1. After the huge acquisition of Altera, Intel won the eASIC dish again, and it is expected that Q3 will complete the acquisition;
Original title: After the huge acquisition of Altera, Intel won the eASIC dish, and it is expected to complete the acquisition in the third quarter.
Abstract: Intel announced the acquisition of eASIC, a semiconductor design company, which is expected to complete its acquisition in the third quarter. At this time, 120 eASIC employees, including CEO Ronnie Vasishta, will also be transferred to the Intel Programmable Solutions Group (PSG). From the perspective of Intel, the acquisition of eASIC can better meet the needs of next-generation computing. eASIC will complement its FPGA business. From the perspective of eASIC, it can better advance the chip of the next process node and the Arm core plan.
Set micro-network news (text / Xiaobei) Beijing time this morning, Intel announced the acquisition of semiconductor design company eASIC, but the specific plan for the acquisition has not been announced. A spokesperson for Intel said that the purchase price is 'not important'. It is reported that The acquisition is expected to be completed in the third quarter, at which time 120 eASIC employees, including CEO Ronnie Vasishta, will also be transferred to the Intel Programmable Solutions Group (PSG).
Intel and eASIC 'engagement' began in 2015, eASIC provides Intel with its ASIC design style Xeon chip.
Dan McNamara, general manager of the Intel Programmable Solutions Group, said, 'eASIC is now a small company, but we believe that Intel's convergence and expansion will give it a technological advantage that differs from Xilinx.' At the same time, Dan McNamara acknowledged that Although Intel has never invested in eASIC, it has always had this idea.
This acquisition is positive for both Intel and eASIC.
From an Intel perspective, the acquisition of eASIC is better suited to the next generation of computing needs, and eASIC will complement its FPGA business.
In 2015, Intel spent $16.7 billion to acquire Altera Corporation and established the Programmable Solutions Group, which is one of Intel's diversification initiatives after the decline in PC processor market revenue. Today, FPGAs are at Intel. The big data strategy plays a crucial role.
Intel's latest quarterly earnings report shows that PSG's annual revenue increased by 17%, and the Client Computing Division (CCG) grew by only 3%. This growth provides the basis for Intel's future market. Currently, FPGA becomes a super data center. For mainstream applications, Intel and rival Xilinx are actively deploying data center services. Intel's acquisition of eASIC is an important step in the future layout of the programmable programmable business.
FPGAs provide a quick way to design circuits in a circuit. In fact, most CPUs and GPUs are first simulated on the FPGA before they are built to ensure they work. It is well known that ASIC chips are highly efficient. Advantages, but its development costs are higher. At present, some small companies will adopt the way to deploy ASIC design to FPGA to save time and cost, but at the expense of power consumption.
The eASIC concept entered the public vision for the first time. It can be said that in 2014, its creativity proposed a design interconnection unit, which can solve the problem that the power consumption, performance and cost of the FPGA are limited by the wiring resources, so that the eASIC integration into the SoC truly becomes Possibly. eASIC can be understood as a 'structured ASIC'. The logic elements on the chip and the corresponding masks are fixed. Users can realize the logic of their own design by interconnecting the logic cells themselves, so they can be manufactured. Significantly reduce manufacturing costs and time, but for flexibility and performance, it is also a discount.
In product planning, Intel said: In the long run, there is an opportunity to build a new programmable chip that uses Intel EMIB technology and combines Intel FPGAs with structured ASICs in a system-in-package solution.
Dan McNamara said that chip programming is not done in the field, but in the factory. The cost of programming the chip at the factory is still hundreds of thousands of dollars, but it only takes several months instead of two years. Currently, The eASIC chip was manufactured by TSMC and GlobalFoundries. Intel has not yet decided whether to produce its own eASIC chip. Dan McNamara said, 'We will further study this issue and determine the strategy for future generation products.
The acquisition of eASIC will not only bring the technology to the Intel product line, but also expand the Intel chip customer base.
Currently, eASIC customers include many suppliers working in the communications industry, such as Huawei, NEC, Violin Memory, Seagate, Microsoft, Flir Systems and Arm. Intel's acquisition of eASIC will also expand 4G, 5G and other customers, in network communications and things. Get more revenue in market areas such as networking.
From the perspective of eASIC, it is better to advance the chip of the next process node and the Arm core plan.
Founded in 1999, eASIC is an IC design company that manufactures customizable eASIC chips at its headquarters in Santa Clara, Calif., with chips for wireless and cloud environments. EASIC investors include Khosla, Kleiner Perkins and Seagate. , a total of $ 149 million in funding. eASIC has undergone a capital restructuring in 2012. Intel did not disclose the financial terms of the acquisition of eASIC, because the price paid is not important to Intel. Intel spokesperson said the purchase price is 'not high'. In the last round of investment in November 2017, its valuation was about 110 million US dollars, and everyone can make a rough guess about its purchase price.
In 2015, eASIC applied for an IPO but eventually decided to cancel the listing plan. In the same year, eASIC talked about plans to support the Arm core, but its investors were not willing to support its future process node advancement (after 28nm) and Arm core plans.
It is reported that if Intel's acquisition of eASIC transactions proceed as planned, eASIC will provide the Arm core solution in early 2019.
eASIC President and CEO Ronnie Vasishta said in a statement that the eASIC team has developed and deployed a truly innovative structured ASIC product. The combination of eASIC technology and Intel's intellectual property and functionality will enable this proven structure. ASIC products are ubiquitously deployed in exciting end applications and markets. This is the best time to start a new chapter in eASIC.
It can be seen that eASIC is very positive about this harvest. This acquisition will also be a beginning for its release of 'ambition'.
2. Autopilot 'killing' is getting worse and worse, how does Intel and Mobileye break through?
Abstract: Compared to other competitors, Intel's entry into the autonomous driving field can be described as a 'latecomer'. However, after acquiring Mobileye for $15.3 billion last year and adding its own in-vehicle computing products, Intel quickly expanded its technical strength. In the autopilot market, can Intel 'get in the future'?
Episode News (Reporter/Mao Mao) In the past two years, the competition in the autopilot industry solution has been particularly fierce. There are established vendors such as Intel, NVIDIA, Qualcomm, NXP, Infineon, Renesas, TI, ADI, ON Semiconductor. There are also newcomers such as Tesla, Apple, and Huawei, as well as horizons, rookies like the Cambrian, and players in the autopilot field spanning multiple fields, and there is a scene of more and more 'smashing drama'.
In May of this year, Vision Fund, a subsidiary of Soft Capital, invested $2.25 billion in the general autonomous driving department of Cruise, and entered the autonomous driving field. On the same day that Softbank invested in Cruise, Waymo announced that it would purchase 62,000 Fiat Chrysler vehicles. To expand its self-driving taxi team; at this year's CES show, NVIDIA released the world's first self-driving L5-class chip, the Drive Xavier; in addition, Apple is also working in this area, currently in California for autonomous driving The number of vehicles has increased to 62...
Undoubtedly, a bloody battle in the field of autonomous driving has arrived, and in this bloody battle, one player that has to be mentioned also includes the chip giant Intel. Last March, Intel announced the acquisition of global assisted driving (ADAS) for $ 15.3 billion. The giant Mobileye not only set a large-scale transaction in the field, but also launched an important shot in the field of strong entry into the automatic driving field.
Building the 'ambition' of the autopilot map
At present, the level of automatic driving is still in the L1-L2 level. It is expected that the commercialization of fully autonomous vehicles will be realized in 2025. According to the reporters of the micro-grid, as the actual application at this stage is more ADAS, the ADAS related industry is also the next three years. With the ability to explode growth, Mobileye is the 'leader' of the field.
Founded in 1999, Mobileye is an Israeli company that specializes in the development of ADAS systems and autonomous driving vision technologies. Mobileye also has EyeQ series of vision processing chips developed independently for the autonomous driving industry. According to Boaz Sacks, Director of Mobileye China, as of now Mobileye has partnered with 25 global OEMs and 13 automakers, with more than 27 million vehicles worldwide powered by the Mobileye ADAS security solution.
According to relevant data, several products of Mobileye currently account for 90% of the market share of ADAS. Even insiders said that Intel's acquisition of Mobileye has suddenly made up for nearly five years of shortcomings in research and development. It can be seen that the value of Mobileye is not unreasonable.
For Intel, the acquisition of Mobileye is a very important part of its construction of autopilot. Before the acquisition of Mobileye, Intel also acquired FPGA chip giant Altera, visual algorithm company Movidius, to form an autopilot end-to-end integrity. solution.
In addition to huge capital investment, Intel has spared no effort in talent competition. In April this year, Tesla said that Jim Keller, head of the automated driving system department, will leave and join a semiconductor-focused company. According to informed sources. Jim Keller's new owner is Intel.
This series of actions all imply Intel's 'ambition' to build autonomous driving maps.
Entering the Chinese market
The synergy effect after the acquisition has begun to appear.
Mobileye China Director Boaz Sacks believes that Intel's acquisition has had a positive impact on Mobileye. 'Unmanned driving is a huge field. Although Mobileye is distributed all over the world, it is still a small company, and Intel One The annual financial report amount is almost equivalent to the annual budget of a small European city, which is a great support for us. 'Boaz Sacks responded.
At the same time, Boaz Sacks revealed that from the beginning Mobileye has established its own driverless team rather than just ADAS. Currently, Mobileye has signed an agreement with a European car manufacturer to provide 800 car autopilot technology. Beginning in 2021, Mobileye will launch the EyeQ5 chip designed for L5 fully autonomous driving. To achieve this goal, Mobileye is conducting intense testing in Israel and Europe.
As the autonomous highland, the Chinese market has become a strategic center for all players, and Intel is no exception.
For the layout of the Chinese market, Boaz Sacks said that Mobileye currently has an office in Shanghai, and the 15-person team is mainly responsible for the aftermarket. In the pre-installation market, Yutong, Dongfeng, FAW, SAIC and other 12 automakers have They are all Mobileye partners. Among them, the cooperation between Mobileye and Zhengzhou Yutong Bus has been going on for a long time, because most of the buses in Israel are manufactured by Yutong.
It is worth mentioning that the first batch of intelligent networked open road test number plates in China has been issued in Shanghai recently, and the first batch of licenses issued by the Shanghai Municipal Government has a total of three licenses. Among them, the GS of the SAIC Group won two license plates. The car ES8 gets a number plate. It should be noted that the first three cars that have obtained the intelligent networked car open road test number plate use the Mobileye EyeQ4 chip.
RSS model guarantees security
For the world's drivers, 'safety' is always in the first place. However, combined with the recent 'auto-driving' accidents, the safety of self-driving cars after commercial use is still controversial. How to Assuring that driverless cars are safer than human driving is still a major problem that plagues companies in this field.
'To ensure the safety of driverless cars, we have done a lot of research and published the results of research in academic journals and publications, proving that the Responsible Sensitive Safety Model (RSS) is very feasible. 'Intel Driverless Solution Jack Weast, senior chief engineer of the program, chief system architect, said in an interview with the micro-network reporter.
In fact, RSS is a set of mathematical formulas. Intel wants to define the 'safe state' mathematically. In the 'safe state', no matter what other vehicles react, autopilot cars can't cause an accident.
To put it simply, Intel's definition of a set of RSS equations mainly includes two aspects: On the one hand, driverless cars must not cause collisions or accidents for their own reasons; on the other hand, when a potential risk is caused by another vehicle When a traffic accident is caused and may occur, the driverless car should take appropriate measures to avoid the traffic accident.
According to Jack Weast, RSS is a good model that deserves to be considered by the entire industry or enterprise. It is also hoped that industries around the world can work together to ensure that RSS models work in all countries. Meanwhile, Jack Weast said that when RSS After the model came to China, because China's driving habits and road conditions will be different, the RSS model will be improved for China's situation.
According to the collection micro-network, on July 4 this year, Baidu announced cooperation with Mobileye, plans to deploy the RSS model in its Apollo open source project and Apollo Pilot commercial project, which means that Baidu will join the Mobileye security system.
Compared with other competitors, Intel's entry into the field of autonomous driving can be said to be a 'later'. However, after acquiring Mobileye for $15.3 billion, Intel quickly expanded its technical strength. Now the bloody battle in the field of autonomous driving has begun. Under the strong enemy, how Intel and Mobileye attack the city, let us wait and see! (Proofreading / Emil)
3. From game drawing to artificial intelligence GPU becomes the corner of AI new era;
Computer graphics is the driving force behind GPU development. Drawing applications are never satisfied with computing power. At the same time, this has produced a huge industry scale. The entertainment industry has a market demand of 1 billion image renderings per year. 400 new games are launched every year. The application of the global lighting technology is like a movie special effects scene. In addition, 500 new films each year, each picture is also the result of numerous renderings. There are currently 12 million designers worldwide, and All product designs in the world also rely on drawing technology. As for 150,000 architects, they also use drawing technology for interior and outdoor architectural design. These immersive rendering algorithms cannot be accelerated. The CPU can only perform rendering jobs, until now the GPU technology makes these applications come true.
The six major markets, including high-speed computing, AI, image processing, transportation, medical and urban security, will bring huge business opportunities.
Creating a high-realistic entertainment world is the goal pursued by computer scientists in the past decade. Rendering means that when the light source hits the surface of the object, the light absorption, reflection and refraction calculation can be performed according to the material of the object. The industry now uses ray tracing technology, but even a single screen requires a few hours of computation on the server by the CPU, and a 2 hour movie has more than 300,000 frames, which can be rendered through the CPU. It takes millions of hours. But if it is accelerated by the GPU, these tasks can be achieved in a shorter period of time.
As for medical treatment, it is currently one of the largest industries in the world. It is estimated to have an output value of about 7 trillion. Early detection of early treatment is the best weapon against disease. Today there are 3 million medical devices, all of which are in AI technology. The operation was completed before the breakthrough. With the maturity of computing technology, from detection, diagnosis to drug development, the medical industry is also undergoing a revolutionary AI technology baptism.
Smart cities bring safe urban industries, and in the future it is estimated that there will be two trillions of output value. Billions of cameras monitor public areas to ensure public safety, control traffic flow and optimize energy use. Without AI, it will not be possible. There are enough people to continuously monitor these details. In the past 12 years, the global population has increased by one billion, and the city has become more and more crowded. The “Amazon effect” has caused frequent shipments of online shopping products, resulting in a serious shortage of truck drivers. The transportation industry is expected to create 10 In the future, all transportation vehicles will adopt a self-driving system.
The invention of self-driving is one of the most important contributions of AI. It will provide a new taxi service without driving, help truck drivers to reduce pressure and expand the scope of work, and more importantly, improve road safety. However, at the same time, self-driving It has also become one of the biggest challenges in the history of computing. The self-driving car computer must be a high-quality, real-time supercomputer that can be used to train, simulate and test AI software.
AI's business opportunities are infinite. It is estimated that when it comes to the end of Moore's Law in 2028, the computing power required by the market is reaching its peak. Six markets including high-speed computing, AI, image processing, transportation, medical and urban security will all be in the market. Bringing huge business opportunities. Looking at these application areas, the single architecture + killer application + global ecosystem + the market value of the US dollar will become the key to the development of high-speed computing. From GPU to game graphics, in the future The development of AI will also become an indispensable corner. It is worth noting that, in the future, the CPU will compete with the GPU in terms of the application of AI. This computing war is expected. CTIMES
4. From voice activation to spatial audio, make the sound more intelligent;
Moshe Sheier, Strategic Marketing Director, CEVA
With the development of a new generation of smart speakers and other voice-activated devices, you can talk more comfortably with your device. What if you use a translating headset when traveling abroad? Although they are not very common, Google just released The smart headset Pixel Buds already supports this feature.
In a recent article, I reviewed the second wave of smart speaker development, characterized by efficient production and volume launches. I speculate that this phase will put pressure on market leaders to promote technological innovation and new manufacturing. The product type, while further improving the entry barrier for others. This is what we saw this month, Google, Amazon, Apple and other companies have released a lot of new audio voice products.
Last year's launch of Google Home responded to the previously successful Amazon Echo. Now, Google is expanding its product line to include Echo Dot, Apple HomePod and AirPods. Google Voice Activation Assistant can therefore have more usage scenarios, Larger price space.
The new Google mini is shaped like a pin pack, smaller and cheaper, and is a replacement for Google Home. For those looking for a better sound experience and willing to pay more, Google Max offers a more high-end sound experience. It also has all the features of Google Home. The new Pixel Buds offers an on-the-go experience, paired with a smartphone to create an in-ear personal assistant, similar to Apple's AirPods for Bluetooth audio. Stream. One of the most interesting features of these headphones is the integration of Google Translate, which supports in-ear simultaneous interpretation as seen in the video. Google says it will support 40 different languages, and this number will undoubtedly evolve with the product. And keep increasing.
Another exciting development in voice activation technology is that the new Gopro Hero6 sports camera can be turned on with voice commands. This is GoPro's second generation of voice-enabled devices. The earliest introduction was Gopro Hero5.
Most voice commands are the same as before, such as 'GoPro start/stop recording' and so on.
The innovation of the new model is that the user can use the voice command - 'GoPro to turn on' to turn on the camera. This feature is optional and remains active for 8 hours after the camera is turned off. Although this is not a complete listening at any time. (always-listening) device, but it took an important step in this direction. The voice-activated Gopro also exemplifies our other vision, voice will become the main user interface. In the process of human-computer interaction, Sound is always the most natural and intuitive way, and soon all equipment will support listening at any time, waiting for our orders.
Alexa's next application is the car, then your face.
Amazon is making this vision a reality, and it is still a step ahead of the 'newcomer' (Google) in the smart speaker market. Before the Google conference, Amazon was the first to release a series of new Echo products, including a new design. The Echo product line and the smart alarm clock Echo Spot continue the style of the integrated screen that Echo Show started. Amazon also revealed that Alexa will soon be integrated into the BMW 2018.
This may be the first time Alexa has become the most popular road travel companion after becoming the most popular home smart speaker.
More interesting than any official news is the speculation about the upcoming Amazon glasses. The rumor of the spread is that it is based on Alexa's non-handheld, battery-powered glasses. According to speculation, Amazon glasses will not provide a visual experience, but as A wearable device that will allow users to talk to Alexa anytime, anywhere. This will be an important step for Amazon to move forward into the wearable field and support the always-on feature.
Obviously, in order to release the full potential of the voice assistant, it is necessary to eliminate the power jack. Here, related technologies can be implemented.
Spatial audio can generate or break virtual/augmented reality
With the addition of Amazon and Google, it will directly lead to the overall improvement of the ear-wearable devices. Many interesting concepts have been proposed for this product, such as the artificial intelligence personal trainer named Vi in the Kickstarter project. Biometric features and thus customized training to achieve moving goals. Now imagine that by adding multi-dimensional spatial audio to create the feeling that Vi is in front of you or behind you, can you provide extra motivation to help you break the individual? Record (see patents applied 10 years ago).
The latest audio innovation concept was introduced in Apple's new mobile phone. The iPhone 8 included dual speakers for the first time. This is a big thing, because dual speakers can enable multidimensional spatial audio, which means that Apple is betting on augmented reality and Virtual Reality (AR/VR) applications. To create an immersive AR/VR experience, spatial audio is a must. Otherwise, even a stunning picture will not produce a convincing real space.
The next step is the neural network enabled sound perception
What is the next step in audio development? The iPhone8 and iPhoneX already include a dedicated neural network engine. For ear-wear and voice-activated devices, neural networks can be used for sound perception and audio analysis. This technology has been used to identify at home. Certain sounds, such as doorbells or broken glass, trigger an appropriate response. It improves the safety of ear-worn devices, such as Vi personal trainers and other immersive AR/VR applications. Neural networks can perceive Some important sounds, and notify users, such as alarms or dog barking. In this way, both security and enjoyment can be taken into consideration.
5.T2 chip has taken over the startup control and security features of iMac Pro and 2018 MacBook Pro
Compared with the MacBook Pro models released in 2016, Apple's new T2 chip on the iMac Pro and 2018 MacBook Pro has further expanded its management functions. It is reported that the T2 chip took over the previous CPU control and security. Other features, including the quality of the FaceTime camera, driver security, and the complete startup process. Although there have been reports and rumors in this area, we only know that it manages a certain degree of boot security, integration and security of the chip. There is not much understanding.
In January of this year, MacWorld's Jason Snell analyzed the T2 chip in detail, detailed its role as a mass storage controller, and pointed out that the T2 chip has complete control over the flash storage array inside the iMac Pro.
It is reported that T2 will encrypt every bit of data sent to the flash memory and decrypt it when the user needs it. Because of this, it is impossible to recover data directly from the flash module of the iMac Pro.
Another major feature of the T2 chip is the full management of the boot boot. The verification process from start to finish includes:
Verify that the cryptographic signature of the bootloader is legal and correct before handing the rest of the process to the other hardware of the iMac Pro.
All of this is managed by the aforementioned secure boot utility, which the user can call up via Command-R during startup.
Snell's report states that the default management level is 'Full Security' - this requires a network connection to verify the legitimacy of the operating system during the installation process - including installing the latest Windows 10 operations with Boot Camp system.
The medium security level eliminates the need for network connectivity, but this feature can also be completely shut down, and users can enable/disable booting of external media in the 'Start Security Utility'.
On the new iMac Pro and 2018 MacBook Pro, the T2 chip also hooks up the FaceTime camera:
Because the image processor (ISP) is also included, it changes all the parameters of the FaceTime camera, much like the iPhone's automatic adjustment of the image.
In addition, the T2 chip has more privileges, such as controlling and protecting the security of the computer microphone, adjusting the fan speed, and controlling the speakers of the iMac Pro.