The application-specific integrated circuit (ASIC) uses a hard-wired fixed mode, while the field-programmable gate array (FPGA) uses a configurable chip approach, which is quite different. Programmable devices are the new force at the moment, and hybrid technology will also play a role in the future. As with other technologies, reports of outdated ASIC technology are immature. The number of new ASIC products may drop significantly, but sales are still quite high, especially in the Asia Pacific region. In addition, hybrid methods, such as structured ASICs, have also injected new life into the technology. At the same time, FPGAs (and other programmable logic devices) are also playing a role, winning important mass markets and moving upwards from low-end applications. Each technology has its supporters. In general, ASICs are used for large projects, and FPGAs are more suitable for small projects that need to be quickly placed on the market and support remote upgrades. ASIC and FPGA vendors are not able to agree on the advantages and disadvantages of these two technologies, and they also have different views on suitable application areas. The above technologies and their derivatives will likely exist for a long time to come. David Greenfield, senior director of high-density FPGAs at Altera Corp., points out that the main advantage of FPGA technology is still the time it takes to bring the product to market. He said: "In the current new design, the choice of FPGA tends to exceed ASIC. ASIC technology has its value, its performance, density and unit capacity are quite excellent, but with the development of FPGA and ASIC development Increasing costs will lead to a shrinking market share of ASICs.†After the above trends, it is the development of FPGAs in terms of performance, density and manufacturing costs. Greenfield pointed out that high performance used to be an advantage of ASICs beyond FPGAs, when FPGAs were inferior in performance and functionality. As the manufacturing process of the chip has grown from 180nm to 130nm or even 90nm, the above situation has changed a lot. Now the performance of the FPGA can meet the needs of most applications (except for the most demanding applications), while the density level reaches the logic design. 80%. He explained: "Some system designers also recognize that the market for ASICs is in extremely high performance/density products, which are very risky in the market. NRE (non-repetitive engineering) and development costs for such devices It is the highest." Altera pointed out that earlier FPGAs were only used for prototyping or low-capacity/low-density applications, and the technology is now widely used in consumer electronics and is also used in high-density applications. Greenfield pointed out that the highest density FPGA (90 nm) is still significantly more expensive than ASIC. He said: "But even for the highest density applications, the results are still inclined to FPGA technology when considering factors such as development and NRE costs." Texas Instruments' ASIC work is based on a unit approach and serves a large number of large customers. These ASIC devices typically have five times the average gate count of industry-standard ASICs and are used in highly complex, high-capacity applications. These applications require a high degree of differentiation between commercial networking and telecommunications technologies. John DiFilippo, a silicon technology designer at TI's ASIC communications infrastructure business unit, said: "The initial investment in ASIC development is high. However, in high-yield situations, ROI will improve significantly because of the smaller chip size and lower unit cost. FPGAs are a better choice if the unit price of the finished product is less important, or if the time to market is shorter, or if the initial investment is low." DiFilippo believes that TI's customers demand good price/performance, which is difficult to achieve for FPGAs and structured ASICs. FPGAs and fabric ASICs are better suited to a wide range of intermediate markets. "FPGAs and fabric ASICs are suitable for low-capacity, short-lived applications, and customers are willing to sacrifice product features and performance, but still achieve system goals," he said. However, TI recognizes both competing technologies. TI introduces new features for cell-based ASIC devices that provide flexibility similar to gate arrays, shorter cycle implementations, and lower cost when devices require redesign. TI has also developed "platform" ASIC products that can be leveraged across multiple customer product lines and noted that it can reduce the development cost per unit of system. TI believes that the ASIC approach is best suited for the following situations: â– The number of doors and storage locations exceeds 10 million; â– There are a large number of gigabit connections; â– The main clock frequency is higher than 300 MHz at the lowest power consumption; â– Cost-sensitive applications. Xilinx noted that the debate over whether FPGAs can be a viable alternative to ASICs and related standard devices has continued for nearly a decade. Erich Goetting, vice president of senior product division at Xilinx, points out that although FPGAs have made significant progress over time, until recently, designers had to use large, expensive devices to achieve high performance, and DSP and RISC processing for specific applications. Or high speed serial connection. Xilinx now offers a new "domain-optimized platform FPGA" (Virtex-4) that can be used to add or subtract chip designs based on the ASMBL (module) architecture for functional requirements and cost targets. Goetting pointed out: "ASMBL is a modular framework for silicon technology subsystems that provides a new approach to FPGA development for fast and inexpensive deployment of platforms for different applications." For example, a design may require high-speed DSP functionality, but not Advanced logic is a must. With the ASMBL architecture, Virtex-4 allows users to choose the right mix of logic, DSP, memory and other functions (column grouping) based on the specific design. It has been pointed out that the columnar architecture can achieve up to 17 device choices and provide more functionality on the "price point". Xilinx pointed out that because NRE costs are almost non-existent (usually shared by FPGA vendors), FPGAs generally have a price advantage. Goetting pointed out: "The development cost of ASICs has risen rapidly and rapidly, and as the capabilities of FPGA platforms continue to increase, this makes the balance of competitive advantages tilt toward FPGAs. In addition to being widely used in analog/mixed-signal applications, ASICs are hard to provide relative to FPGAs. Other significant functional advantages. FPGAs can also save costs in other ways, with software downloads to fix errors and to facilitate system performance when adding new features. Figure 1: The graph shows the growth rate of the global market for FPGAs (a type of programmable logic device) and ASICs published by the Semiconductor Association. GE Fanuc Automation believes that the "real advantage" of FPGAs has two aspects: First, it can be developed quickly with reliable standard components, and can be easily modified to add new features; second, it can be developed or during product life. Corrected the error internally. GE Fanuc senior engineer Richard Reed pointed out that unlike ASICs, FPGAs have more features as built-in standards, such as testability or JTAG interfaces, which saves design time and cost. FPGAs have accelerated the introduction of products. Reed pointed out: "The use of standard components in large quantities makes the price of FPGAs more competitive with ASICs. For applications with longer life cycles and higher yields, it is sometimes more appropriate to convert designs into ASIC-specific chips." In terms of the advantages of ASICs, Reed pointed out that ASICs can be run immediately after power-up, with more package options for unit logic size, and some analog logic. In contrast, the FPGA load configuration takes time to enter the memory and therefore does not work immediately. In addition, the packaging of the FPGA is also more complicated. Nallatech, a developer of FPGA computing systems and hardware and software, acknowledges that ASICs achieve "high performance levels" for the specific types of functions and specialized applications for which they are designed. However, Craig Sanderson, application engineer at Nallatech Systems, points out that using ASICs for high-performance processing functions such as industrial simulation, modeling, or imaging can have "commercial impact." Figure 2: FPGAs successfully applied to industrial products, such as NI's CompactRIO FPGA chips embedded in reconfigurable acquisition and control systems play an important role. GE FANUC's PA CSystems RX3i controller also uses FPGA technology. The above "high performance" applications are usually of the small to medium size. Although Sanderson did not give a critical point of cost efficiency for implementing applications with ASICs, he pointed out that from a cost/risk perspective, ASICs are not feasible for relatively small-scale applications. He added that regardless of the size, "FPGA vendors generally say that they will promote FPGAs, not ASICs." †Nallatech also believes that FPGAs avoid high NRE costs and have other advantages. The reprogrammability of FPGAs enables a more flexible development path, reducing risk and cost. In contrast, ASIC development must be “first and for sureâ€. The on-site reprogrammability of the FPGA allows developers to modify the chip by running the program on-chip with a software upgrade package instead of replacing the chip. FPGAs can even be upgraded remotely over the Internet. Obsolescence control refers to the available resources of existing FPGA application design as a new generation of device recompilation. For many applications, FPGA vendors have said that performance is comparable to ASICs. Sanderson points out: "In terms of high-performance applications, FPGAs provide ample resources to achieve ASIC-like functionality while being much more powerful than standard processors." Due to the reprogrammable nature of the FPGA, applications can be debugged and tested in real hardware. Sanderson added: "As far as ASICs are concerned, all tests must be simulated before entering the physical implementation of the ASIC hardware phase. It is too late to find problems in the hardware phase." Gricha Raether, manager of industrial control and distributed I/O products at National Instruments (NI), pointed out that ASICs and FPGAs were used early in large-scale applications, such as machine manufacturing and OEM-based integration, which helped to spread traditionally higher Development costs. The reason for the higher cost, he believes that the above-mentioned devices have a long development cycle and designers need to master a large amount of expertise in development tools, especially the design work and manufacturing steps of ASICs are time-consuming. FPGA products are well designed and can be programmed directly. He pointed out that in this regard, FPGAs will gradually replace the actual integrated circuits. Because FPGAs have customizable flexibility, vendors may charge more. Designing IC packages and printed circuit boards adds more cost, which is the same for both technologies, but especially for ASICs. Industrial life cycle Raether believes that FPGAs are also good for industrial products with longer life cycles. This is mainly due to the fact that the technology can be easily reprogrammed according to the new version and can be reprogrammed in the field. He said: "Designers using FPGA technology should consider the expansions and modifications that may be required, and should be prepared in advance when selecting the number of FPGA gates." This requires the number of gate arrays and chip programming required to implement the functions. A subtle balance is achieved between the performance achieved, in addition to the required "storage space". Altera also believes that FPGAs are "very beneficial" for industrial products with longer life cycles, even though such products will decline in sales over time. Greenfield pointed out: "The FPGA process does not require a minimum number of reservations and a longer life, which is an important reason for its uniqueness. Many industrial customers who have been designing ASIC products for five years now use FPGAs instead of ASICs." There are many reasons. For example, ASIC requires a minimum number of reservations, which is very inflexible; ASIC process technology is outdated, or need to be converted to lead-free chip packages. The aging of process technology is a problem that chip manufacturers must face. Greenfield pointed out: "This problem is especially serious for ASIC companies because their customer base is very limited and it is very likely that they will be difficult to get out of trouble." The role of software tools Developing FPGA solutions is quite complex and requires the right software tools. Nalltech's Sanderson points out that FPGA design tools are constantly evolving, especially those that use high-level languages ​​or interfaces for application development, such as MatLab/Simulink from Mathworks. He said that high-level languages ​​are especially important for FPGA companies because they can package the necessary application functions into one or more FPGA devices. Sanderson pointed out that previously, this functionality had to be implemented on one or more DSPs or microprocessors, plus some fixed-function ASICs to implement the connection. ASICs and FPGAs are integrated circuits (ICs), but they are different. As the name suggests, an application-specific integrated circuit (ASIC) is a hard-wired silicon chip that is designed to meet the specific application needs of an electronic product or family of products for use in a variety of consumer electronics and industrial products. Field Programmable Gate Array (FPGA) is an emerging IC technology that includes thousands of logic cells connected by programmable switches to meet different design requirements through the logical interconnection of cells. In addition to the logic blocks, the other programmable components of the FPGA are I/O blocks (as an interface between the internal single-line and external pins of the chip) and interconnect interfaces (route the I/O signals of other components to the appropriate network). The ability to reprogram is the biggest advantage of this type of device. The structured ASIC forms the middle ground of the above method, and it uses a metal base layer to pre-manufacture design elements (logical units, memories, I/O, etc.) common to many applications. Data for specific applications can be added in the final few metal layers, which greatly reduces the number of mask layers and reduces the cost of pre-development for development. One of the design complications faced by designers is to enable communication between multiple functional blocks in a single FPGA. Nallatech's DimeTalk tool (currently only available for Nallatech hardware) is said to solve the problem of FPGA communication system development. Design tools are required for each chip technology. Xilinx pointed out that due to the characteristics of the FPGA design flow, FPGA users do not have to consider manufacturing yield and sub-micron issues. In addition, FPGAs are also easy to use, low cost and short time to market. Goetting added: "As a standard product, FPGAs have been thoroughly tested and can function properly because FPGA vendors have solved physical design, verification and characterization issues." Xilinx is a logic, DSP and embedded processing device. Provides integrated design and debugging tools, as well as interfaces for third-party tools. Depending on the vendor, the software that programs the FPGA differs in terms of content and value-added features such as compilation and editing tools. NI's Raether emphasizes that skilled use of the above tools requires years of experience and training. He said: "Some of the more advanced tools are entering the market, but you need to have a good understanding of the internal mechanisms of the FPGA to use these tools." VHDL (Extreme [High Speed] Hardware Description Language) is the most commonly used development language. Raether said that NI's LabView software completely abstracts the internal operating mechanism of the device, and it is the only software that currently implements this feature. The software programs the FPGA in the programmable automation controller through a graphical development environment. FPGAs are also facing challenges. Xilinx pointed out that the static power and size limitations of high-density chips are FPGA issues because programmable chips require more transistors to perform logic functions. While FPGA processes have evolved toward new, smaller process technologies, process-level, circuit-level, and architecture-level innovations seem to be increasingly constrained by power issues. Goetting pointed out, for example, that Xilinx reduced the power consumption of its 90nm Virtex-4 family by half compared to the 130nm processor by using trioxide technology and integrated platform capabilities. National Instruments' Raether noted that FPGA development faces issues such as development time, compatibility with industry specifications, and the allocation of appropriate development resources for board and package design. An FPGA similar to NI CompactRIO (see photo) integrates an FPGA to help with product development. GE Fanuc's Reed is interested in dedicated standard product (ASSP) components derived from different traditional ASIC designs. GE Fanuc uses the available IP (intellectual property) cores for FPGAs to increase its productivity; vendors use the same technology to introduce many modified versions of standard components to accommodate many smaller market segments. Reed concluded: "We can introduce embedded processors to better match the functions required for the combination, and don't have to pay for the features we don't need. This is because the IP cores can be reused. We can quickly match these IP cores. , made into standard components." What we are currently trying to solve is the dispute between "application-specific integrated circuits" and "programmable devices." And whether the final solution is hybrid chip technology, let us wait and see docking station for macbook air,docking station for laptop,docking station usb c,USB C HUB,thunderbolt 3 usb type c hub Shenzhen Konchang Electronic Technology Co.,Ltd , https://www.konchangs.com
