Set micro-grid launch WeChat IC: "Every day IC", big news immediately released, every day IC, every day set micro-network, micro-into a! Scan the end of the two-dimensional code to add attention.
1, IC application-driven automotive electronics market strong growth in 2021 will account for 9.8% of the global sales of electronic systems;
According to the 2018 edition of IC Insights data report, from 2016 to 2021, sales of automotive electronics systems is expected to rise 5.4% CAGR, the six major terminals The highest user system category (Figure 1).
figure 1
With the increasing demand for new car electronic systems, people are increasingly concerned with autopilot technology, (V2V) and vehicle-to-infrastructure (V2I) communications, as well as automotive safety, convenience and environmental features, including This shows that people are increasingly interested in electric vehicles.With the widespread use of these technologies in mid-range and entry-level cars, and the strong promotion of aftermarket products, automotive electronics products are in a growing situation.For semiconductor suppliers , After all, many of these automotive systems require analog ICs, microcontroller MCUs, and a large number of sensors, as shown in Figure 2. The global market for electronic systems in 2017 is estimated at $ 1.49 trillion and the estimated number of vehicles Market share accounted for 9.1%, up from 8.9% in 2015 and 9.0% in 2016. The share of automobiles in the global production of electronic systems is expected to increase only slightly in 2021 as a share of the total electronics market. Automotive electronics Products account for 9.8% of global sales of electronic systems.Although many electronic systems are being added to the new car, IC Insights believes that during the forecast period, Under pressure from ICs and electronic systems, the share of automotive end-use applications far exceeds the current share of the entire electronic system.
figure 2
The IC market drivers are also relevant for other electronics and integrated circuit markets and the auto market is expected to be the fastest growing market for electronic systems by 2021. This is good news for the entire automotive IC market and is expected to be 2017 The year will grow by 22% and by 2018 by 16%.
Industrial electronics systems are expected to reach the second-fastest rate of growth (4.6%) by 2021 as robots, wearable health devices, and IoT-promoting systems are pushing aggressively for growth in this area.
It is estimated that analog ICs will occupy 45% of the industrial IC market by 2017. CAGR of communication systems in 2016-2021 is expected to reach 4.2% as sales of smartphones and other mobile devices in the world are saturated, and it is expected that the growth of regional communications systems in the Asia Pacific region The most powerful, 2017 will account for 69% of the total communications IC market.
The consumer electronics market is expected to show a CAGR of 2.8% by 2021. The logic chip will be the largest consumer IC market for the entire forecast, with the consumer IC market projected to grow 2.4% CAGR in 2016-2021.
The average demand or marginal demand for personal computing devices (desktops, laptops, tablets) is expected to lead to a low CAGR of the computer system market by 2021. It is estimated that in 2017, due to the average price of computer DRAM and NAND flash memory Different levels of improvement, the total market for computer integrated circuits will also be pushed 25%.
2, fingerprinting into the screen will be a new trend;
Apple recently canceled the fingerprint recognition in the iPhone X and switched to face recognition, a move that led many analysts to see the development of bad fingerprinting technology.However, the fingerprinting module maker Kim Jong Technology pointed out that the current Bank ATMs have also adopted large-scale fingerprinting technology, which is sufficient to prove that the market has a high degree of trust in fingerprinting. In addition, fingerprinting next year will smoothly integrate with the monitor to form a screen recognition scheme. Therefore, it is bound to face facial recognition Long-term coexistence in the market.
Xiao Wenxiong, senior vice general manager of Jinjian Technology, said that in terms of smartphone applications, face recognition and fingerprint identification will exist simultaneously in 2018, and face recognition at the Android camp will have to be quite difficult at this stage to achieve 3D. So it will use 2D technology for face recognition, supplemented by the primary authenticity recognition to prevent image fake.In view of this hybrid technology architecture, the current Jin 佶 has a corresponding solution.
Xiao Wenxiong further pointed out that in 2018, fingerprint identification will make great strides toward screen recognition, that is, the design framework for the direct integration with the display.At present, most mobile phones are fingerprinted on the Home key, Allowing fingerprints that would otherwise have to be exposed to hide beneath the screen, and as the additional area is no longer required to install the sensor, the proportion of the phone's screen becomes larger.
In addition, supply chain sources pointed out that the reason why Apple gave up fingerprint recognition in iPhone X is mainly because the technical maturity of fingerprint recognition on the screen can not catch up with the mass production of iPhone X, so iPhone X can not keep fingerprinting Features For the iPhone X development team, having a completely different user experience across the full screen is the supreme guideline.
For this argument, Xiaowen Xiong frankly, the current optical fingerprint recognition maturity and capacity indeed can not meet the needs of smart phones to meet the level, but until 2018, optical screen fingerprint recognition technology will be quite mature. With the current status It seems that mass production in the second to third quarters of 2018 has been identified as having no problems, therefore, fingerprint identification and face recognition will still have the opportunity to coexist in the market in the future and will not result in the situation of face recognition elimination of fingerprinting.
In fact, fingerprinting applications in the smart phone business will not disappear, but must be converted to other technical principles, in order to allow mobile phone operators to design a "hidden" fingerprint recognition, leaving more space for the screen.In addition to optical technology, Chinese mobile phone brand Vivo also released invisible fingerprint technology based on Qualcomm's ultrasonic fingerprint sensing in Shanghai MWC earlier this year. Qualcomm's latest ultrasonic fingerprint sensing technology can penetrate a cover glass thickness of 800μm, or Aluminum sheets up to 650 μm thick outperform previous generation technologies that penetrate only 400 μm glass or metal, and Qualcomm predicts that the terminal will be available in the first half of 2018.
Xiao Wenxiong analysis, the difference between ultrasonic and optical fingerprinting solutions lies in the fact that the technical difficulty of the ultrasound is really high. Because the directivity of the ultrasound is poor, the emitted wave is divergent and needs to be transmitted through a fixed sound speed To calculate the difference between the fingerprint peak and trough, and then use the sonic sensor to receive the principle of the optical scheme is similar to the camera through the CMOS image sensor to take the image, with each person's fingerprint unevenness, so the brightness of the image is not The same, and then come to different fingerprints of the picture
3, research: the second half of next year DRAM growth will stop production growth of only 11.8%;
Continued DRAM gains, sequelae gradually emerged, channel providers said that due to excessive memory growth, downstream applications, especially the laptop and some smart phones and other markets, eroded profits, has begun to reduce the carrying capacity to resist, plus some New output one after another in 2019 output, research that the second half of next year, DRAM gains will be halted, the price downward pressure on repair.
According to Gartner's latest research report, Samsung, SK hynix and the mainland Fujian Jinhua, Hefei Rui Li and other companies increase production plans for the future DRAM market dropped a new variable, DRAM output may grow 11.8% next year, though still, However, growth has fallen below 67.8% this year. By 2019, with an increase in new production capacity and a resumption of price war, the annual output value will decline by 25.9% and there are two years of estimated killings.
Memory access providers said Samsung and other major memory maker channel is still up under DRAM contract price, DRAM prices in the first quarter of next year is still expected to remain high, however, DRAM from the second half of last year rose after six seasons have risen, More than 70% increase this year, many downstream applications have end too much.
Although the industry still stresses that the market is in short supply, it will not be able to be solved by next year. It is estimated that about 21-22% of DRAM supply will be added next year while the demand growth will reach 22-23% with a still short supply. DRAM Rongjing Let Samsung, SK Hai Lux and Micron have made brilliant earnings every quarter of this year. However, DRAM prices have risen too much. It is estimated that by the first quarter of next year, the price and the ratio of the beginning of this year will increase by as much as 80% and will also break the industry's moderation expansion plan.
For example, Samsung announced earlier plans to convert South Korea's Hwaseong Plant Line 16, the original production for the production of 2D memory storage, to produce DRAM, monthly increase of 80,000 to 100,000, total production capacity from the current 390,000 Increase up to 500,000, the output of the first quarter of next year began to release.
In addition, SK Hynix also announced that it will invest 8.6 billion yuan in Wuxi to build a second wafer fab, which is expected to put into operation in 2019. It is expected that the yield-increasing effect may prevent the buying trend from turning around in the second half of next year.
4, the strong demand for silicon wafer global crystal, Taiwan Sheng branch market Wang;
Semiconductor silicon wafer market demand is strong, the price of the first quarter of next year to determine the rise of about 15%, driven by Global Crystals (6488) last Friday (10) direct attack daily limit, closed at 368.5 yuan; and Taiwan Branch (3532) Is benefited from one of the stocks, the market optimistic about the theme of price increases, revenue, profits are expected to grow simultaneously.
According to the industry analysis report of Silicon Manufacturers Group (SMG) of the International Semiconductor Industry Association (SEMI), the global silicon wafer shipments in Q3 reached 2,997 million square inches for 6 consecutive quarters Shipments hit a record high record.Because the silicon wafer in short supply, the price of the first quarter of next year, up about 15%, will be bullish until the second half of next year.
With the new capacity of the mainland semiconductor plant one after another out, and 8-inch silicon wafer is also panel-driven IC, power management chip, fingerprint identification and sensor-related components, the growth driven by the global crystal October revenue 3.942 billion yuan, Minus 5.9%, with an annual growth of 175.7%. Total January-October revenue was 37,707 million yuan, an increase of 184.7%.
Japanese silicon wafer giant SUMCO recently gains significantly, SUMCO is the largest shareholder of Taishwondo, the subsidiary SUMCO TECHXIV Co., Ltd. holds a 47% stake in Taisheng Branch SUMCO last quarter earnings, demand for all sizes of silicon wafers, driven Revenue from the previous quarter: Taishengke reported a revenue of 1.132 billion yuan in October, a 2.3% increase over the previous year, an annual increase of 19.7%. The revenue from January to October was 10.522 billion yuan, a year-on-year increase of 17.8%, a record high.
5, do not know the power of Confucianism? Supply chain guarantee, the rapid development of the new Chinese website services in China market;
One of the most important factors in the thriving electronics industry is the stable and reliable supply chain of electronic components, which is the key for enterprises to get enough electronic components to successfully complete the project.
As China's electronics industry continues to thrive, it has attracted major overseas distributors to grow its business domestically, including Rutronik Elektronische Bauelemente GmbH, which provides a fixed supply of 28 billion for a wide range of fields and applications Kinds of components to ensure that customers can meet different needs of different categories.
Founded in 1973 in Germany, the distributor of electronic components has successfully developed from one-person business to large-scale enterprise after many years of hard work. Now it has become the third-ranked and tenth-largest distributor in Europe, mainly in the automotive, medical, Semiconductor, passive and electromechanical components and displays, embedded motherboards, storage solutions and wireless solutions are available to target markets such as industrial, home appliances, energy and lighting.
Ruudzhuar has spared no effort in promoting sports development. It not only supports professional racing teams but also actively supports students in automotive electronics. For example, the company has been supporting e-Tech Racing Team, which consists of 50 students from University of Barcelona. The single-seat electric race car manufactured by the team leads in all competitions, including gaining valuable experience in formula racing in Spain in August and winning a place in the Czech Republic in September.
To tie in with the company's strong goal of opening up its business in Asia, especially in China, Ruud Zhuang sponsored the HCB-Rutronik racing team to participate in the Macau Grand Prix since last year to establish closer ties with its local customers. At the same time, the company established The new China website (www.rutronik.com.cn), hoping to better serve domestic customers.
The distributor also features the Rutronik24 eCommerce Platform (www.rutronik24.com), which offers online product catalogs and advanced datasheets, smart search capabilities, volume quoting, and purchasing advanced features. The site is truly unique around the world 24/7 service and 24/7 customer support through a sales force in the country One of its special features is not only focused on the needs of large customers but also those who support smaller orders in the hope of motivating 'small customers' to grow and change For larger buyers, this support is commendable for many small and medium-sized electronics companies in development in the country.
In order to establish closer relations with electronics industry professionals and help engineers gain experience and knowledge of key trends, needs and specific issues in order to facilitate and successfully complete the project, Ruozhu Li also opened a WeChat account (WeChat ID: Rutronik_Electronics) and more Quickly provide the latest news, product services and industry trends.
As long as you pay attention to our account by December 31, 2017, you will have the opportunity to get a small gift, do not miss! To understand the participation method, please visit Ruzhu WeChat page.
6, this is Intel / AMD co-create the real reason for the new CPU! Nuclear significant goodbye;
Looking back ten years of PC chip industry, SIS obscurity, S3 attributed to VIA eventually part of China KC, Matrox and Imagination integration of outsiders have no soul, veterans continue to dying .There are many people who like this evolution as the Han and the Three Kingdoms Treacherous battle stories, as if the remnants of Intel, AMD, NVIDIA are highly personal palace fighting experts .This kind of story reading is actually more boring.
The processor chip industry is the laurel of the semiconductor industry, there is not much "human conspiracy" factor in this shift. 'I destroy you, what is it with you', 'The law of numbers determines everything, 'This is the law of the development of the processor industry.
In the processor industry, the company profits from the hard and soft ecological, hard and soft eco-dependent basic design, technical design with hard power to speak. No matter how the players turn the towel bottles, vertical and horizontal, around the basic strength of the game. The Intel / AMD joint show, fundamentally speaking, is a typical conspiracy, there is no need to whitewash things.
We are here to talk about Intel / AMD in heterogeneous CPU issues, and briefly answer the following:
Question 1: CPU / GPU heterogeneous fusion of how much is the need? Answer: For game graphics is not meaningful, for the AI, the cloud computing is uncertain.
Second: CPU / GPU heterogeneous fusion will oppress the graphics card market, what is the difference between long-term and short-term? Answer: Intel, AMD have their own integrated graphics have been tapped the maximum market potential, differentiation has been completed, regardless of future cooperation between the two How, within three years will not really change the pattern of the PC market.
Intel's hybrid packaging technology is the foundation
Third: Intel, AMD in the heterogeneous level, what kind of development prospects, the latest convergence of the two products on the desktop, notebook, one machine market, what kind of direct impact? Answer: Intel needs AMD's GPU to continue their Iris Pro Unfinished product line, Iris Pro Impermanent area AMD's GPU is difficult to bring special improvements. Intel, AMD's latest converged processor may be limited impact less than 2kg heat dissipation less frivolous games in this market, but its main Focus is still in one machine, NUC and Apple products.
Fourth: Intel and AMD integration of the product, the continuous development of their own product line how much? Answer: Built-in GPU is doing two things, built-in high-performance GPU hybrid processor is difficult to promote the popularity of both Things. The Ryzen APU's complete body may have some market overlap with the new converged products, but with little impact. Intel's existing product line almost does not consider how much to improve for the integration.
To put it bluntly, the integration of the eighth generation of Core H series mobile processors and AMD high-performance Radeon GPUs is an event that is more attractive to the industry but there is no prospect that the internal driving force will really change the overall PC market Perhaps this is a positive seed, AMD will bring more profits to weather the storm, Intel's 'big chess' to provide some help, but this may become a high probability of fantasy.
Necessity of heterogeneous fusion
Heterogeneous fusion is AMD's CPU development direction after the acquisition of ATI in 2006. Considering the weak performance of Intel's GMA series integrated graphics, it is doubtless that the AMD Fusion strategy is targeted and very targeted.
Differences in logic and data throughput between CPUs and GPUs make the temptation of heterogeneous general purpose processors very large - for the GPU to be given to the GPU, to the CPU for the CPU, to design a suitable bus, to cache the communication mechanism Enough.
However, this beautiful idea has not been fully fulfilled for ten years.Due to the defects of software environment optimization, GPU has a very weak presence in the desktop common application, and even applications that support GPU acceleration often fail to make full use of GPU Performance, the real GPU can be used to scientific computing, big data and other projects have entered the commercial area, with the basic needs of consumer-level do not take the edge.
Intel embarked on the road of nuclear significant
Therefore, AMD has long been, whether AMD's own APU or Intel's later follow-up 'core' graphics card, the selling point can be proposed only about three:
One: HD hardware performance GPU advantage. Compared to high-performance discrete graphics, integrated GPU heterogeneous processor can save power and PCB area, thereby reducing the size of the product. Second: the face of entry-level users, the built-in GPU has been To meet the vast majority of people's 3D graphics entry-level needs, browsing the web, running outdated games is not a big deal, but very ordinary appetite. Third: the face of the needs of professional users, application software on the GPU Accelerated demand is limited, often only requires less than the strong, alone significantly less advantage, built-in GPU heterogeneous processor power consumption, design cost advantage.
However, with the passage of time, desktop 1080P or even 4K video decoding without having to consider the issue of soft solution hardware solutions, nuclear show enjoy the advantages of hard-ware power saving may also have, but not much.Really need to MadVR rendering decoding Of users, the flagship graphics card to meet, even less likely to consider nuclear display.
Given the ever-decreasing iteration of the AMD architecture and the never-ending graphics stability Intel has ever had, the entire built-in GPU industry will not find a truly functional high-performance product until Ryzen APU is introduced in 2017. Because of the expertise of GPUs Accelerate the optimization efforts to increase, the nuclear showed a limited comparative advantage is completely lost.
For the most part, the significant advantages of nuclear visualization, including AMD's single-in-one display, only save space and extend notebook life.
The concept of Fuison has never been more meaningful to end users
Although, the launch of Ryzen APU and Intel Kabylake-G series (merged AMD Radeon PoLaris) to a certain extent, the performance of integrated GPU to take a big step, get rid of the embarrassment of years of stagnation.
However, both the cost of packaging or future technological iteration speed can not be truly compared with the discrete graphics, consumer-level integration heterogeneous processor demand has never been rigid, is a useful, but not nor is not unacceptable state, This state will continue.
Watching video CPU usage is not a big deal
Faced with the necessity of end-consumer users is not equal to the necessity of all industries. Under the influence of Google NPU, Cambodian Core in China, under the guidance of Apple iphone, ARM and NVIDIA campuses such as QUALCOMM under external pressure, Intel's move to buy a big chess game like FPGA is not enough to solve all the problems.
Intel's breakthrough at the GPU level is indeed an inevitable requirement as it is a requirement for cutting-edge macro-industries such as deep learning, artificial intelligence, big data computing, etc. However, this need is product specific at the product level , There is no consumer-grade product profits and design practices, no objectionable beauty.
The combination of Intel and AMD, indeed, will certainly be considered in this regard, which for the patent, technology exchange details are unknown.We know that Intel is not willing to hype, AMD also agreed to hope that a low profile I / A compression NV? To be calm
I / A this joint, rendering more exaggerated by the media, half a century (48 years) for the first time since the cooperation, secretly crowded NVIDIA is a foregone conclusion of the judge were released.This attractive approach is still more successful, but Not objective.
Ten years ago AMD acquired ATI, Intel's chipset camp not only Sis, there are ATI, ATI Core Duo series of out-of-print chipset in Ebay but can not buy high-priced items.
Conversely, NVIDIA's Mr. Jen-Hsun Huang was originally a former employee of AMD, a rumor is not what he wants to become CEO of the acquisition of AMD is NVIDIA NVIDIA for a long time also designed for Intel and AMD motherboard chipsets, early years There are even models such as the MCP79 that have graphics cards for Intel.
Intel and ATI jointly against AMD and NVIDIA PC industry was the hottest year of the grand scene.Even now, AMD CPU ashes fans group there are many predecessors more favor NVIDIA graphics card, AMD public relations staff also have a deep understanding.
NVIDIA has no X86 CPU, GPU relatively strong lead; ATI was acquired after the integration of AMD is only part of the RTG graphics department later formed a competitive disadvantage with NVIDIA; Intel since i740 in 1999 no longer launched a discrete graphics core graphics card problem Heavy, energy consumption than the tragic - this is the situation we are facing.
Historically, Intel had hit NVIDIA's motherboard chipset business with a patent stick and had paid huge sums of money to AMD to help the latter tide over the storm.
Intel's Cross Fire patent license from ATI, and the AMD GPU division often introduced the Intel Core i7 platform game console, CPU department and promotion in the always with the potential of NVIDIA graphics together... This is the history of the Spring and Autumn nonsense Excerpts of the battle-style three historical romances.
AMD is not a firm but a single personality carrier but the body will issue multiple voice interest groups
In general, AMD has been the Intel standard follower most of the time since AMD was designated a subcontractor and is not unacceptable to subordinates.
Intel and AMD have a common interest in X86 business bigger and stronger, the compromise between the two is inherent, as much as the competitiveness of the two.NVIDIA does not have the X86 business, from the outset can not embark on I / A Well-designed Boulevard, the graphics card and then stronger to expand their sphere of influence, which is more than a decade has been well-known thing.
Even if AMD merged ATI, but also set up RTG, GPU technology has just resigned from AMD Daniel Raja for a long period of time has a semi-independent RTG sector leadership, even the yellow in his interview to describe him to switch to Intel AMD will bring great losses.
AMD internal is not a piece of iron, which co-operation with the unequivocal, and not too difficult to understand about the "pacing around." Microsoft in the Apple conference platform for the Office does not prevent the release of the Surface for MacBook, AMD and Intel It is also very natural to engage in 'basic', and it is not necessary to have much to do with NVIDIA.
Intel and AMD never cut the chaos
Intel recognized the basic direction of AMD's convergence and the need for powerful GPU to achieve consumer-level breakthroughs, or even more GPU-based graphics technology in the future AI artificial intelligence, in-depth layout of advanced learning, it is unmistakable conspiracy. AMD graphics The department has always had its own voice, and AMD as a whole is inextricably intertwined with Intel, all of which are public intentions that are not covered up at all.
Since ancient times, strategists, not many people intent to achieve, no matter what kind of cooperation Intel and AMD for their own development, have escaped the inherent pattern of chip design.
Intel can not figure out ATI / NVIDIA's monopoly of graphics patents and GPU technology. AMD can not afford to take X86 alone. Instead of having I / A co-operate for NVIDIA's intentions, NVIDIA does not have its own vertical integration channel Technical puzzles inevitably occupy a relatively independent position. Core Radeon direct impact
Intel announced the new Core Duo information has been disclosed almost the same, there are two models known i7-8809G and i7-8805G. Both are Kabylake architecture four-core eight-thread design, embedded multi-die interconnect bridge (EMIB) approach will be GF foundry AMD Radeon (new sources pointed out that it should be Polaris Polaris series) strong GPU hybrid package into the CPU.
Different positioning of the new i7 have different frequencies and the corresponding AMD series GPU, the same is that they will have built-in HBM 2 high bandwidth video memory (up to 4GB 1024bit bit width.) Intel CPU with AMD graphics super heterogeneous CPU has finally become a reality Rumor is no longer a rumor.
Intel has the highest known model CPU reference frequency of 3.1GHz, the highest Core frequency of 4.1GHz, and desktop mainstream Core Duo similar 7. GPU, AMD GPU GPU rumors earlier is VEGA 24 core, later confirmed as the Polaris family , With up to 1536 stream processor units, up to a maximum frequency of 1190MHz.
Relevant results have been leaked a lot, this heterogeneous processor 3Dmark 11P file GPU score more than 14,000 points, close to the NVIDIA MaxQ version GTX1060, and the desktop version of the Polaris 11 RX 470 / RX 470D theoretical performance close to (unit Stream processor performance more), the lower limit is also NVIDIA GTX1050Ti or RX 560 (R9 380X) above.
This performance is really staggering, is a very dramatic performance leap.
Games, 1080P low-effects i7 8809G singularity ashes reached 62 frames, and the desktop version of the RX 470 is almost the same, running mainstream games should be no problem.
Just yesterday, equipped with this technology platform NUC one motherboard PCB has appeared, and we imagine the same, this motherboard is very high level of integration, work fair and deft.
So this heterogeneous processor in the product form for what product line, what kind of competitiveness will it?
In our opinion, the adaptability of this competing product line ranged from professional computing platform to all-in-one (especially Apple) platform, high-performance ITX platform such as NUC, and medium-sized pan-thin entertainment gaming platform.
This is because, although Intel has designed the specialized EMIB interconnection channel, the communication efficiency even surpasses the traditional way between AMD's own GPU and HBM2 memory, but in essence, this is not the Intel architecture. Between the GPU and the CPU The communication is still independent, only to replace the PCI-E bus in other ways (there are rumors that the original HD620 part of the Die in the Intel CPU still exists.) This degree of heterogeneous packaging is not a true fusion, Probably only Intel quad-core mobile I7. And mobile RX, 570 graphics compact package.
Although packaged together as a single heat source, compared to the original I + N combination is conducive to the placement of heat fins, but can not reduce the thickness too much, only to reduce the vertical projection area of the notebook from the horizontal, so this hybrid chip you want to do Into a real 13, 14-inch thin and light notebook is impossible.
Even though the power consumption of the AMD GPU in the low-frequency state is not bad at the moment, the pressure on the heat sink will be severely exaggerated because he is too close to the Intel CPU.
Currently in the Razer, Gigabyte, MSI 14, 15-inch thin and light on the service of the i7-7700HQ itself on the brink of heat dissipation is on the verge of falling, often down the state, and then plugged into an AMD GPU can only exacerbate the CPU cooling burden.
Intel and AMD's masterpiece
What's more, HBM is still an expensive technology product, only appearing on AMD's flagship high-end card. Intel's hybrid chips do not save as much crystal area as discrete graphics, but at the cost of packaging and design cost.
This hybrid chip is definitely more expensive than the same performance of the graphics card a lot, and can put his product must be high enough profit margins Caixing - only the boss did not discuss the entry of thin games or XPS series of entertainment only There is such a premium treatment, otherwise it can only return to the traditional NUC and one machine line.
The game market capacity of about one-eighth of the notebook market, the desktop market does not require such a high degree of integration of things; NUC and one machine has always been the same for many years the popular market.
Core H + Radeon combination only eight thousand to twenty thousand yuan in the frivolous games get some influence in this market, due to the absolute performance of the GPU has no advantage, only four core CPU to face the eighth generation Core Duo standard six-core voltage Products are also vulnerable, these chips can only be temporarily in the meaning of 'meaning' of the trial stage, would like to squeeze out from the marginalized market will undoubtedly be very difficult.
As for the next platform development, due to the Polaris and Vega unit GPU stream processors are very close performance, energy consumption gap is not significant, AMD's next-generation Navi platform mature release is close to a year, Intel's built-in energy consumption ratio The huge Iris Pro series is temporarily shut out, and the level of integration we see so far is not hard at all until at least early 2019.
According to Intel's latest roadmap, in 2018, the Coffee Lake H family of gaming notebook platforms will popularize six-core products and may have eight nuclear products by the end of the year. As the 10-nm maturation iteration may take time, To adhere to 14nm, power consumption will not allow and AMD's high-performance co-exist alone, so the current integration in the CPU / GPU are stable optimal solution.
Of course, Iris Pro did not completely die, GT4E in addition to the i7 6770HQ basic did not really appear, they also appear in the Coffee Lake H family, the game performance is far less than the integration of AMD GPU Kabylake-G series, the CPU On the contrary, the performance is stronger.
Six nuclear join the card and the quad core with the mainstream card how to choose? Perhaps next year we will face such a puzzled Intel mobile CPU purchase, in fact, not a bad thing is not? ZOL