Set micro-network to launch integrated circuit micro-letter public number: ' Every day IC ', major news immediately released, every day IC, every day set micro-net, product micro-Cheng! Replication Laoyaoic Micro Letter public number search adds attention.
1. Gradually get rid of dialog, Apple's independent power management chip is coming out next year;
Integrated micro-network reports, according to Nikkei news, Apple's own design of the power management chip will be the fastest to use next year's iphone models, to reduce the British semiconductor company dialog dependence.
Over the past few years, Apple has increasingly been designing its own chips, such as a-series CPU, GPU, and so on, using the iphone and ipad. Apple's next independent chip-design field appears to be battery and power management. Nikkei said Apple's power management chip ' will be the most advanced in the industry '.
According to Apple's current plan, the Company intends to use its own power management chip to replace part or half of its current iphone power management chips starting next year. Another person familiar with the report said Apple did develop a power management chip for the iphone, but the timing was uncertain and could be postponed until 2019.
Dialog currently designs power management chips for the iphone. If the report is true, Apple will not have to pay royalties to dialog in the future. Dialog shares tumbled 15% after reports from the Nikkei Asia Review were released. However, Apple and dialog have yet to formally announce a change in relations between the two sides.
In addition, like A11, A10 and other apple-designed chips, the Apple power management chip will be produced by TSMC.
In 2016, more than 70% of dialog revenue came from Apple. Dialog's share price fell 36% in the first time this April when Apple had to abandon dialog to develop its own power management chip. In recent years, Dialog has also tried to reduce its reliance on Apple and several other smartphones. In 2014, for example, attempts to merge with the Austrian sensor chip manufacturer AMS AG, in 2015, attempted to acquire Atmel (Atmel), but both ended in failure.
2. Intel launches the largest Soc/fpga:stratix SX series chip in history;
Intel (Intel) has announced that the Stratix SX series of chips will begin shipping. Stratix-SX Series consists of 10 devices, the logical unit (logic element; LE) number between 400,000 ~ 5.5 million. Each device has 1 dual-core or 4-core ARM cortex-a53 processing subsystems. and its closest competitive product, Xilinx (Xilinx) zynq UltraScale + MPSOC EG series has about 1.1 million logical units. According to Eejournal reports, Stratix SX cortex-a53 operation Pulse up to 1.5GHz, embedded memory up to 229Mb, as well as up to 5K DSP module (block), 11K 18x19 multiplier, As well as up to 144 SerDes transceivers running at the highest 30Gbps. Stratix the size of the ZYNQ, is actually more expensive, larger, more power consumption of different product categories, the two product lines there are many differences. But in a similar sense, the combination of the programmable logical structure and the processor subsystem is a surprisingly successful combination. The combination of powerful application processors and FPGA architectures on the same chip is far more than simple integration and can dramatically increase the number and type of links available between processors and other related logic, interfaces, accelerators, peripherals, and memory. The FPGA/SOC combination brings amazing functionality that can not be achieved by combining the FPGA with the processor. Starting with marketing, Xilinx and Intel are taking a very different path in the Fpga-soc field. Xilinx has been very careful not to refer to Zynq as a FPGA, preferring to position it as an advanced function of the SOC. Altera has been referred to as the SOC FPGA from the beginning. In fact, the engineering community has a broad understanding of SOC and FPGA, and the decision to combine these terms reflects the preconceptions of each of the two companies and the types of major designers they try to win. Intel is serving customers who already understand the FPGA and want the FPGA to have a powerful processor. Xilinx is pursuing a processor design and may want the processor to have a few programmable customers. This difference is evident in marketing materials, tools and devices. The Xilinx ZYNQ device contains many additional SOC, such as instant and drawing processors. The Intel FPGA Soc is a higher-order FPGA that includes the arm core and more mature. The FPGA itself has some significant differences. The Hyperflex architecture of Intel has placed a series of small latches throughout the data path, creating 1 of micro-pipeline structures that allow the timing optimizer to more evenly balance latency on critical paths to achieve a faster maximum time pulse. In arithmetic, Intel contains support for Single-precision floating point in its DSP module, and Xilinx does not claim faster fixed-point operations. Depending on the application, it can be valuable or wasteful to support floating operations. However, if you want to purchase a higher-order FPGA with an embedded ARM application processor, Intel's new 14 nm FinFET process Stratix is the only option. These devices are monolithic design, so they do not increase the cost of advanced 2.5D package, and the strong specification should be able to make it in many applications, including Intel locked 5G wireless communication, software defined radio, military security Operations, network function Virtualization (NFV) and Data center acceleration. Especially in the field of wireless applications, it would be interesting to observe the adoption of Stratix-SX and Sailing Zynq RFSOC. The former is a mature higher-order SOC FPGA, built-in application processor; The latter is equivalent to the built-in RF/analog components of the medium-scale SOC FPGA. For each specific application, the design team will need to weigh in on the larger FPGA capabilities, as well as the ability to integrate RF components into the same device, as well as costs, power consumption, and other common factors. Digitimes
3. Intel and Qualcomm grudge difficult solution arm server processor fear of escalation;
While Intel is attempting to chase Nvidia in the artificial intelligence (AI) field, efficient operations (high-performance computing; HPC) Market competition has also intensified. Ultra Micro (AMD) has launched the EPYC server processor, and IBM has published its Power 9 processor, as well as Cavium has previously released the "X2 arm processor. These 3 great cause include the x86 camp and the ARM Camp server processor supplier, in fact, the 3 industries, including the ultra-Micro, IBM and Cavium, have been plowing the server-processor market for a long time, but have yet to reach a significant market share, despite the impact of Intel's influence, but Although Qualcomm (Qualcomm) is more of a server processor later than the industry, industry analysts believe that Qualcomm or will be able to give Intel a small threat. Gao Tongfu published its first ARM architecture Centriq 2400 server processor in early November 2017, and was produced by Samsung Electronics (Samsung Electronics) in 10 nm process nodes. Qualcomm claims that this 64-bit server processor delivers better cost performance and performance per watt compared to Intel's Xeon Platium 8180 processors. Qualcomm is locking up its first Centriq processor family of products, targeting high-density cloud data centers (cloud data center) customers who desperately need blade server architecture (Blade server architecture), low cost, and higher power efficiency ( Power efficiency). Given that Intel has taken over 90% per cent of the market for this data center, Qualcomm's move is bound to target a share of the company's share of Intel, so Qualcomm must encourage Intel's customers to Centriq 2400. So far, Qualcomm seems to be ready to plunge into the competitive server-processor market, with the endorsement of large cloud partners, including Alibaba, Microsoft's Azure, Red hat, and server vendor benefits and technology (HPE) support; At the same time, according to Qualcomm, the company, including Google and Facebook, is actually welcoming Qualcomm's ability to compete in this highly oligopolistic server market, helping to maintain price levels and stimulate technological innovation. True, the server processor market leader Intel will not wait until the enemy, as the technology industry turns to AI to drive the need for a large amount of data processing, Intel is also with the time to revise its product development blueprint to meet customer needs. According to Trish Damkroger, general manager of Intel Data center group, the company has halted Knight Hill development, and Intel is now working to build another processor, Knights Mill, To handle the workload of AI and deep learning. Although Intel has yet to formally provide a clear message, the Intel Data center group general manager has hinted that Intel is going to build a new platform that will support the convergence of efficient computing, large data, and AI, which industry analysts believe is Intel's recently extended micro-drawing processor director Raja Koduri, which may also be related to the platform, is also likely to contribute to the developing platform, including the Altera and Nervana acquisitions.
What you can imagine is that Intel's expectations for Knights Mill are not just competitors in efficient computing, such as Qualcomm, but perhaps also include Nvidia. The feud between Intel and Qualcomm, the first major IDM in the United States, and the first major IC design industry, has cost Intel 10 years, spending $1 billion on the handset chip market not only to reach a high pass, and Intel finally backed down in 2016, where Qualcomm now crosses the mobile field, Walk into the data center market to Intel battling, and Intel should probably be ready to take its own medicine. Digitimes
4. Intel also does not want Qualcomm to be swallowed by the Bo-Tong, analyst: Qualcomm shares are doomed to frustration;
Set of micro-network news, the acquisition of Qualcomm will bring the opportunity to enter the server market. Because of cost advantages and energy consumption, the industry generally believes that the ARM architecture in the data center processor market will ultimately outweigh the x86 architecture. Prior to this, Intel's x86 processor has proved to be not cost-competitive in the Tablet PC market. Cloud computing + terminals, if Qualcomm can win two big markets at the same time, the prospect is very imaginative. Boton put forward to Qualcomm's acquisition, it is Rob in the Qualcomm server market, imagination has not landed before the "sweep goods." The biggest capital of Qualcomm's board and executive team will be the future story and imagination of the server market, under the aggressive offensive of Bo.
On the one hand, Qualcomm is flipping through the lines of the two sides, attempting to take the ARM architecture server processor into the Intel-dominated data center market. On the other hand, in the industry's long-awaited 5G market, Intel is also blowing the horn of the counter-attack, on the battlefield of the 5G revolution, to the High pass war, 5G to connect smartphones, cars, drones and industrial devices, etc. simultaneously connected to the Internet, Qualcomm has enjoyed a pioneer advantage, and Intel is not willing to miss. The 5G network is expected to be launched in 2020, while Qualcomm is expected to import into the commercial phase by 2019. In the 5G era, car connectivity is expected to be far more powerful than today's smartphone links, and car connectivity will become more popular with self-driving cars.
In this regard, Qualcomm and Intel have clashed again. Digitimes pointed out that if the success of the Qualcomm and the grace of the semiconductor merger case, Intel in the automotive communications chip market more than Qualcomm-eun-chi, the enemy, Intel will have to take the lead in order to not replicate the automatic driving processor market deficiencies, The possibility of Qualcomm and Eun-Zhi's fitness has posed a significant threat to intel. Therefore, the other is not willing to see Bo Qualcomm-Qualcomm purchase together success, to be considered as Intel.
Industry analyst Karl Ackerman points out that after the combination of three companies, Qualcomm, High-pass and Grace, it will be expected to have most of the high-end 0 components of the smartphone market and have a deeper impact on the 5G standards, and it is worth noting that the 5G standard is also important in the automated driving market and in the industrial networking market. The company will be in the future in the automotive processor, automotive communications chips and the cloud network market, such as the location can not be ignored,
In addition, the proposal may be resisted by Qualcomm's existing young-generation board. Paul Yas Jacobs (Paul Jacobs) is the executive chairman and the founder son of Owen Yas Jacobs (Irwin Jacobs), which may be a very difficult point to break. Steve Mollenkopf, the incumbent chief executive, Steve Mollenkov from the original processor designer to become a manager. It remains to be seen whether these emotional factors can overturn interest-driven claims.
Whether the takeover succeeds or not, Qualcomm's share price will be frustrated
After CNBC received a message that the company was prepared to hostile takeover of rival chip maker Qualcomm, a Wall Street analyst Team believes there is a downside risk to high pass stocks, whether or not the deal succeeds.
Kevin Cassidy, a Stifel Nicolaus Semiconductor stock analyst, downgraded its high-pass stock rating in California State Santiago from buying to holding, while raising its target price from $65 to $75. He expects the share price to rise 12.7% in 12 months.
In a research report to clients, Stifel analysts wrote that a hostile takeover by Qualcomm could lead to lengthy negotiations, leading to a slide in Qualcomm's share price. Two weeks ago, Qualcomm rejected the initial bid for the Bo Tong because the purchase price was too low. It is now reported that Bo is preparing to nominate a group of people to replace Qualcomm's current board of directors.
The uncertainty brought about by the Apple, Eun-zhi
If the acquisition is unsuccessful, Cassidy believes that the Qualcomm stock valuation will lose One-fourth, down to $50, a price close to the $54.84 close on November 2, before the news of the potential takeover.
If it succeeds, analysts say the stock is likely to be discounted in 2018, as the takeover is likely to face an enlarged regulatory review and a delay of 12 months until a final deal can be closed. He believes that Qualcomm stock prices will be ' up to $70 to $80 per share '.
These two bad results, coupled with the ongoing legal battles between Qualcomm and Apple, and the uncertainty over Qualcomm's takeover of the company will make Qualcomm ' bogged down '.
5. Division: Intel claims to have twice times as strong as the driving chip Nvidia;
Set micro-network news, according to foreign media reports, Intel CEO Division 29th that Intel developed EyeQ5 SOC Chip, in the subsidiary Mobileye Advanced Technology Plus, the depth of learning efficiency is 2.4 times times the opponent.
Nvidia's October launch of a new generation of Pegasus (Pegasus), which supports the highest level of automatic driving standards, is expected to be used by unmanned taxis, and Nvidia is thought to be the driver of a driving chip.
However, EYEQ5 supports full autonomous driving (level 5), and Mobileye estimates that 2020 will have a EyeQ5 chip from the car listing. Intel and the Google parent company Alphabet, BMW and auto parts supplier Delphi have established a partnership in self-driving.
6. MediaTek pay, early dividend "year-end determination"
Netizens share in PTT, MediaTek (2454) adjusted the salary algorithm, the subsidy into the salary calculation, on behalf of the year-end bonuses will be followed by more, and the dividend calculation method also changed "early to take", netizens have praised "double Cai after a lot of policy than before", shouting "Thanksgiving Ming-Jie!" Praise and practise! 」.
The Netizen in PTT pastes the text "M to adjust salary? "To share what I've heard?" the payroll algorithm merges the benefits and gets more at the end of the year. And the dividends have to change, there is gossip? 」
The result seems to be to catch some insiders to share, "year-end determination more", but the overtime will not be more controversial, but it can be determined that the salary "next big outbreak","overall, next year must be more than this year."
In addition, "the relatively big change is the dividend early to take, do not have to wait for next year","next year the end of the first half of next year's bonus."
Netizens have praised "Cai Dongfu heart to","entrepreneur Model","double Cai after a lot of policy than before good","Thanksgiving Mingjie praise","Buddha Heart Company." (Financial centre/comprehensive report)
Apple Daily