A Guide to Semiconductor IP Core

Ⅰ What does semiconductor IP mean?

IP (intellectual property) in the semiconductor industry is also called IP core, which refers to the mature design of circuit modules with independent functions in the IC chip.

Wikipedia: “In electronic design, a semiconductor intellectual property core (SIP core), IP core, or IP block is a reusable unit of logic, cell, or integrated circuit layout design that is the intellectual property of one party.”

The circuit module design can be applied to other chip design projects that include the circuit module, thus reducing the design effort, shortening the design cycle, and improving the success rate of the chip design. The semiconductor IP core can also be understood as the middle building block of a chip design.

Generally speaking, a complex IC chip is composed of a circuit part designed by the chip designer himself and several outsourced IP cores connected. As shown in Figure 1, to design a chip with such a structure, the design company can outsource all the IP cores in the chip (different color modules) and design only the creative and self-designed part of the chip (indicated by green) and connect the parts.

Figure 1. The self-designed part and multiple IP cores are integrated together to form the design of a complex chip

The IC design process in Figure 1 is like the system board development process in Figure 2, which is the process of layout, placement, and signal connection with existing, mature IP cores (or chips), and this process can be called the reuse of IP cores. The difference is that in the system board in addition to the chip and connection lines, system developers rarely develop their own ICs. In the chip design process, in addition to the use of outsourced IP cores on the chip, generally speaking, the chip designer also has to design part of their own circuit, and complete the signal connection between the various parts, and finally, the entire chip's function and performance of the pre-manufacturing repeated checks and verification.

Figure 2. Building a complex chip from IP cores, similar to building a complete system board with chips

If the above introduction still seems too professional, you can also use the analogy of the IC design with a puzzle, you can abstractly understand the chip shown in Figure 1 into the puzzle shown in Figure 3. The outsourced IP cores of different functions in the chip are represented by different colored blocks, and the self-designed circuit parts are represented by green blocks, so the design process of a complex chip is like trying to put together this picture. The same thing is that the existing blocks (IP cores) are used to put together the beautiful picture (complex chip). The difference is that, while the picture only needs to consider the shape of the block, the chip design has to consider many parameters and indicators of the IP core and to connect each IP core and the self-designed part correctly to ensure the correct function and performance of the whole chip.

Figure 3. IP cores make the design of complex chips easier, and IP reuse is similar to a "puzzle"

IP cores are adopted by other chip design companies and are known in the industry as IP multiplexing. Specially designed relatively independent circuit function module, the purpose is to promote to other chip design companies for reuse, this design work is called IP development. The company specializing in IP development is called IP vendor, or IP provider. IP vendors selling IP to chip design companies is a kind of IP trading behavior.

Ⅱ The origin and role of IP

The origin of IP has to be told from the early chip design process. The early chip integration scale is limited, the design complexity is not high, all the circuitry on the chip is done by the chip designer independently. IC companies with low design levels and limited capability could only design small-scale and simple chips. IC companies with high design levels and capabilities can only design chips with large-scale and complex functions. The early high-end chips are basically held by a small number of large international chip companies.

With the modern information society to enhance the requirements of the chip, the size of the chip is exponentially increased, the complexity has increased sharply. Small and medium-sized chip companies to independently complete a complex chip design has become almost impossible. Especially in the late 1980s, the IC industry, the emergence of the foundry business model, a large number of small and medium-sized chip design companies (Fabless) came into being. During this period, the IC design industry urgently needs to solve the problem of small chip companies can not design large chips.

There are many inspiring ideas to solve this problem. For example building blocks and puzzle drawing toys; designing large machines from standard parts; designing large software by software subroutine (or middleware) calls; building large electronic systems with chips, etc... The idea is to reuse pre-designed mature components to build more complex systems, eliminating the consideration of internal problems of the components.

IP core development and IP reuse in the IC design industry are inspired by these ideas, and IP cores are similar to the above-mentioned building blocks, which are pre-designed circuit modules with independent functions. With such building blocks as IP cores, large and complex chip designs become easier, shorter in cycle time, and easier to succeed.

At present, many small and medium-sized IC design companies have limited design capability and level, but out of the need to seize the market and shorten the chip design cycle, they will outsource many IP cores to complete their own chip design projects. The number of IP developers and IP providers in the industry is increasing and becoming more and more professional. Various functions and types of IP cores are emerging, and IP trading activities are becoming more and more common, and the transaction amount is getting larger.

Ⅲ Types of IPs and examples

There are three levels of IP cores: Behavior, Structure, and Physical, which correspond to three types of IP cores, namely IP Soft Core designed by hardware description language, IP firm core which completes structure description, and IP hard core which is based on physical description and verified by process.

1. IP Soft Core: It is a circuit module designed in hardware description language (HDL) with independent functions. In terms of the degree of chip design, it has only been optimized and functionally verified at the RTL level of design and is usually submitted to the user in HDL text form. Therefore, it does not contain any physical implementation information, and thus the IP soft core is independent of the manufacturing process.

With the purchase of an IP soft core, the user can synthesize the correct gate circuit level design netlist and can perform subsequent structural design with great flexibility. With the help of EDA synthesis tools, users can easily synthesize with other IP soft cores, as well as independently designed circuit parts, and design into chips with different performance according to various semiconductor processes.

2. IP Firm Core: It is designed between IP soft core and IP hard core, which not only completes all the design of IP soft core, but also completes the design aspects such as gate level circuit synthesis and timing simulation. Generally, it is provided to the user in the form of a gate level circuit netlist.

3. IP Hard Core: It provides the circuit modules at the mask level in the final stage of circuit design. IP hard cores have both predictability of results and can be optimized for power and size for a specific process or a specific IP provider.

Therefore, the three types of IP cores are products of the circuit function module design at different design stages, as shown in Figure 4.

Figure 4. Different types of IP cores are available at different stages of circuit function module design

After careful evaluation and selection, users purchase IP cores from IP vendors and start designing their own chips. As mentioned earlier, a complex chip generally consists of a purchased IP core and a circuit part designed by the user himself. The chip design process includes three phases: behavioral, structural, and physical. The work in the behavioral and structural design phases is generally referred to as front-end design, and the work in the physical design phase is generally referred to as back-end design. The diagram in Figure 5 illustrates that different types of IP cores are added to the overall chip design in different design phases.

Figure 5. The three types of IP cores are combined into the design at different stages of the chip design to form a whole

Examples of IP cores are typically ARM's various types of CPU IP cores. Many IP suppliers provide DSP IP cores, USB IP cores, PCI-X IP cores, WiFi IP cores, Ethernet IP cores, embedded memory IP cores, etc. There are a wide variety of IP cores.

If divided into broad categories, they can be broadly classified into processor and microcontroller IP, memory IP, peripheral and interface IP, analog and hybrid circuit IP, communication IP, image and media IP, etc.

Some of the world's large EDA suppliers are also IP suppliers. For example, the United States Synopsys can provide thousands of various types of IP, covering Logic Libraries, Embedded Memories, Analog Libraries, Wired and Wireless Interface, Security, Embedded Processors, and Subsystems, etc.

Ⅳ Development and application of IP

From Figure 3 and Figure 4, it can be seen that semiconductor IP core development is relatively similar to the IC design process, both need to keep up with the development trend of IC technology, continuously accumulate chip design technology, and understand the pace of chip manufacturing process development. The difference is that the business model of chip design companies is to design complete chips and bring them to the market to realize profits from chip sales. IP developers do not design the complete chip, they only design certain functional modules of the chip, the pursuit of these functional modules (IP cores) is the most optimal design. IP developers seek these IPs can be purchased by many chip design companies to adopt, and realize the IP sales profits.

However, the technical difficulty and requirements of IP development are very high, and it is not something that any IC design company wants to do. IP cores as the building blocks of chip design can be compared with the building blocks of software design (middleware). Software building blocks can hide small flaws, integrated system software can also be put into practical applications, through the patching and upgrading of the way to continue to improve.

The IP core must be very perfect, with no defects at all. Because, any small defect in the IP core may affect the success or failure of the entire complex chip design, affecting the millions of dollars or even hundreds of millions of dollars spent on the entire chip development input and output. Therefore, complex IP cores are extremely expensive, ranging from hundreds of thousands to millions of dollars. In order to facilitate the promotion and reduce the user's R & D investment, IP vendors generally use the business model of charging the initial IP license fee + later royalty calculated by chip shipments.

The second reason is the multi-parameter requirements of IP, including process nodes, power, power consumption, performance, etc... These require IP developers to be very familiar with IC design, manufacturing processes, and industrial applications. The third is the user's requirements for the credibility of the IP provider, including the company's technical strength, influence, and stability. It is difficult for users to accept IP from a smaller, unstably developed IP vendor, even if these IPs are cheap or even free. Users would rather buy more expensive IPs provided by famous and big IP vendors because such IPs have quality assurance and follow-up service guarantees.