Set micro-network launched integrated circuit micro-letter public number: ' Every day IC ', major news real-time release, every day IC, every day set micro-nets, accumulate micro-Cheng! Replication Laoyaoic Micro-letter public number search adds attention.
1.vivo will use the new SI chipset starting screen fingerprint identification cell phone;
A few days ago, Synaptics, a well-known chip maker, announced that it would join one of the world's top five handset makers in launching the first handset to adopt a screen fingerprint identification scheme, which is vivo by Forbes, a well-known Chinese handset maker.
At present, Synaptics and Vivo have solved the problem of blocking the large-scale production of the program's mobile phone, the mobile phone fingerprint identification with Synaptics clear ID FS9500 scheme can run safely and stably, this series of sensors is designed for smart phone, need to use OLED screen, Activating only when fingerprint recognition is needed can save cell phone power.
When the handset detects the user is preparing to unlock, the screen will show the fingerprint area, the finger touch can unlock the handset, the Synaptics program supports the OEM manufacturer to carry on the customization, because embeds in the screen, therefore FS9500 in the identification fingerprint aspect also naturally has the oil-resistant, the water stain, carries the dust and so Solves the embarrassment of not having a full screen phone with face recognition at the back.
Prior to that, Vivo had shown support for screen printing of engineering machines, but Synaptics could go a step further, Vivo revealed that they will start next January with screen fingerprint technology handset products. It house
2. Qualcomm Snapdragon 845 Chip 4 strokes try to surpass Apple A11;
Qualcomm (QUALCOMM) next-generation flagship processor Snapdragon 845 of the 8 core Kryo 385, may not be the market's most powerful mobile processor (CPU), but Qualcomm does not matter, Qualcomm at the Snapdragon Peak will spend a lot of time explaining how to use different parts of the chip, such as digital signal processors, image processors and graphics processor (GPU) to handle different workloads to achieve heterogeneous operations. According to PCMag, the Qualcomm approach is different from Apple's, with Qualcomm pursuing more cores, while Apple is reducing its core number. In processor benchmarks such as Geekbench, A11 performance may still be higher than Qualcomm's 4 arm Cortex-a75 and 4 A55 cores, especially on a single core workload. But Qualcomm's strategy will still be rewarded, with Snapdragon 845 likely to pick up the Galaxy S9 of Samsung Electronics. In color capture, Apple's photos and images use the DCI-P3 wide gamut standard, covering the entire CIE 1931 color space about 53%, Snapdragon 845 supports REC 2020 gamut, nearly 76% of CIE 1931. If the panel technology fits, you can create a more high-end function of the Snapdragon 845 mobile phone, the chip supports 2K screen, up to 120 frames per second, while supporting REC 2020 and HDR for the most realistic image. As a result, the images of Galaxy S9 may look more vivid, and the screen is more high-end than the Apple device. In terms of wireless speed, although Apple's data machine is not on the A11, Snapdragon 845 of LTE and Wi-Fi are better than the version Apple is using. Snapdragon 845 also supports WiGig, a super fast way to transfer data over a short distance, and faster access to the network on the 802.11AC, which can be a very reliable and easy way to connect to Wi-Fi networks on behalf of users. For voice assistants, Apple has a handheld Siri, Snapdragon 845, which supports multiple handheld wake-up words directly on its voice codec, representing Google, Alexa and Cortana will be able to coexist easily on Snapdragon handsets. HTC U11 also runs Google Assistant and Alexa on Snapdragon 835, but Snapdragon 845 is more efficient. In support of augmented Reality (AR), Apple's Arkit and Google's Arcore can only draw flat horizontal surfaces, which are completely different from the dynamic spatial mappings required for augmented reality experiences. Qualcomm said the use of Snapdragon 845, as long as a camera can be used for indoor 3D mapping, representing dual-lens mobile phones such as Galaxy Note 9 will be able to track objects in the room, and even try to identify people. Digitimes
3. Qualcomm Centriq 2400 compete with Intel Xeon for the data center chip market;
The Intel (Intel) X86 architecture Xeon Chip is still standing on the global server chip market, but Ultra Micro (AMD), Qualcomm (QUALCOMM) and IBM have also launched their own server chip to challenge Intel's hegemony, which Qualcomm launched in late November, code-named ' Amberwing ' Centriq 2400 server chip, its effectiveness has become the focus of attention, there will be a Qualcomm server application partner CloudFlare will Centriq 2400 and Intel's first two generation Xeon Chip, It was found that, although Centriq 2400 was inferior to Intel Broadwell and Skylake for each core performance comparison of the partial operational load, Centriq 2400 per system performance was significantly better than Broadwell and Skylake, per thread, Performance comparisons of Centriq 2400 per watt are also preferable. According to the Next platform Web site, CloudFlare This performance test is a single slot server with 2.5GHz, 46 core Centriq 2452 processor, with dual-slot server with pulse 2.2GHz, 10 core Broadwell The Xeon e5-2630 v4 processor and another dual-slot server performs a performance comparison with the Intel Architecture server with 2.1GHZ, 12 core Skylake Xeon SP-4116 processor. Since Xeon is hyperthreading with multiple threads at the same time, it is equal to 40 Broadwell threads, 48 skylake threads are compared with 46 Qualcomm Amberwing threads, two of which have a cost of 1 chips, 334 dollars, two skylake chips cost 2.004 U.S. dollars, a single Qualcomm Amberwing chip is 1,383 U.S. dollars. By CloudFlare, it can be found that each core performance of Amberwing chip is close to Broadwell in most test workloads, and Skylake has better performance than amberwing and Broadwell. However, from the performance of each system, because the amberwing system has more core, so in the encryption and compression load transmission volume on the amberwing often better performance. For each thread performance, the high pass thermal design power consumption (TDP) was almost identical to the Xeon Platinum 8180 of the Intel TDP 205 Watts in the 120 Watts Centriq 2460, and the integer test was 13.7 and 13.9 respectively; The Centriq 2452 of the Qualcomm TDP 120 watts is compared with the Xeon Gold 6152 of the Intel TDP 140 watts, with the integer test being 13.8 and 12.8 respectively, and the Qualcomm TDP 120 Watts Centriq 2452 with a slightly better performance of about 8% per thread; The Centriq 2434 of the Qualcomm TDP 110 watts is compared with the Xeon Silver 4116 of the Intel TDP 85 watts, with an integer test of 14.1 and 13.6, with a high-pass TDP 110 watts Centriq 2434 per thread performance slightly better than 4%. In comparison of the performance of each watt of each chip under TDP maximum heat transfer (heat dissipation), the Qualcomm TDP is 5.5 and 3.8 points compared with the Xeon Platinum 8180 of the Intel TDP 205 Watts in the 120 Watts Centriq 2460 respectively, Qualcomm per watt performance is superior to Intel up to 45%; Qualcomm 120 Watts of Centriq 2452 and Intel TDP 140 Watts Xeon Gold 6152 compared to 5.3 points and 4 points respectively, Qualcomm per watt performance is still better than the Intel 32%; The Qualcomm 110 Watts Centriq 2434, compared with the Intel TDP 85 Watts Xeon Silver 4116, was 5.1 and 3.9 respectively, and Qualcomm per watt performance was also superior to that of Intel 31%. Finally, compared with the Xeon Platinum 8180 of the Intel TDP 205 Watts, the Qualcomm TDP 120 Watts Centriq 2460 compares the cost per unit cost of the spec integer performance to 0.33 and 0.08 USD per unit, and Qualcomm is over 4 times times better than Intel; The Centriq 2452 of the Qualcomm TDP 120 watts is compared with the Xeon Gold 6152 of the Intel TDP 140 watts, each unit cost is 046 and 0.15 USD respectively, Qualcomm surpasses the Intel more than 3 times times; The Qualcomm 110 Watts Centriq 2434, compared with the Xeon Silver 4116 of the Intel TDP 85 watts, is 0.64 and 0.33 USD per unit cost, and Qualcomm is also better than Intel nearly twice times. Digitimes
4. Smart Speaker: 2017 Q3 Amazon and Google shipments accounted for a total of 92% of the global market share;
Set of micro-network messages, the market for the adoption rate of smart speakers continue to exceed expectations, there are clear indications that the market began to shift from the early stage to the mass market. In addition, thanks to Alibaba, Millet, Jingdong in the double 11 of the sharp price reduction and promotional activities, as well as the much-anticipated Baidu's first Duer OS speaker launch, China's smart speaker market in the fourth quarter has excellent performance.
Key findings in the recently released 2017 Q3 Smart speakers Quarterly report by Strategy Analytics include:
• In 2017 Q3, North America led the smart speaker market with a 75% sales share, although Western Europe and Asia-Pacific will weaken their share of sales in Q4.
Google home in a number of countries and regions to promote the growth of Google assistants, resulting in Amazon Alexa's AI operating system market share from the first half of 2017 80% down to 2017 Q3 69%. The launch of more sales countries and new models will drive the world's total sales of Google home and Amazon Echo to 12 million in the last quarter of 2017.
• Alibaba and Millet to launch new smart speakers, as well as in Q3 has been ranked first in China Jingdong Buzz box to promote the Chinese market began in 2017 Q3 growth momentum. "With the addition of new manufacturers such as Harman Kardon, Sony and Sonos, the competition in the smart speaker market is getting fiercer," said David Watkins, director of Strategy Analytics's new smart speaker service. These vendors have higher quality audio than Amazon or Google. As a response, Silicon Valley giants such as Google and Amazon have improved the audio quality of their flagship products, but their real competitive advantage is the ability to sell a large number of extremely price-enabled echo Dot and Google home Mini, making it far ahead of rivals in the foreseeable future. '
' Competition will continue to warm up in 2018 and we will see the launch of Third-party devices that support the Alexa or/and Google Smart voice AI platform. All of this will certainly add to the pressure on Apple ¬--'s Siri homepod Smart speakers were not released in time for the Christmas season and will be unveiled in 2018 years. Apple has lagged behind Amazon and Google in the field of AI and virtual assistants, and its laudable hardware innovations may not be enough to counter its rivals ' rapid AI software innovations. '
5.Tesla development AI chip implementation of vertical integration of mass production capacity can amortize the cost of research and development?
Tesla CEO Elon Musk, although the development of AI technology is more concerned about the development of AI seems to be more negative, but Musk recently in the California Long Beach held the 31st session of Neural Message processing system (NIPS), Says Tesla is developing its own AI chip, which is tantamount to confirming the rumors that Tesla is developing its own AI chip. In this case, whether the future Tesla and Nvidia chip cooperation may split up, and Tesla can afford to develop the chip may need to invest the huge cost, etc., have become the focus of attention. Tesla and ultra-micro cooperation possibilities high according to the Motley Fool, a number of media reports, Musk confirmed that Tesla is developing self-driving AI chip, but did not mention whether it is with micro (AMD), or in order to replace the current with Nvidia partnership, However, it is speculated that Tesla chip design may have cooperation with ultra Micro technology. Tesla also revealed its AI chip development project, led by Jim Keller, a legendary chip engineer who worked for micro-and Apple (AAPL), and has a number of super microchip veterans in charge of important roles. While it is thought that Tesla to develop AI chips on its own, it may not be good news for partner Nvidia, but Tigress financial analysts believe that Tesla's own development of AI chips is good news for chip manufacturers, Because it is possible to obtain Tesla to entrust the business opportunities. Tesla may only be responsible for AI chip design work, manufacturing part of the company with external vendors or outsourcing production mode, in the future if more manufacturers to invest in self-driving AI chip design, will be able to drive more AI chip production requirements, which will form a virtuous cycle, so on Nvidia or Skyworks FX and other manufacturers, such a trend of market development is optimistic. Loup Ventures analysts point out that even if Tesla did not work with Nvidia at the end, with Nvidia's excellent self-driving chip technology, it would be possible to obtain cooperation from other carmakers, such as Ford and General Motors, even if the Tesla partner was lost. GM) and so on. Because these traditional carmakers do not have the technology developed and manufactured by these AI chips, they tend to have this kind of self-driving AI chip Development Committee from outside the chip professional, in which case, the NVIDIA chip technology will naturally form a demand. Develop AI chips conform to Tesla vertical integration beliefs Generally, the reason why you want to develop a customized chip is that you want to improve chip performance, power consumption and cost savings, and have greater control over its own technology, without being subject to external chip makers, which is the most successful case of Apple. Musk also said at the nips that the development of AI chips could help Tesla potential improvements in power performance while reducing costs. Part of the media analysis, the move is Tesla to reduce the dependence on other companies a part of the strategy. At present, Nvidia's graphics chip (GPU) is also widely used in the AI field, but this kind of GPU is not suitable for specific purposes, this may be the Tesla body to recognize the development of its own self-driven by the need for the production of the chip. Tesla has always been a proponent of vertical integration of beliefs, the design of their own chip fully conform to Tesla this concept, it can also understand why Tesla had to find Keller reasons. It is expected that the AI chip will also help Tesla to the level 5 technology milestones, to achieve Musk previously announced that in 2019 to let the full auto technology market commitment. It is not only time-consuming, but also expensive to invest in the development of chip cost, so it is inevitable that large scale production will be needed to amortize the huge cost. This is not a problem for Apple, after all, Apple shipments are so impressive that its a-series chips are even used by other devices, such as Apple TV, homepod and ipod Touch, to scale the cost of research and development as much as possible. However, Tesla at this stage still do not have such a large terminal port, in the past year, only 95,000 vehicles shipped, if the next few years Tesla production and delivery can be significantly increased, of course, will help to flatten its AI chip research and development spending, but if the failure to achieve the mass production target, which would become Tesla another cost burden, Will test Tesla operational intelligence. The current technology industry to create a customized chip has gradually become a new trend, such as Apple and Huawei has been in the mobile phone using a customized chip, Google's own cloud AI services also using a customized chip, even the UAV and consumer cameras began to carry a customized visual chip. Digitimes
6.2017 November North American Semiconductor equipment shipments of 2.05 billion U.S. dollars
Set of micro-network news, the International Semiconductor Industry Association (SEMI) released the latest shipping report (billing) showed that November 2017 North American semiconductor equipment manufacturers shipments of 2.05 billion U.S. dollars. It grew by 1.6% compared with the $2.02 billion trillion in final data in October, up 27.2% from 1.61 billion dollars in the same period last year.
Semi Taiwan District President Cao Shi said the November North American semiconductor equipment shipments in the last four months after a recession for the first time is growing, semi estimates with China's new wafer factory began operating, 2017 years of growth momentum will continue to the 2018.
Semi's billing report is based on the average global shipments of North American semiconductor equipment manufacturers over the past three months.
