What changes will the iPhone7 use FPGA?

Lei Feng network : According to the author of the Semiconductor Industry Alliance, the article mainly introduces the FPGA chip used in the iPhone7, and introduces the development history of the FGPA chip.

These days, the Apple iPhone 7 has been officially launched. The iFixit team in San Luis Obispo, California, disassembled the iPhone 7 Plus. The dismantling of pictures and commentary reports quickly occupied the world’s technology media. In this dismantling report, we were surprised to find a new member.

The motherboard of iPhone 7 Plus integrates a very small FPGA with a size of 2mm x 2mm. This chip is just a few years ago Lattice's iCE40 low-density, low-power, small-package FPGA, model iCE5LP4K, with a density of 3.5K LUT-4 equivalent logic resources. Its peripheral devices include audio CODECs, audio amplifiers, diversity receivers plus atmospheric pressure sensors on the back, NFC controllers, WIFI/Bluetooth and other controllers.

After some speculation by some technical experts, what the FPGA implements here may be the function of the Sensor HUB, which is to preprocess the various sensors (possibly including bus control, bus conversion, time division multiplexing, interface conversion, etc.). The functional module), when necessary, performs data exchange with the host processor CPU and further calculations to reduce the monitoring and operating frequency of the host processor CPU related modules, thereby reducing power consumption and improving processing performance.

Of course, there are also some other technical people who have different opinions. They think that the main role of this FPGA is to control the USB Type C. At present, due to standards and other reasons, there is no ASSP vendor to provide a complete solution, and Lattice uses FPGA. Features, first to become one of USB Type C suppliers. This is also the advantage of FPGAs over ASSPs, which accelerates the time-to-market and helps customers save time and money. Of course, these are just speculations before Apple or Lattice unveiled its design intent.

The iPhone7 Plus dismantling component catalog is as follows :

Apple/Cirrus Logic 338S00105 Audio Codec

Cirrus Logic 338S00220 Audio Amplifier(x2)

Lattice Semiconductor ICE5LP4K FPGA

Skyworks 13702-20 Diversity Receive Module

Skyworks 13703-21 Diversity Receive Module

Avago LFI630 183439

NXP 610A38

Apple A10 Fusion APL1W24 SoC

Samsung 3 GB LPDDR4 RAM

Qualcomm MDM9645M LTE Cat. 12 Modem

Skyworks 78100-20

Avago AFEM-8065 Power Amplifier Module

Avago AFEM-8055 Power Amplifier Module

Universal Scientific Industrial O1 X4

Bosch Sensortec BMP280 Barometric Pressure Sensor

Toshiba THGBX6T0T8LLFXF 128 GB NAND Flash

Murata 339S00199 Wi-Fi/Bluetooth Module

NXP 67V04 NFC Controller

Dialog 338S00225 Power Management IC

Qualcomm PMD9645 Power Management IC

Qualcomm WTR4905 Multimode LTE Transceiver

Qualcomm WTR3925 RF Transceiver

TDK EPCOS D5315

Texas Instruments 64W0Y5P

Texas Instruments 65730A0P Power Management IC

Lattice's FPGA devices have never been to Apple's supply chain before. This is the first time ever. It is also the first time that Apple has introduced an FPGA device into the iPhone. Why? According to common sense, if Apple wants to reduce the cost of 100 million iPhones, it should use ASIC chips as much as possible. The traditional tradition is that FPGAs are usually used only in low-volume markets, so this time Apple's design is real. Has caused everyone's interest.

Many years ago, the Samsung Galaxy S4 and S5 were the first to use Lattice's FPGAs of the same model. At the time, the main application direction was for infrared recognition and two-dimensional code recognition. However, after the S5 Galaxy model, Samsung used other things. The program replaced the FPGA, so it did not cause the industry to shake, but the significance of Apple's use of FPGA is still very different.

Apple's iPhone 7 adopts the FPGA scheme to set a benchmark in the smart phone industry. This also means that FPGA is the first time this popular device has appeared in the consumer electronics market so that it has won wide attention, and this situation was many years ago. It is totally impossible. At least we can get a concise conclusion: It may be that the future integration of FPGA will become a watershed to differentiate high-end smart phones , perhaps in the future we will find more consumer electronics will be applied to FPGA. This should be a good news for the FPGA industry.

The Lattice chip used by Apple is the ICE series. In 2012, Lattice released the first FPGA iCE40 for mobile applications on the market. As for the iCE product's naming, it is consistent with the product's features, which is low power consumption like ice. Prior to Lattice’s introduction of iCE, the consumer market accounted for less than 5% of its total revenue, and now consumption is already approaching 30% of sales, all thanks to new market applications.

With the rapid evolution of handheld market technology, many functions are still in the trial period, there is no corresponding agreement and a stable market, ASSP vendors are not willing to develop, so in the six months to one year window, FPGA can help customers solve The issue of the transition period. Lattice uses the "allah oil" as a metaphor for the iCE series, which means the most flexible wearable device. In addition, FPGA can help customers save debugging time, because once ASSP has a bug, customers can only wait for the next batch of products, FPGA can be modified at any time, shortening the development cycle.

In 2015, Lattice introduced the iCE40 UltraLite for mid-range phones and wearables. As its name implies, it is a simplified version of the iCE40 Ultra. However, the configuration has not been reduced. The biggest change is that the storage space is reduced, and the logic unit is reduced from 1k/2k/3k to 0.64k and 1.2k, customers can choose for their own applications. For example, a bracelet or a mid-range mobile phone, UltraLite can be considered at the same time to achieve low power consumption, differentiation and low price product design. The package size is as small as 1.4x1.4x0.45mm, which is 60% smaller than the competitor's. The power consumption is only 35uA, which is 30% lower than that of the competitors. It also has the highest hardware integration.

iCE series applications include: wearable devices, pedometer, always online navigation, voice input, LED breathing effects, schedule reminders, activity monitoring, infrared control, user identification and so on. At the same time, UltraLite also supports a variety of industrial applications, including hand-held tools, POS and other traditional industrial applications, but also includes portable medical devices such as blood pressure monitors, dental, gastroscopy and other in vivo medical applications can also be used.

FPGA is known as “universal chip”. From a technical point of view, it is a programmable product that can be used repeatedly without changing the hardware composition of the chip itself. According to market changes, different functions are implemented on the same carrier. It sounds good, but it is very difficult to develop. Manufacturers that can make FPGA devices must be big cattle, because semiconductor giants like Intel and Texas Instruments have tried to enter this market, but Because of this kind of reason, it's gone. Currently, there are only a handful of companies worldwide that can mass-produce FPGAs. Xilinx, Altera, Lattice, and Microsemi are familiar American companies.

FPGA is a kind of semi-custom circuit in the field of Application Specific Integrated Circuit (ASIC). Its advantages include that it can be finished quickly, can be modified, errors in the program can be corrected, and the cost is cheaper. Engineers reconfigure the logic blocks and I/O modules inside the FPGA to implement the desired logic. FPGAs also have static repeatable programming and dynamic reconfiguration in the system so that the hardware functions can be modified programmatically just like software. FPGA can accomplish the function of any digital device, even high-performance CPU can be realized with FPGA. FPGA is like a white paper: When power is on, FPGA chip reads the EPROM data into the on-chip programming RAM and enters the working state after the configuration is completed. After power-off, the FPGA recovers into a white chip and the internal logic relationship disappears, so the FPGA can Repeated use. Using FPGAs to develop digital circuits can greatly reduce design time, reduce the PCB area of ​​printed circuit boards, and increase system reliability. Built-in hard CPU or soft core CPU in FPGA, can realize digital logic and adapt to embedded development.

In 2014, the total size of the global FPGA market reached 5 billion US dollars, of which, China’s market share was nearly 1.7 billion US dollars, and the Chinese market accounted for one-third of the global market. Analysts expect that the annual compound annual growth rate of the FPGA market from 2015 to 2020 will be 9%. By 2020, the global FPGA market will reach US$8.4 billion. At present, FPGA is in an accelerating market trend, and the growth rate is far greater than that of other chip markets. At the same time, the average gross profit of the FPGA industry is considerable. According to market data analysis, the average gross profit of the industry is greater than 60%. The FPGA industry also needs a larger market size to attract more users. It is expected that with the gradual increase of FPGA output and further cost reduction, its market share will continue to increase.

Currently, the main application areas of FPGAs are communication, industrial control, defense, and consumption. One of the most interesting trends in FPGAs in recent years is the continuous expansion of the application field.

In the traditional application market of FPGA, communication gradually realizes high speed and complex protocols, while consumer electronics applications focus on low power consumption and low cost. In addition, FPGAs are also widely used in medical electronics, security, video, industrial automation, voice networks, semiconductor manufacturing equipment, and home appliances. The application of FPGA in industry mainly involves four major fields, namely motor control, industrial network, machine vision and robot control. Taking motor control as an example, the emphasis is on reducing noise, reducing vibration, and reducing EMI while obtaining higher control accuracy and reducing energy consumption.

Currently in the global market, Xilinx, Altera's two major companies still occupy an absolute monopoly on the technology and market of FPGA. The two companies have a market share of nearly 90%, with more than 6,000 patents; the remaining market share is mainly occupied by Lattice and Microsemi, and the patents of these two companies also reach more than 3,000. In 2014, Xilinx and Altera had revenues of US$2.38 billion and US$1.93 billion, respectively; and in 2014, both Lattice and Microsemi (FPGA only) had revenues of US$366 million and US$275 million. The top four FPGA market leaders have almost monopolized the FPGA market share.

The market prospect of FPGA is attractive, but the high threshold is unsurpassed in the chip industry. More than 60 companies worldwide have spent billions of dollars successively and have gone to the summit of FPGAs. There are many industry giants such as Intel, IBM, Texas Instruments, Motorola, Philips, Toshiba, and Samsung. However, only the top ones succeeded. Four companies in the US Silicon Valley - Xilinx, Altera, Lattice, Microsemi. The above four companies hold more than 9,000 patents. The technical barriers constituted by so many technical patents are of course unattainable.

Last year, Intel took Altera, the second-largest FPGA company, for $16.7 billion. Intel CEO Brian Krzanich once stated that Intel’s growth strategy is to expand its core assets to enter a highly profitable and complementary market space. Through this acquisition, Intel can push Moore’s Law to a better and stronger next-generation solution. Enterprise users expect to get more cost-effective performance in the field of enhanced networks, large cloud data centers, and the Internet of Things, and this is exactly what Moore's Law and Intel and Altera can jointly achieve.

So why did Intel buy this company at such a high price (almost 8 times the sales rate)? The author noted that in 2014, Microsoft and Amazon, together with several top universities, published an article entitled "A Reconfigurable Fabric for Accelerating Large-Scale Datacenter Services." The long paper, the paper's participants also include technical support from the two companies Altera and Quanta, the project involved in the article was named "Catapult", which is a radical project that Microsoft Research Institute actively promoted. In the era of cloud computing, Microsoft Research has been actively thinking about how to improve the current cloud infrastructure. The current cloud infrastructure is growing too fast. When the cloud data center exceeds a certain scale, it will face the challenge of stability and sustainability. Microsoft Research actively considers whether there are other alternative technologies to achieve the sustainability of cloud data centers. With growth, this technology is FPGA field programmable gate array.

Since the birth of the world's first FPGA product in 1985, the application of FPGAs has gradually expanded from digital signal processing and high-speed serial transceivers to embedded processors. FPGAs are widely used in micro-electronics fields such as communications, consumer electronics, industrial control, and test and measurement, but they are hardly used in large-scale data centers.

Microsoft Research announced in June 2014 that it successfully completed the test and applied FPGA technology (instead of GPU technology) to the Bing cloud data center. The results include 40 times faster than the traditional server in processing Bing's custom algorithm. Two times faster than existing Bing system, can reduce the number of existing servers and so on. FPGA technology accelerates the processing of large amounts of data on servers, helps solve big data challenges, and meets the needs of huge distributed workloads. The paper "A Reconfigurable Fabric for Accelerating Large-Scale Datacenter Services" describes how to test FPGA technology on Bing. In September 2014, Baidu quickly followed suit and announced that it will use FPGA technology for its own data center.

This means that the huge market space of the future data center will probably be achieved by FPGA. For Intel's relatively low market share of communication and network market, the combination with Altera is expected to open this market space.

The application of FPGA in the cloud data center will be discretely used from the CPU and the FPGA, packaged with the CPU and FPGA, and then integrated with the CPU and the FPGA. In the convolutional neural network algorithm for image recognition, encryption algorithm for security control, compression algorithm It plays an important role in carrying out big data processing. Microsoft's research shows that FPGAs consume about one-tenth the power of high-end GPUs, but the high-end GPUs have two to three times the image processing power of FPGAs, and a cluster of multiple FPGAs can achieve the GPU's image processing capabilities and maintain Low power consumption features . According to Intel’s estimates, by 2020, one-third of cloud data center nodes will use FPGA technology.

In terms of the Internet of Things, Altera's FPGA technology combined with Intel's Atom chip is expected to replace the existing technology in the industrial automation control market and advanced driver assistance system market, generating a $11 billion incremental serviceable market in 2020. Thanks to FPGA technology, users can self-programmatically replace existing ASSPs and ASICs to add new features, improve performance, reduce costs, and reduce time-to-market for key IoT applications by up to 50%.

At present, the competition in the FPGA market is changing and the application field is constantly expanding. In the past, FPGAs were largely dominated by the communications market, but industrial intelligence, automotive electronics, and the development of the Internet of Things have led to a large increase in the demand for flexible programmable FPGAs. FPGA vendors have tended to change to solution providers and actively develop Consumer market. Cooperation with big manufacturers such as Intel will determine the market's dominant solution status. Once Intel and Altera's solutions dominate, programmers and engineers will flock to this technology and will further consolidate Intel's market position.

In June 2014, Intel announced that the industry-leading Xeon processors and FPGAs were packaged into a single structure with slots compatible with standard Xeon processors. Intel said that FPGAs will provide Intel customers with a more powerful algorithm execution capability and higher performance. Industry studies have shown that FPGA-based accelerators will provide more than 10 times performance improvement, while Intel claims that it can also bring a more than double boost on this basis. It is also used between FPGAs and processors. Ultra low latency interface. In other words, Intel can bring more than 20 times the speed of server-based applications by using FPGAs!

By September, Altera began to use Baidu's deep learning application in Baidu to establish partnerships with FPGAs and Convolutional Neural Networks (CNN). Baidu said that the acceleration of FPGAs has huge application potential and will help FPGAs penetrate into the mainstream of heterogeneous computing. Utilizing programmable capabilities and features such as Altera's FPGA hard-core IEEE 754 floating-point multipliers and adders, servers and data centers meet the increasingly complex needs of search, big data, and deep learning.

It is estimated that the power consumption of global data centers accounts for 1%-10% of the world's total electricity consumption, and as the volume of data increases, power consumption also increases rapidly. Now, while reducing energy consumption growth, it continues to ensure that providing faster computing power has become a huge economic and ecological problem. Today's data centers are built on cheap electricity, and the size and throughput of data centers are limited by the ability to import electricity and heat. Companies that rely heavily on servers, such as Microsoft, Google, Facebook, and eBay, have plenty of incentives to increase code execution speed and reduce electricity bills.

At the same time, as Moore's Law has been coming to an end after about 50 years of long-distance running, the power of our single-core processors has reached its limit, and power consumption has increased faster than the speed of performance improvement. Then we have dual-core, quad-core or even 8-core core core chips. Finally, in order to make up for the lack of continuous improvement of the underlying semiconductor process, we are looking for other ways. If only waiting for better semiconductor materials to solve data center problems is not a viable option.

Why can FPGAs provide a significant increase in computing power with lower power consumption? For specific algorithms, FPGA-based hardware can provide highly segmented parallel computations—lower latency, higher throughput, and lower power consumption.

It is also worth noting that Intel is not the first company to plan production of heterogeneous processors with FPGAs. Xilinx has been fighting for this market for several years with its Zynq family, which uses an ARM processor with a Xilinx FPGA architecture. Although Zynq is not a data center-level processor, the gap is still very large, but Xilinx's experience and tools accumulated on Zynq will enable it to occupy a place in the future of low-power server market competition.

From the perspective of the five major markets for FPGAs, industries and communications have occupied approximately 70% of the market share. Computer applications and medical equipment have developed rapidly, and low-cost, low-power FPGAs have entered the consumer market. With the development of communications, industrial control, robotics, video analytics, and even consumer markets, the market size of FPGAs is even greater. Recently, there have also been opinions expressing the possibility that the FPGA is going to replace the CPU as the main chip, and at the same time, whether the FPGA with high investment threshold can continue to develop steadily in terms of localization.

Can FPGA become an antidote to the failure of Moore's Law?

As I mentioned earlier, the entire global FPGA market is now monopolized by four U.S. companies, and Altera and Xilinx monopolize 88% of the market. For Chinese companies, FPGA is almost a blank, present and future. Will still rely heavily on imports. If you encounter a ZTE ban sales crisis again, it may be a disaster for any business. The rapid growth of FPGA will bring more challenges to China. When will China be able to complete an important piece of this semiconductor puzzle? It is the focus of the government and major manufacturers.

However, according to a product information that the author has obtained, there have been domestic companies successfully developing compatible devices with the Lattice iCE series, and have been sold in the market. Therefore, the author is not too concerned about the fact that the Chinese people can finally overcome the fortress of the FPGA built by the Americans. This company not only has Lattice-compatible devices, but also more compatible with Altera's low-end devices. (In order to avoid suspicion of advertising, the company name will be erased)

Samsung's Galaxy series has also appeared in this domestic manufacturer's figure, can be confirmed on the side is not a reverse design, or Samsung's big manufacturers are scruples intellectual property issues will not be purchased.

As the demand for intelligent market changes faster and faster, the trend of giant-scale reduction of SoC asic has become irreversible. The amount of investment in the ASIC solution has increased significantly, with a long cycle and a substantial increase in market risk. Relatively speaking, FPGA technology is gradually moving towards the mainstream.

FPGAs have the advantage of parallel computing and can replace some ASICs and DSPs in high-performance, multi-channel applications. FPGA parallel computing can optimize traditional ASIC solutions for multi-channel processing.

With 50,000 currents as the critical point, FPGAs have a tendency to replace ASICs in high-value, relatively small-volume, multi-channel computing dedicated devices (such as radars, space shuttles, and routers), and this will be the Xilinx-held site.

The FPGA development cycle is 55% lower than ASIC and can be used to quickly capture the market. This may be Lattice’s most important winner in the larger consumer electronics market.

The ultimate goal of Intel’s Altera FPGA is to use its advantages in high performance, low power consumption and deep learning, and eventually integrate into the CPU technology to achieve the future of heterogeneous computing data centers.

From the capital market point of view, before the Ziguang Group had just completed a small stake in Lattice, Lattice (LSCC) is currently listed on the Nasdaq, the total market value of not more than 800 million US dollars, compared to Altera 16.7 billion US dollars purchase price, such a body The volume is an ideal acquisition goal. The global FPGA companies only have a total of four benchmarks (Xilinx, Altera, Lattice, and Microsemi). Intel has already bought Altera. Xilinx will not be sold as the industry leader for the time being, and the remaining two Scarcity bids sell one less. Taking advantage of the trend of entering the iPhone7 supply chain, I don't know if Lattice will soar in market value this year and become a target for bigwigs to fight for. However, it seems that the U.S. government will not approve Chinese companies to acquire such key technology-sensitive companies. Therefore, the Chinese mainland FPGA companies must be swiftly attempting to achieve alternative import as soon as possible so that they can get rid of the situation that is subject to human beings.

Lei Feng Network (search "Lei Feng Network" public concern) Note: This article is authorized by the Semiconductor Alliance Lei Feng network, if you need to reprint, please contact the original author, and indicate the author and source, not to delete the content.

Snowflake

Big Snowflake Light,Led Snowflake Lights,Christmas Snowflake Light

XINGYONG XMAS OPTICAL (DONGGUAN ) CO., LTD , https://www.xingyongxmas.com