Current Situation and Optimization of FPGA Heterogeneous Computing

The general CNN accelerated design based on FPGA can greatly shorten the FPGA development cycle and support the rapid iteration of service depth learning algorithm; It provides computing performance comparable to GPU, but has the delay advantage of orders of magnitude compared with GPU, and builds the strongest real-time AI service capability for business.

WHEN Current situation of deep learning heterogeneous computing

With the rapid growth of Internet users and the rapid expansion of data volume, the demand for computing in data centers is also rising rapidly. At the same time, with the rise of computing intensive fields such as artificial intelligence, high-performance data analysis and financial analysis, the demand for computing power has far exceeded the capacity of traditional CPU processors.

Heterogeneous computing is considered to be the key technology to solve this computing gap at this stage. At present, "CPU GPU" and "CPU FPGA" are the heterogeneous computing platforms most concerned by the industry. They have the advantages of higher efficiency and lower latency than traditional CPU parallel computing. In the face of such a huge market, a large number of enterprises in the science and technology industry have invested a lot of funds and manpower, the development standards of heterogeneous programming are gradually maturing, and the mainstream cloud service providers are actively layout.

The industry can see that giant companies such as Microsoft have deployed a large number of FPGAs to accelerate AI information. What are the advantages of FPGAs over other devices?

Flexibility: programmability naturally adapts to the rapidly evolving ML algorithm

DNN, CNN, LSTM, MLP, reinforcement learning, decision tree, etc

Arbitrary precision dynamic support

Model compression, sparse network, faster and better network

Performance: build real-time AI service capability

Improved low latency prediction capability compared to GPU / CPU

Improved single watt performance compared to GPU / CPU

Scale

High speed interconnection IO between boards

Intel CPU FPGA Architecture

At the same time, the short board of FPGA is also very obvious. FPGA uses HDL hardware description language for development, which has long development cycle and high entry threshold. Taking separate classic models such as Alex net and Google net as examples, it often takes months to customize and accelerate the development of a model. The business side and FPGA acceleration team need to consider both algorithm iteration and adapting FPGA hardware acceleration, which is very painful.

On the one hand, we need FPGA to provide low latency and high-performance services that are competitive compared with CPU / GPU. On the other hand, we need the development cycle of FPGA to keep up with the iteration cycle of deep learning algorithm. Based on these two points, we design and develop a general CNN accelerator. Considering the general design of mainstream model operators, the compiler generates instructions to drive model acceleration, which can support model switching in a short time; At the same time, for the emerging deep learning algorithm, the rapid development iteration of relevant operators is carried out on this general basic version, and the accelerated development time of the model is reduced from a few months to one to two weeks.

The overall framework of the general CNN accelerator based on FPGA is as follows. The CNN model trained by Caffe / tensorflow / mxnet and other frameworks generates the instructions corresponding to the model through a series of optimization of the compiler; At the same time, the image data and model weight data are preprocessed and compressed according to the optimization rules, and then distributed to the FPGA accelerator through PCIe. The FPGA accelerator works completely according to the instruction set in the instruction buffer. The accelerator executes the instructions in the complete instruction buffer once to complete the calculation and acceleration of a picture depth model. Each functional module is relatively independent and is only responsible for each individual module calculation request. The accelerator is separated from the deep learning model, and the data dependency and pre and post execution relationship of each layer are controlled in the instruction set.

In short, the main work of the compiler is to analyze and optimize the model structure, and then generate the instruction set efficiently executed by FPGA. The guiding ideology of compiler optimization is: higher MAC DSP computing efficiency and less memory access requirements.

Next, we take the Google net V1 model as an example to make a simple analysis of the accelerator design optimization idea. The network of inception V1 combines 1x1, 3X3, 5x5 conv and 3x3 pooling stack, which increases the width of the network and the adaptability of the network to scale. The following figure shows the basic structure of inception in the model.

Data dependency analysis

This part mainly analyzes the pipelined and parallelized computing in the mining model. Pipelining design can improve the utilization of computing units in the accelerator, and parallel computing can use as many computing units as possible at the same time.

As for pipelining, the analysis part includes the operation of loading data from DDR to SRAM on FPGA chip and the pipelining calculated by PE. Through this optimization, the memory access time is overlapped; DSP calculates the calculation control process of the whole column to ensure the improvement of DSP utilization.

As for parallelism, we need to focus on analyzing the parallel relationship between PE computing array and "post-processing" modules such as activation, pooling and normalization. How to determine the data dependency and prevent conflict is the key to the design here. In inception, it can be seen from its network structure that 1x1 convolution calculation in branch A / B / C and pooling in branch D can be calculated in parallel, and there is no data dependency between them. By optimizing here, the calculation of 3x3 Max pooling layer can be completely overlapped.

Model optimization

In the design, two aspects are mainly considered: finding the optimization of model structure and supporting the fixed-point adjustment of dynamic accuracy.

FPGA is a device that supports a large number of parallel computing. Looking for higher dimensional parallelism from the model structure is of great significance for computing efficiency and reducing memory access. In inception V1, we can see that the input data of 1x1 convolution layer in the first layer of branch a branch B branch C is completely consistent, and the strip and pad of convolution layer are consistent. Can we align and overlay on the output feature map dimension? After superposition, the memory access demand for input data is reduced to 1 / 3 of the original.

On the other hand, in order to give full play to the acceleration characteristics of FPGA hardware, the influence process of the model needs to carry out fixed-point operation on the model. In FPGA, the performance of int8 can be twice that of int16. However, in order to enable customers in the company and on Tencent cloud to deploy their trained floating-point model without perception, and do not need retrain int8 model to control accuracy loss, we adopt a customized int16 scheme that supports dynamic accuracy adjustment. Through this method, the user trained model can be deployed directly through the compiler without any loss of accuracy.

Memory architecture design

Bandwidth is always one of the bottlenecks restricting the performance of computer architecture. At the same time, memory access directly affects the power consumption efficiency of accelerator devices.

In order to minimize DDR memory access during model calculation, we designed the following memory architecture:

Input buffer and output buffer ping pong are designed to maximize pipelining and parallelism

Support the inner copy operation between the input buffer and the output buffer itself

Cross copy operation between input buffer and output buffer

Through this architecture, for most current mainstream models, the accelerator can hold all the intermediate data on the FPGA chip. In addition to the loading of model weight, the intermediate does not need to consume any additional memory operation. For the model that cannot completely store the middle tier feature map on the slice, we introduce the concept of slice slice slice in the channel dimension and part slice in the feature map dimension. Through the compiler, the primary convolution or poolingnorm operation is reasonably split, and the DDR memory access operation is pipelined with FPGA accelerated calculation. On the premise of giving priority to ensuring the DSP computing efficiency, the memory access demand of DDR is minimized.

Calculation unit design

The core of the general CNN accelerator based on FPGA is its computing unit. The current version of the accelerator is designed based on Xilinx ku115 chip. The PE computing unit is composed of 4096 MAC DSP cores working at 500MHz, with a theoretical peak computing capacity of 4tflops. Its basic organizational framework is shown in the figure below.

Ku115 chip is stacked by two die pairs, and two groups of processing units PE are placed in parallel in the accelerator. Each PE is composed of four groups of Xbar composed of 32x16 = 512 MAC computing DSP cores. The key of the design is to improve the data multiplexing in the design, reduce the bandwidth, realize the multiplexing of model weight and each layer feature map, and improve the computing efficiency.

Application scenarios and performance comparison

At present, the mainstream of deep learning uses GPU for the training process of deep learning. When deploying online inference, it is necessary to comprehensively consider the characteristics of real-time, low cost and low power consumption and select the acceleration platform. Classified by deep learning landing scenarios, advertising recommendation, speech recognition, real-time monitoring of picture / video content belong to real-time AI services and real-time low-power scenarios of terminals such as intelligent transportation, intelligent speakers and driverless. Compared with GPU, FPGA can provide strong real-time and high-performance support for business.

What is the platform performance, development cycle and ease of use for users?

Acceleration performance

Taking the actual Google V1 model as an example, the CPU test environment: two 6-core CPUs (e5-2620v3) with 64g memory.

When the CPU of the whole machine is full, the performance of a single ku115 based accelerator is 16 times higher than that of the CPU, the detection delay of a single picture is reduced from 250ms to 4ms, and the TCO cost is reduced by 90%.

Meanwhile, the prediction performance of FPGA is slightly better than that of NVIDIA GPU P4, but the delay is optimized by an order of magnitude.

development cycle

The general CNN FPGA acceleration architecture can support the deep learning model in the rapid iteration and continuous evolution of business, including classic models such as Google net / VGg / RESNET / shufflenet / mobilenet and new model variants.

For classical models and algorithm variants based on standard layer self-developed, the existing acceleration architecture can already support, and the corresponding instruction set of the model can be implemented through the compiler within one day to realize deployment online.

For self-developed special models, such as asymmetric convolution operator and asymmetric pooling operation, iterative development of relevant operators needs to be carried out on this platform according to the actual model structure, and the development cycle can be shortened to support within one to two weeks.

Ease of use

FPGA CNN accelerator encapsulates the underlying acceleration process and provides easy-to-use SDK to the business parties of the acceleration platform. The business party can complete the acceleration operation by calling simple API functions, and there is almost no change to the business logic.

epilogue

The general CNN accelerated design based on FPGA can greatly shorten the FPGA development cycle and support the rapid iteration of service depth learning algorithm; It provides computing performance comparable to GPU, but has a delay advantage of orders of magnitude compared with GPU. The general RNN / DNN platform is under intense research and development, and the FPGA accelerator builds the strongest real-time AI service capability for the business.

In the cloud, at the beginning of 2017, we launched the first domestic FPGA public cloud server in Tencent cloud. We will gradually introduce the basic AI acceleration capability to the public cloud.

The battlefield of AI heterogeneous acceleration is great and wonderful. Providing the best solution for the company's internal and cloud business is the direction of the continuous efforts of the FPGAs team.

If the online model needs to be changed, just call the model initialization function and initialize the corresponding model instruction set to FPGA. The acceleration service can be switched in a few seconds.

About the author:

Derickwang (Wang Yuwei), graduated from Huazhong University of science and technology in 2014, with a master's degree. His main research and practice direction is FPGA heterogeneous computing in data center, and his project

Current Situation and Optimization of FPGA Heterogeneous Computing 1

GET IN TOUCH WITH Us
recommended articles
FAQ
Apple still Uses Cheaper LCD Screens and Decides to Produce Them by Itself, Endangering the Interests of Many Screen Suppliers
Apple still Uses Cheaper LCD Screens and Decides to Produce Them by Itself, Endangering the Interests of Many Screen Suppliers
It is reported that Apple plans to use cheaper LCD screens on most iPhone models next year, while delaying the full conversion to high-tech organic light emitting diode (OLED) screens, which also reflects consumers' price sensitivity.According to the Wall Street Journal, apple is expected to launch an LCD model and two OLED models this fall, and the LCD model is expected to account for most of its iPhone sales. It is expected that Apple will continue to use LCD next year, which is completely inconsistent with the previous report of a complete transition to OLED.Apple announced in March that it was testing its internal microled screen.Apple's decision to insist on using cheaper LCD screens has brought more uncertainty to its screen suppliers; In addition, Apple's decision to produce its own screen endangers the interests of many suppliers.Since apple is not expected to use these products in its OLED screen, the shares of Apple's three LCD suppliers, including sharp, LG display and Japan display, have fallen sharply in the past year. Subsequently, when it was revealed in March that Apple was already testing its microled screen, the shares of many display suppliers, including Samsung, sharp and LG display, fell again. This also damages the interests of general display and Synaptics. After reports showed that Apple would use more LCD screens than expected, the share price of general display company, which provides OLED display materials, began to fall.The Wall Street Journal pointed out that Apple's decision to postpone the implementation of OLED screens does not necessarily bring opportunities for LCD suppliers. For example, LG display has planned to invest in OLED screens.Apple has been facing the problem of supply shortage in the past. If suppliers have to change their production plans, Apple may face supply shortage again. It is reported that Japan display raised $320 million in April to meet the expected demand for its advanced LCD.According to recent reports, Apple has communicated its transformation plan to its suppliers, including plans to reduce parts procurement from suppliers by nearly 20% this year. At the same time, the uncertainty of the continuous expansion of Apple's supply chain will also make the new iPhone more expensive than ever.
Apple still Uses Cheaper LCD Screens and Decides to Produce Them by Itself, Endangering the Interests of Many Screen Suppliers
Apple still Uses Cheaper LCD Screens and Decides to Produce Them by Itself, Endangering the Interests of Many Screen Suppliers
It is reported that Apple plans to use cheaper LCD screens on most iPhone models next year, while delaying the full conversion to high-tech organic light emitting diode (OLED) screens, which also reflects consumers' price sensitivity.According to the Wall Street Journal, apple is expected to launch an LCD model and two OLED models this fall, and the LCD model is expected to account for most of its iPhone sales. It is expected that Apple will continue to use LCD next year, which is completely inconsistent with the previous report of a complete transition to OLED.Apple announced in March that it was testing its internal microled screen.Apple's decision to insist on using cheaper LCD screens has brought more uncertainty to its screen suppliers; In addition, Apple's decision to produce its own screen endangers the interests of many suppliers.Since apple is not expected to use these products in its OLED screen, the shares of Apple's three LCD suppliers, including sharp, LG display and Japan display, have fallen sharply in the past year. Subsequently, when it was revealed in March that Apple was already testing its microled screen, the shares of many display suppliers, including Samsung, sharp and LG display, fell again. This also damages the interests of general display and Synaptics. After reports showed that Apple would use more LCD screens than expected, the share price of general display company, which provides OLED display materials, began to fall.The Wall Street Journal pointed out that Apple's decision to postpone the implementation of OLED screens does not necessarily bring opportunities for LCD suppliers. For example, LG display has planned to invest in OLED screens.Apple has been facing the problem of supply shortage in the past. If suppliers have to change their production plans, Apple may face supply shortage again. It is reported that Japan display raised $320 million in April to meet the expected demand for its advanced LCD.According to recent reports, Apple has communicated its transformation plan to its suppliers, including plans to reduce parts procurement from suppliers by nearly 20% this year. At the same time, the uncertainty of the continuous expansion of Apple's supply chain will also make the new iPhone more expensive than ever.
Will the LED Transparent Screen Become the "Darling" of the LED Screen Market?
Will the LED Transparent Screen Become the "Darling" of the LED Screen Market?
Led transparent screen is a popular subdivision screen in the LED field in recent years. We often see the figure of transparent screen whether it is glass curtain wall, glass window or shopping mall. As one of the high-end display products, the price problem of the original transparent screen makes many users can only wait and see from a distance.With the rapid development of LED transparent screen, its price has also decreased significantly to the acceptable range of users, and various spacing and special-shaped display products emerge in endlessly. Transparent screen is gradually emerging in the commercial display field because of its unique display style.Scientific and technological progress is the core of social development and the first productive force. The transformation and upgrading of science and technology has not only greatly improved the modern production capacity of enterprises, but also further improved human vision in social development. In recent years, with the strong support of the construction industry in big cities, the glass curtain wall area of commercial buildings has reached a new peak. In this regard, led transparent screen manufacturers have made major technical upgrades to the transparent display scheme of large-area glass curtain wall. As a provider of indoor and outdoor LED transparent screen display solutions, Shenzhen Jingang vision is also very happy to contribute to our strength under the mainstream of this market.In order to expand the scope of commercial display application market, we continue to integrate new technologies and introduce various innovative technologies, products and solutions to the market. In recent years, led transparent screen solutions have become more and more common due to the use of advanced technology and most market participation. This popularity has also prompted the outbreak of urban commercial exhibition industry, which is also due to the continuous improvement of infrastructure and the rapid development of urbanization.When it comes to transparent screens, the market segments are relatively broad, and the market segments are based on products, technologies, components, display types, screen sizes and application scenarios. Based on various segments, led transparent screen can help enterprises choose growth niche market and strategic development, and determine the difference between the core application field of products and the target market.
Principle of Touch Screen Capacitance Screen_ What Are the Application Areas of Touch Screen
Principle of Touch Screen Capacitance Screen_ What Are the Application Areas of Touch Screen
Principle of touch screen capacitance screenTouch screen is the most friendly computer user interface at present. It is more and more widely used in consumer electronic products. The touch detection device and touch screen controller are good. The mechanical structure of the detection device is damaged in the process of touch screen device, resulting in the decline of touch accuracy and shortening of service life of the touch screen; In addition, the cost of commonly used capacitive and resistive touch screens is relatively high. Inductive touch screen can avoid damage to the touch detection device during touch and prolong the service life of touch screen products.The working principle of inductive touch screen, as well as the hardware circuit and part of the program design of the controller, all of the controller circuits have passed the verification. The non-direct contact touch of inductive touch screen avoids the damage to the touch detection device, prolongs the service life of the whole touch control system, and the repetition accuracy and touch sensitivity of touching the same point meet the expected design requirements.The capacitive induction touch screen is not affected by dirt, dust or oil stains, but it will be affected by the surrounding environment to varying degrees because it adopts the electric field coupling principle. At the same time, the capacitive induction touch screen has serious drift phenomenon.What are the application areas of touch screenTouch screen has become a hot spot in the market. In the market, shopping malls, hotels and other places, it is still dominated by smart phones. The growth momentum of the smartphone market slowed down and entered a period of gentle development. The application of touch screen outside mobile phones, such as automotive electronics, retail, medical treatment, public facilities, leisure and entertainment, will promote the rapid growth of touch screen market.In addition to applications in large fields, touch screens can be innovated in subdivided fields. In the field of "pen writing tablet", innovate the touch screen.The products have been extended to tablet with pen writing function, e-book, all-in-one machine, electronic whiteboard, etc. And the pen and hand can be used at the same time. When one hand is touching, the pen can also write at the same time.The traditional materials of capacitive touch screen are ITO, metal grid, nano silver wire, carbon nanotube, conductive polymer, graphene, etc. Metal mesh and nano silver wire have more supporters. Both materials are metals (silver or copper) and their conductivity is better than ITO.
Screen Fingerprint Identification Will Be Applied to LCD Screen, Not Just OLED
Screen Fingerprint Identification Will Be Applied to LCD Screen, Not Just OLED
Fingerprint identification on mobile phones has become an indispensable part of our daily life.Especially during the epidemic, face recognition is mostly paralyzed due to wearing masks, and fingerprint recognition has become the current optimal solution.Of course, if you use a mobile phone without fingerprint identification such as iPhone x, you have to find a way by yourself...9to5mac also gave a tutorial on using face ID with mask Screen fingerprint identification has become the mainstream, and the technology is also very mature. However, Tony saw a news while surfing the Internet today, saying that Tianma, a domestic LCD and LCM design company, released the world's first LCD Screen multi-point fingerprint identification scheme (Ted finger print, referred to as TFP).The strength of this scheme is that the off-screen fingerprint identification mobile phones we use now are basically OLED screens, while the LCD screen basically has no on-screen fingerprint identification...One plus 6T off screen fingerprint identification diagram As we all know, the screen fingerprint unlocking needs the illumination of the light source for fingerprint identification, but the LCD screen is relatively thick and has poor light transmittance; Second, the LCD screen cannot emit light actively, and the backlight will interfere with the identification of optical fingerprints.These two problems make LCD basically bid farewell to screen fingerprint recognition.Therefore, mobile phones with LCD screen basically adopt side fingerprint identification, and some will be combined with the power key on the side of the mobile phone.But OLED screens don't have this problem. Because the OLED screen is thinner and more transparent, it is natural to use fingerprints on and off the screen.Then why do manufacturers not directly on the OLED screen, but want to operate madly on the LCD screen? The reason is also simple: because LCD screens are cheap...Although Lu Weibing, vice president of Xiaomi, said publicly some time ago that Hongmi team successfully realized fingerprint recognition under LCD screen on a redmi note8 Pro prototype, so far, no manufacturer has officially announced this technology, and Tianma is the first.So Tony was surprised when he saw the news. This fingerprint identification scheme can be said to break through the two obstacles of current screen fingerprint identification technology - LCD screen and identification area.If the multi-point fingerprint identification in the LCD screen can achieve mass production, it means that the LCD screen can also use fingerprint identification!Another point is the identification area of off screen fingerprint identification.So far, the identification area of screen fingerprint is basically fixed, almost at the lower position of the screen. People will subconsciously put their fingers at the approximate position when unlocking the mobile phone.But no one can guarantee that every time the finger is pressed on the fingerprint identification area, because the identification area is so large, it is inevitable that there will be deviations...The scheme announced by Tianma this time shows that it can make the whole LCD screen recognize fingerprints!Tony mended his brain, that is, I could identify my fingerprint and unlock my mobile phone by placing my finger anywhere on the screen.If this multi-point fingerprint identification technology can be mass produced, the efficiency of unlocking the screen will be much higher than now. You don't have to find the identification area and poke it on the screen.It can be said that Tianma's operation directly makes the LCD screen ride on the head of OLED in fingerprint identification. Once LCD screens didn't even deserve fingerprint identification. Now they not only have fingerprint identification, but also can identify 5 at a time...As for why the multi-point fingerprint solution in the TFP screen can realize these functions, high integration has played a great role.In short, it is to integrate touch, display and fingerprint identification, replace the previous design of fingerprint identification module and reduce the thickness of the screen, so that the screen fingerprint identification is feasible on the LCD.At the same time, in order to realize fingerprint identification on the whole screen, Tianma covers the whole screen with optical path and sensors.As for how multi-point fingerprint identification can play new tricks, it still depends on the subsequent operations of manufacturers.As for the fate of LCD screen after mass production of this scheme, we still don't know, but before that, LCD screen has never been the first choice for mobile phone manufacturers.After all, the OLED screen is thinner, can emit light by itself, and the color is more bright. The experience is much better than that of LCD, but there are still problems of stroboscopic and screen burning, and the price is relatively expensive.When it comes to the LCD screen, the advantages and disadvantages are reversed...The price of iPhone 11 with LCD screen is lower For most users, OLED screen is their first choice, but there are still a group of people who stick to the LCD screen.In addition to having a larger viewing angle, another reason is that they can't accept the strobe of OLED screen. They think it's really hurting their eyes...Therefore, once the TFP scheme realizes mass production, the defect that the LCD screen has no off screen fingerprint will be supplemented. If the price can be controlled well, the current situation of the screen in the mobile phone market may change a lot.However, before that, the mobile phone market was still dominated by OLED screens, not to mention the suoper AMOLED screens such as Samsung. The evolution of LCD screens has a long way to go...
What Is the Forming Process of the Flexible Touch Screen
What Is the Forming Process of the Flexible Touch Screen
Every day, new products, new experiences, professional membrane switch proofing factory, 12 hour expedited shipmentFrom the hot sale of Samsung Galaxy edge flagship series to the advent of various curved screen bracelets and TVs, it is no doubt that curved screen technology has been widely valued and sought after by consumers and the industry in recent years. In the future, with the continuous development and improvement of fixed curvature curved screen technology, the flexible touch screen that can realize arbitrary bending will become an important development direction. On the 25th, at the 13th International touch screen and display technology development forum, Huang Hanfeng, deputy general manager of touch R & D from Shenzhen oufeiguang technology, made a detailed introduction to the flexible touch screen from the perspective of the industrial chain around the four key elements of flexible display, flexible touch, flexible cover plate and flexible bonding.Flexible displayFor LCD display, the LCD is usually responsible for controlling the display gray, and the three primary color filters control the display color, which requires them to correspond one by one in order to display normally. If the LCD panel is used as a curved screen, the inner circumference and outer circumference will be unequal, and the original one-to-one corresponding liquid crystal and filter will have position deviation, resulting in color distortion. In addition, it will also cause a series of problems such as light leakage, dark screen ghosting and snowflakes. Therefore, for the curved screen, the OLED display technology represented by AMOLED has almost become the best choice.Huang Hanfeng said that at present, Samsung is the undisputed AMOLED overlord in terms of patents, production capacity, yield and technology accumulation. The sixth generation AMOLED panel that began mass production at the end of 2014 can now produce 50000 pieces per month, and the yield can reach 60%. The second is LG, but LG has been focusing on large screen curved TV. At present, with the increasingly obvious trend of small screen AMOLED, LG is also accelerating the layout of small and medium-sized flexible screens. Although its output is not as good as Samsung, its yield can also reach more than 60%. The latest release of Xiaomi note 2 is the flexible AMOLED screen of LG. In addition, Japan's JDI and sharp (already acquired by Hon Hai), Taiwan's Youda, domestic BOE, Rouyu and other screen manufacturers are also accelerating the layout of their own AMOLED technology. It is the general trend for AMOLED to replace LCD.Flexible touchIn terms of touch control, ITO (indium tin oxide) transparent conductive film has become the most important touch layer material for display screens such as non curved LCD and OLED due to its good light transmittance, low thickness, excellent hardness and conductivity, mature manufacturing technology and many other reasons. However, as ITO itself is a brittle material, it is not suitable for making flexible touch layer with large curvature or even free bending, and the cost and cost are high. At present, it accounts for about 30% - 40% of the cost of the upstream material part of the whole touch screen industry, and uses "indium", a rare metal with limited reserves. Therefore, with the advent of the era of curved surface and flexibility, it is likely to be replaced.Huang Hanfeng said that at present, the main substitutes for ITO materials are graphene, carbon nanotubes, nano silver and metal mesh. Graphene and carbon nanotubes are very good substitutes for ITO from the characteristics of the material itself. However, graphene is still in the R & D stage and is still far from mass production. The industrial mass production technology of carbon nanotubes is not perfect, and the conductivity of the thin film products made of carbon nanotubes is not as good as that of ITO. Therefore, from the perspective of technology and marketization, metal grid and nano silver technology will be the protagonists of development in recent years. Metal grid technology has been applied in some PC display markets.The metal mesh technology mentioned here is a conductive metal mesh formed by pressing metal conductive materials such as silver and copper or oxides on thin film substrates such as pet. Its main advantages are low raw material cost and good folding ability. However, due to the problem of Murray interference ripple caused by yield, output and high-speed pixels, it is more suitable for desktop all-in-one computers, laptops, televisions and other products with low resolution and relatively long-distance use.Nano silver technology means that nano silver ink material is coated on pet or glass substrate, and then laser lithography is used to depict the conductive network of nano silver wire. Its main advantages are high yield, small linewidth, good conductivity and folding resistance, while its disadvantage is high cost. Moreover, compared with metal mesh, nano silver material has smaller radius of curvature and small resistance change rate during bending. In addition, due to the linewidth, it is more suitable for high-resolution close-up scenes such as mobile phones, smart watches and bracelets."By comprehensive comparison, I think nano silver has more advantages than metal grid," Huang Hanfeng said.Flexible cover plateIn order to bend at will, in addition to solving the flexible display and flexible touch, the last and most critical part is the flexible cover plate and fitting. However, due to its hard and fragile characteristics, glass cover plate is not the best choice in the field of flexibility.Huang Hanfeng said that compared with glass, pet and PI (polyimide) materials with high surface hardness and high light transmittance are better choices, but at present, pet and PI also have many problems to be solved. On the one hand, they are not resistant to high temperature and have poor light transmittance: pet and PI will have problems of partial melting and property change at low ambient temperature, and they do not have good light transmittance compared with glass.Secondly, it is difficult to achieve a balance between hardness and folding. Huang Hanfeng said that if you want to meet certain hardness requirements, you need to apply the hardened layer, but after coating the hardened layer, after repeated folding, the hardened layer will crack and reduce the hardness. Therefore, how to achieve a balance between the two is the biggest bottleneck at present.Compared with the flexible cover plate that needs further development, the 3D glass cover plate necessary for curved screen mobile phones has a very good development prospect.Some data show that 2018 may become the outbreak year of 3D glass cover plate. It is expected that the mobile phone penetration rate of 3D glass cover plate will exceed 50% around 2020. With the maturity of 3D glass cover plate processing technology and the decline of unit price, the penetration rate will further improve. It is expected that the market scale will reach more than 19 billion yuan in the future.Huang Hanfeng said that the common manufacturing processes of 1R (single radian bending) and 2R cover plates have been basically mature. They basically achieve a bending effect through hot bending technology. The difference is mainly in the coating of hardened layer. However, the yield is a problem to be solved, which is basically only 30%, and there is no room for improvement in the short term. If 4R is to be done in the future, the yield will be lower. Therefore, the yield is one of the urgent problems to be solved in the manufacture of cover plate.Flexible fitAfter the cover plate is selected, it is fitted, that is, the cover plate is integrated with the touch and display layer below to form a complete module. At present, there are two common methods: roller type and vacuum bonding: the roller has high efficiency and needs to roll the pasted object. The efficiency of vacuum type is relatively low, but it does not need to roll the pasted object, so it is not easy to cause damage to the pasted object.Huang Hanfeng said that if the curvature is large, you can only choose vacuum bonding. However, for the possible 4R bonding in the future, due to its special four side bending characteristics, it is difficult to avoid wrinkles and bubbles in places with large radians even if vacuum is adopted. At present, the industry has adopted customized extrusion airbags or plastic parts to alleviate this problem, but it has not been completely solved. Therefore, the industry still has room for progress in multi surface and high curvature fitting technology.Huang Hanfeng frankly said: "from the perspective of the supply chain, it will take several years for the flexible screen to really realize mass production. At the same time, it also needs the upstream and downstream supply chain to jointly overcome the relevant technical problems."For future development, Huang Hanfeng believes that the integration of various modules will be the general trend. He said: "in the incell technology of LCD, ITO touch film and liquid crystal layer are integrated together, which is a good integration. Now the fingerprint module is independent, and it can also be integrated with touch in the future. I think there will be a process of mutual integration and progress between functional modules in the future."
LG Will Mass Produce Mobile Phone Flexible Screen in the Fourth Quarter of This Year
LG Will Mass Produce Mobile Phone Flexible Screen in the Fourth Quarter of This Year
In the early morning of June 20, Beijing time, according to the English version of Korea Times, in order to further consolidate the leading position in the display market, LG display will mass produce a new version of smart phone flexible display from the fourth quarter of this year and provide it to major customers.LG researchers and technicians said that the purpose of producing flexible display is to meet the growing demand for innovative solutions. LG spokesman Frank Lee said that the company is trying to meet the rapidly growing market demand for advanced display screens. Li stressed that the new display screen is foldable and not fragile, and the product will become a major innovation in display technology.LG said it was investing in flexible displays and high-tech panels for mobile devices, namely OLED panels that support high-definition (UHD) displays. "We have completed the development of the first flexible display and will start mass production from the fourth quarter of this year," LG said in a statementThe E2 production line of the company's display manufacturing plant in pozhou, South Korea, will be responsible for the production of flexible displays. LG said that the OLED flexible display will adopt generation 4.5 glass cutting technology, and the monthly production capacity of the production line will be 12000 pieces.Koo Bon moo, chairman of LG, hopes to revive the enterprise by launching leading products. Therefore, as a major shareholder of LG display, LG Electronics will launch its first flexible display smartphone later this year.LG predicts that flexible display may become the latest trend of mobile devices, so the company may also consider selling flexible display to major mobile device manufacturers.It is reported that LG's competitor Samsung display is also developing similar products. At present, Samsung display is testing flexible display products to improve product quality. Due to manufacturing problems, the upcoming Galaxy note 3 should not be configured with a flexible display in the short term.Kim Yang Jae, an analyst at Woori investment securities in Korea, said: "mobile phone manufacturers are increasingly pursuing bright images, beautiful shapes and thin display screens, while flexible display screens cater to this demand. LG display is expected to help them seize a greater market share." Kim said, LG's strategy of switching to flexible display can break the industry conventions and make the company earn stable profits.In fact, flexible display is only one step for LG to invest in different businesses and seize the global market share of OLED. LG is the first company to mass produce OLED display and foldable OLED screen. At the same time, the company is also developing OLED TV supporting high-definition display.Lim Joo soo, manager of LG OLED Technology Strategy Department, said that in order to remain competitive, LG is trying to reduce costs, reduce power consumption and improve image quality. Lim Joo soo said: "LG display is developing high-definition OLED TV screen. In order to improve the brightness, we plan to replace the existing 'bottom light' technology with 'top light' technology. In addition, the company will develop new materials to extend the service life of OLED display."In the next few years, OLED TV will become the mainstream product, and LG plans to expand the TV screen size to 65 inches and 77 inches. LG said that the above two OLED displays will be produced by the M1 test production line of pozhou factory.
LG Will Mass Produce Mobile Phone Flexible Screen in the Fourth Quarter of This Year
LG Will Mass Produce Mobile Phone Flexible Screen in the Fourth Quarter of This Year
In the early morning of June 20, Beijing time, according to the English version of Korea Times, in order to further consolidate the leading position in the display market, LG display will mass produce a new version of smart phone flexible display from the fourth quarter of this year and provide it to major customers.LG researchers and technicians said that the purpose of producing flexible display is to meet the growing demand for innovative solutions. LG spokesman Frank Lee said that the company is trying to meet the rapidly growing market demand for advanced display screens. Li stressed that the new display screen is foldable and not fragile, and the product will become a major innovation in display technology.LG said it was investing in flexible displays and high-tech panels for mobile devices, namely OLED panels that support high-definition (UHD) displays. "We have completed the development of the first flexible display and will start mass production from the fourth quarter of this year," LG said in a statementThe E2 production line of the company's display manufacturing plant in pozhou, South Korea, will be responsible for the production of flexible displays. LG said that the OLED flexible display will adopt generation 4.5 glass cutting technology, and the monthly production capacity of the production line will be 12000 pieces.Koo Bon moo, chairman of LG, hopes to revive the enterprise by launching leading products. Therefore, as a major shareholder of LG display, LG Electronics will launch its first flexible display smartphone later this year.LG predicts that flexible display may become the latest trend of mobile devices, so the company may also consider selling flexible display to major mobile device manufacturers.It is reported that LG's competitor Samsung display is also developing similar products. At present, Samsung display is testing flexible display products to improve product quality. Due to manufacturing problems, the upcoming Galaxy note 3 should not be configured with a flexible display in the short term.Kim Yang Jae, an analyst at Woori investment securities in Korea, said: "mobile phone manufacturers are increasingly pursuing bright images, beautiful shapes and thin display screens, while flexible display screens cater to this demand. LG display is expected to help them seize a greater market share." Kim said, LG's strategy of switching to flexible display can break the industry conventions and make the company earn stable profits.In fact, flexible display is only one step for LG to invest in different businesses and seize the global market share of OLED. LG is the first company to mass produce OLED display and foldable OLED screen. At the same time, the company is also developing OLED TV supporting high-definition display.Lim Joo soo, manager of LG OLED Technology Strategy Department, said that in order to remain competitive, LG is trying to reduce costs, reduce power consumption and improve image quality. Lim Joo soo said: "LG display is developing high-definition OLED TV screen. In order to improve the brightness, we plan to replace the existing 'bottom light' technology with 'top light' technology. In addition, the company will develop new materials to extend the service life of OLED display."In the next few years, OLED TV will become the mainstream product, and LG plans to expand the TV screen size to 65 inches and 77 inches. LG said that the above two OLED displays will be produced by the M1 test production line of pozhou factory.
Design Scheme for Replacing Ch7511b DP to LVDS Screen
Design Scheme for Replacing Ch7511b DP to LVDS Screen
Ch7511b is a DP to LVDS screen conversion chip, and ch7511b is an EDP to LVDS conversion chip. Ch7511b converts the embedded DisplayPort signal to LVDS (low voltage differential signal). Through the advanced decoding / encoding algorithm of ch7511b, the input EDP high-speed serial video data can be seamlessly converted to LVDS. LVDS is a popular high-speed serial link display technology for medium / large LCD displays. This chip is specially designed for the multi-functional integrated PC, industrial computer and notebook computer market.Cs5211 is also an EDP to LVDS bridge chip. Cs5211 has flexible configuration and is suitable for various low-cost display systems. Cs5211 is compatible with EDP 1.2 and supports 1-channel and 2-channel modes. The speed of each channel is 1.62gbps and 2.7Gbps. Cs5211 adopts powerful SerDes technology, which can recover high-speed serial data with low bit error rate.Compared with ch7511b, the advantages of cs5211 in designing DP to LVDS conversion circuit are as follows:1. Cs5211 LVDS transmitter supports single port and dual port modes.2. The maximum resolution supported by cs5211 is WUXGA (1920x1200).3. Cs5211 has 4 configuration pins and can support 16 different combinations of panel resolution and LVDS working mode of an EEPROM image.4. Cs5211 provides a simple tool to edit, generate and update EEPROM images for custom configuration.5. Cs5211 has simple circuit design and fewer peripheral circuit components, which is more conducive to saving BOM cost.6. Cs5211 integrates clock source to save external crystal.7. Cs5211 supports a wide range of power supply (core power is 1.8 1.2V), saving on-board power supply equipment.8. The total power of cs5211 is less than 300MW, which simplifies the design of power supply network and is easy to integrate into the current popular low-cost display system.9. Cs5211 is qfn68 package and ch7511b is also qfn68 package. Overview: cs5211 directly replaces ch7511 without changing the circuit board and pin to pin replacement mode.10. Cs5211 on-chip MCU can update the program online to adjust various parameter values of the screen, improve the universality of the design and product compatibility, and increase the applicable scenario and development and design cost of the scheme product.The detailed characteristic parameters of cs5211 are as follows:Overview:1. 2-channel DisplayPort v1.1 compatible receiverSupports 18 bit single port, 18 Bit dual port, 24 bit single port and 24 bit dual port LVDS outputsSupports open LDI and spwg bit mapping for LVDS applicationsBuilt in oscillator without external crystalEmbedded linear voltage drop regulator (LDO)On Chip MCUSelect the support panel through the GPIO pin controlBoot ROM is loaded automatically after power onUpdate serial boot ROM data via I2C bus or auxiliary channelAutomatic chip power mode control.EMI reduction of EDP and LVDWide core power range from 1.8V to 1.2V.Qfn68 package.HBM 4KV2. DP receiver characteristics:Compliant with embedded display port (EDP) specification 1.2Support 1 or 2 main link channels at 1.62gbps or 2.7Gbps link rate.Enter the color depth of 18 / 24 bits per pixel in RGB format.Support enhanced frame mode.Support VESA and CEA timing standards, resolution up to 1920x1200, 60Hz, 24 bit per pixel mode.Supports dynamic refresh rate switching.Support quick link training and full link training.Support EDP authentication: optional scrambling seed reset and optional frame.Support HPD interrupt.3. LVDS output characteristics:Support 18 bit single port, 18 Bit dual port, 24 bit single port and 24 bit dual port LVDS output interfaces.Supports openldi and spwg bit mapping for LVDS applications.Keep LVDS output when the input video is not ready.Flexible LVDS output pin switching.Programmable swing / common mode voltage.Output conversion rate control to reduce electromagnetic interference.4. Output panel and backlight controlProgrammable LCD panel power sequenceTwo PWM modes are supported, including backlight brightness level control PWM pin and BLUP / bldn pinEditing: JQ
Design Scheme for Replacing Ch7511b DP to LVDS Screen
Design Scheme for Replacing Ch7511b DP to LVDS Screen
Ch7511b is a DP to LVDS screen conversion chip, and ch7511b is an EDP to LVDS conversion chip. Ch7511b converts the embedded DisplayPort signal to LVDS (low voltage differential signal). Through the advanced decoding / encoding algorithm of ch7511b, the input EDP high-speed serial video data can be seamlessly converted to LVDS. LVDS is a popular high-speed serial link display technology for medium / large LCD displays. This chip is specially designed for the multi-functional integrated PC, industrial computer and notebook computer market.Cs5211 is also an EDP to LVDS bridge chip. Cs5211 has flexible configuration and is suitable for various low-cost display systems. Cs5211 is compatible with EDP 1.2 and supports 1-channel and 2-channel modes. The speed of each channel is 1.62gbps and 2.7Gbps. Cs5211 adopts powerful SerDes technology, which can recover high-speed serial data with low bit error rate.Compared with ch7511b, the advantages of cs5211 in designing DP to LVDS conversion circuit are as follows:1. Cs5211 LVDS transmitter supports single port and dual port modes.2. The maximum resolution supported by cs5211 is WUXGA (1920x1200).3. Cs5211 has 4 configuration pins and can support 16 different combinations of panel resolution and LVDS working mode of an EEPROM image.4. Cs5211 provides a simple tool to edit, generate and update EEPROM images for custom configuration.5. Cs5211 has simple circuit design and fewer peripheral circuit components, which is more conducive to saving BOM cost.6. Cs5211 integrates clock source to save external crystal.7. Cs5211 supports a wide range of power supply (core power is 1.8 1.2V), saving on-board power supply equipment.8. The total power of cs5211 is less than 300MW, which simplifies the design of power supply network and is easy to integrate into the current popular low-cost display system.9. Cs5211 is qfn68 package and ch7511b is also qfn68 package. Overview: cs5211 directly replaces ch7511 without changing the circuit board and pin to pin replacement mode.10. Cs5211 on-chip MCU can update the program online to adjust various parameter values of the screen, improve the universality of the design and product compatibility, and increase the applicable scenario and development and design cost of the scheme product.The detailed characteristic parameters of cs5211 are as follows:Overview:1. 2-channel DisplayPort v1.1 compatible receiverSupports 18 bit single port, 18 Bit dual port, 24 bit single port and 24 bit dual port LVDS outputsSupports open LDI and spwg bit mapping for LVDS applicationsBuilt in oscillator without external crystalEmbedded linear voltage drop regulator (LDO)On Chip MCUSelect the support panel through the GPIO pin controlBoot ROM is loaded automatically after power onUpdate serial boot ROM data via I2C bus or auxiliary channelAutomatic chip power mode control.EMI reduction of EDP and LVDWide core power range from 1.8V to 1.2V.Qfn68 package.HBM 4KV2. DP receiver characteristics:Compliant with embedded display port (EDP) specification 1.2Support 1 or 2 main link channels at 1.62gbps or 2.7Gbps link rate.Enter the color depth of 18 / 24 bits per pixel in RGB format.Support enhanced frame mode.Support VESA and CEA timing standards, resolution up to 1920x1200, 60Hz, 24 bit per pixel mode.Supports dynamic refresh rate switching.Support quick link training and full link training.Support EDP authentication: optional scrambling seed reset and optional frame.Support HPD interrupt.3. LVDS output characteristics:Support 18 bit single port, 18 Bit dual port, 24 bit single port and 24 bit dual port LVDS output interfaces.Supports openldi and spwg bit mapping for LVDS applications.Keep LVDS output when the input video is not ready.Flexible LVDS output pin switching.Programmable swing / common mode voltage.Output conversion rate control to reduce electromagnetic interference.4. Output panel and backlight controlProgrammable LCD panel power sequenceTwo PWM modes are supported, including backlight brightness level control PWM pin and BLUP / bldn pinEditing: JQ
no data
Contact US

    Bao’an District  Shenzhen City, China

     +86 189 3806 5764

NEWSLETTER SUBSCRIBE
no data
Copyright © 2021 CrowBerry

   

chat online
NEED HELP? WE'RE HERE!