efficient and comprehensive than DSRC. With the previous year 3GPP first edition The development of C-V2X standardization specification is completed, and the development of C-V2X technology and related ecological chain is faster.
In addition to automakers, traffic management and infrastructure providers are now launching more and more C-V2X-related initiatives to equip road infrastructure (such as traffic signals, road signs, etc.) with more C-V2X-based roadsides. Units (RSUs), for their successful deployment.
The two relative ratios are different. In the future, the global V2X market may have a DSRC or C-V2X technology, or there may be DSRC and C-V2X at the same time, but the C-V2X's stamina is obviously more.
China's acceleration
As the world's largest auto market, China's C-V2X should be the first choice for domestic V2X technology standards, from the perspective of the world's largest LTE network status and C-V2X evolution.
Since it is a multiple-choice question, China is also spared no effort at this level. In the C-V2X standard, China is making efforts to strengthen the right to speak, and in 3GPP has led the development of some C-V2X standards and the study of subsequent evolution technologies. In 2017 At the 49th Plenary Session of ISO TC 204 held in April, China's C-V2X standard project application was approved, and C-V2X was identified as a candidate for the ISO ITS system. In September 2017, the second phase C- was completed. The V2X standard is released.
At the level of spectrum research, at the end of 2016, the Ministry of Industry and Information Technology officially divided 5905-5925MHz for C-V2X technology test, and conducted testing and experimental verification through Beijing-Baoding, Chongqing, Zhejiang and other vehicle networking demonstration zones. The stage has been completed. The “Guide to the Construction of the National Vehicle Network Industry Standard System”, which was recently introduced, is planned to be 2025. After China has formed an intelligent network standard system capable of supporting high-level automatic driving, the Ministry of Industry and Information Technology Networked direct communication uses the 5905-5925MHz frequency band to publicize for public comment. The introduction of these measures means that the Chinese government has taken a big step forward in promoting the policy and implementation conditions of C-V2X.
The industrialization level is also in force. Huawei launched the prototype of the vehicle terminal supporting C-V2X in 2016, and recently released the first RSU (roadside unit) that supports Uu+PC5 concurrency. Another major is Datang. - The first phase of V2X technology - LTE-V technology has carried out a lot of development work, and the industry's first chip-scale LTE-V pre-commercial products.
According to the current industrial development, China's C-V2X is expected to achieve scale pilot or commercial trials in 2018.
Advance to automatic driving
As the industry enters the world of 5G and autonomous driving, the C-V2X is also ready.
The C-V2X has a clear 5G evolution path and supports forward and backward compatibility. In addition to its superior performance and cost-effectiveness, the Rel-14 C-V2X is also the future of the 5G NR-based Rel-16 C-V2X. Driving technology paved the way.
Compared to the Rel-14 for basic safety communication between vehicles, the C-V2X based on the new 5G air interface provides high throughput, wideband carrier support, ultra-low latency and high reliability, making the autonomous vehicle effective. Sharing information on roads, road conditions and surrounding conditions with other vehicles and road infrastructure. In addition, vehicles can use the 5G NR-based C-V2X to transmit intent and planned travel routes to other vehicles with greater predictability and Accuracy, can further optimize the route planning of autonomous driving.
Therefore, the upcoming 5G commercialization feast will make the C-V2X play an important role in the future of autonomous driving. It is important to follow up in time instead of swinging around. (Proofreading/Yachuan) 4. Is the hardware industry still in the golden age?
If your industry has a history of 60 years, the public's perception may be that it has come to 'mature age' and is about to enter the next year, just like the semiconductor industry... but now more and more saying, The hardware has quietly re-emerged as a market trend.
At the Young Under 40 Innovators Award Panel during the annual Design Automation Conference (DAC 2018) in San Francisco, USA at the end of June, the platform architect from Google, Li Wei (Sheng Li) said to the audience: 'I think this is the golden age of hardware people.'
Another winner of the 'Innovator under 40', Arijit Raychowdhury, associate professor of Georgia Tech, also echoed Li Wei, saying: 'I think it is a good time for hardware engineers, there are many opportunities.'
As a member of the hardware IC industry, they should all agree with the above statement, believing that the actual situation is beneficial to them; the explosive growth of data centers and emerging AI applications depends on the development of faster, more powerful, power consumption by chip design engineers. It has a lower processor and can be built on different computing architectures.
However, the author is skeptical and believes that this 'hardware trend reappears' need to be further examined. As the moderator of the forum, the author does not believe that these 'innovators under 40' award winners are optimistic about the hardware industry. The view is fraud, and no one says that the semiconductor industry is returning to the glorious era of doing business as usual.
We can all feel the dramatic changes in this industry, but I don't think that hardware people can master the next step well. So I asked every young engineer and scholar on the stage, in their respective fields, 'innovation bottlenecks' Where is it?
David Atienza, an associate professor at the Ecole polytechnique federale de Lausanne (EPFL) in Switzerland, pointed out that the recent boom in artificial intelligence (AI) and machine learning has caused him to worry about a very large gap between the hardware sector and the field, because the former It is a large amount of software that focuses on solving all aspects of AI, but it is still based on a very traditional architecture at the hardware level.
Atienza's question is that AI software can continue to evolve, but 'what should we do with hardware?' He pointed out that accelerators can be developed, but in terms of 'applications require hardware to do' and 'hardware can handle things' There is still a very large gap.
And we obviously need to rely on EDA tools to design new accelerators, but do those tools really understand the system requirements? Atienza bluntly: 'To survive, the EDA industry has to move to a higher level and must be able to support a full range of solutions.'
The hardware and software boundaries are becoming more and more blurred and the hardware design community is uneasy. The abstraction layer is clearly disappearing. Li Wei, who worked at Intel Labs, pointed out: 'I have been working as a platform architect for my career. In the past, there was an abstraction layer that separated the software from the hardware structure, so both sides could operate independently; 'but not so.
His observations are: 'The abstraction layer has begun to collapse,' we have more and more data and more advanced algorithms that require more computing power; but Moore's Law has slowed down for a while: 'We can't provide so much computing power, or we can't improve computing power as fast as we used to,' making a huge gap between software and hardware architecture.
Li Wei’s view is, 'There will be no pure software or hardware architecture; 'but there will be 'architecture-driven apps' or 'application-aware architecture' (application-aware architecture) ).
EPFL's Atienza agrees with the above opinion. He said: 'It is true. We are seeing more and more types of education methods in the academic field. We don't want people to only do hardware or software. Everyone needs to be educated and trained for both. Can do. '
Raychowdhury, a former Georgia Tech school at Texas Instruments, observed: 'When I first entered the industry, we had an idea that we had to build a fixed-function accelerator, but after a long time, the situation was not so clear. We are not sure what kind of tasks or functions require accelerators, and we don't know the actual bottlenecks in the hardware field.'
As machine learning becomes a critical application, the goal of hardware design engineers is becoming clearer. 'We now know that even with GPUs, it takes a long time to train large networks; from a design perspective, people are looking for It's to get more high-bandwidth memory, and to build fixed-function components to speed up certain tasks - we now know that this can make a big difference, and it's reasonable; 'Raychowdhury optimistic: 'So I feel that there will be a lot of things happening in this field . '
He added: 'There are other things that I think are about to happen, and I see that the foundry industry is particularly active, is the advent of high-density memory -- SRAM, DRAM or embedded non-volatile memory, all kinds of Higher bandwidth technology. '
Raychowdhury believes that those memories will occur in the back end of line (BEOL), in 3D stacking or in a single 3D structure, at low temperatures; he expects dramatic changes in this area: 'Moore's Law is on the way Slow, the transistor size may not shrink any more, but I think there are other opportunities, especially in the latter stage, people will try to innovate.'
Machine Learning and EDA Challenges Yu Wang, an associate professor in the Department of Electronic Engineering at Tsinghua University in Beijing, and co-founder of DeePhi Tech in Beijing, said that everyone at the symposium is talking about the integration of applications and hardware components. , 'This is of course a trend, we have to link these two areas, and we think this is something we can do.'
But another big opportunity that Wang Yu sees is to use more than one machine to perform deep learning tasks; he explains that today's model is that we can build a machine that is very efficient in deep learning, but usually The resources of each machine are limited: 'If we operate these machines together?'
In his opinion, if those machines can work together, it may be a way to narrow the gap between the theoretical model and the actual situation. 'Today we have a host, and then there are a lot of agents... if we Get rid of the host and let all the agent machines work together. You don't need the host to monitor it all the time? It's like finding a person in the building, having multiple agent machines in the building, and trying to cover the entire area together.'
Wang Yu believes that many system-level innovations may be realized by more powerful agent machines: 'If we combine them, I foresee that there will be innovations in algorithm level, system level and hardware level, and there are many opportunities.'
Zhiru Zhang, an associate professor at Cornell Univ., echoed the views of other participants, co-founder of AutoESL, an EDA tool developer, which was later acquired by Xilinx; 'I am talking about Specialized computers are very excited, 'He believes that there will be a series of different accelerators coming out that will drive innovation.
Zhang pointed out: 'I think it is the best time to study such problems; but on the other hand, this will be the most challenging moment. 'What is the challenge? He explained: 'The accelerators will be built very much. Time consuming, and their programming can be quite difficult. '
As for revitalizing the semiconductor industry, Zhang believes that it is not to attract new talents, but to 'need to create more open hardware tool solutions'; he uses the open source online community GitHub as an example, 'If you want to expand some Machine learning research, as long as you can find a lot of machine learning related repost resources on GitHub.
But if you want to search for design automation-related forwarding resources on GitHub, there may be less than 20; Zhang believes that the lack of open tools is one of the biggest challenges facing hardware engineers today.
Compilation: Judith Chengeettaiwan (Reference: Hardware: Golden Age or Golden Ager?, by Junko Yoshida)5. Three-camera mobile phone, car electronics will drive CMOS sensor demand;
The timing entered the third quarter of 2018, but the demand for smart phones in the market is not as good as expected. Familiar with semiconductor manufacturers, it seems that the demand for mobile phone related chips is expected to maintain the basic disk. The growth space is still to be observed, but this year's noteworthy focus In order to redeem upgrade requirements, such as the previous use of dual-lens mobile phones, and will continue to advance to the three lenses, will drive the amount of CMOS sensing components can be improved. In addition, the most important next-generation automotive electronics market outside the mobile phone, a car The number of CIS components is more than 40. For the professional CIS packaging and testing factories such as Shengli, Jingyuan, Tongxin, and Sony CIS components, the company still has operations in the second half of the year. It is expected to look forward to it. The CMOS image sensor for vehicles has been first stepped under the popularization of Advanced Driver Assistance System (ADAS), and the demand continues to rise. In the future, the number of sensors will be improved in the era of autonomous vehicles. Need to lay out in advance, the global front-line powerhouse layout is also basically set, supply chain manufacturers believe that in the field of CMOS sensors, can master the Japanese lead Sony orders Business, the potential for growth in follow-up operations is the most promising. For example, the vehicle CMOS image sensor packaging and testing plant Shengli, recently announced the June revenue and the second quarter combined self-settlement, of which the combined revenue in June was 193 million yuan, compared with last year. In the first half of the year, Shengli Motor Co., Ltd. used CIS packaging and testing orders to maintain high-end. In the second quarter, the single-season self-consolidation and revenue of 595 million yuan has reached a record high in the quarterly consolidated revenue of Shengli CIS packaging and testing business. Compared with the same period of last year, it increased by 10.9%. In addition, the second quarter self-combined business income and pre-tax net profit increased by 22.4% and 51.8% respectively compared with the same period of last year. Familiar with CIS component manufacturers, Shengli has gradually focused after years of layout. In the first half of the year, Shengli's accumulated self-combined revenues reached 1.167 billion yuan, a growth rate of 12.5% compared with 1.037 billion in the same period of last year. Shipments also increased significantly compared with the same period of last year. Li's current business focuses on high-margin car and safety control products with strong demand, in which the proportion of vehicle products in revenue and shipments exceeds 60% of the total. Further promoted the first half of this year's accumulated self-sufficiency 272 million yuan, In the same period of last year, it increased by 28.7%, with a pre-tax net profit of 296 million yuan, a 23% increase over the same period of last year. The net profit after taxation for the first half of the year was 5.16 yuan, a 23.5% increase over the same period of last year. Shengli continued to benefit from Anson this year. US (ON-Semi) released the order, and cut into the Sony supply chain since 2015. The first half of this year from Sony CIS component packaging and commissioning orders have been stable and heavy volume, currently the second largest customer of Shengli. Looking forward to the second half of the automotive CIS components The market demand is still strong. Shengli's operating performance will be better than last year. It is expected to expand the production line in the third quarter. The channel manufacturer who has been acting for Sony CIS components for many years is still in addition to the US department. The supply of Korean manufacturers is led by Sony. Most of the first-tier mainland brand manufacturers have indirectly obtained Sony CIS components through Shangli. Shangli's main customer is Guangbao's camera module division, although Lite-On was once robbed by Huawei and other land-based manufacturers in the second half of last year. In the first half of the year, Lite-On Branch announced that it will transfer its camera business unit directly to the mainland connector manufacturer Lixun Group's Lijing Innovation in the form of business transfer. Lixun is the main force chain of many first-line mobile phone manufacturers. The company also obtained a 10% stake in Lijing Innovation through the transaction, and Shangli returned to the Huawei supply system. Familiar with CIS manufacturers, Huawei has already pushed the P20 three-lens mobile phone, the number of CIS components increased, and the market also passed 2019. The Apple iPhone is also expected to adopt three lenses. Under the help of the leading brands, the three lenses are expected to become the upgrade trend of smart phones. Although the overall mobile phone shipments have limited room for growth, some functions will be upgraded and will benefit CIS. Component supply system. Huawei, Oppo, Vivo did not significantly reduce sales targets this year. Under the trend of emerging market demand, Huawei will challenge the shipment of 170 million standards this year. It is worth noting that Huawei is targeting the next generation. The layout of the car is quite active. As the CIS sensor becomes one of the key components in the automotive electronics field and the amount of energy is greatly increased, the professional CIS component agent channel is still in the second half of the year and the subsequent operational growth momentum is worth looking forward to. DIGITIMES