The Superiority Of The Embedded FPGA (EFPGA) IP In ADAS Applications

Understand the basic knowledge of EFPGA IP, its advantages, and why it will become the key element of future advanced driving assistance system (ADAS) technology.

One of the main aspects of increasing vehicle electrification and autonomous driving is the popularization of advanced driving assistance systems (ADAS).Today, these systems are quickly applied to almost all vehicles in the market, and as technology matures, this trend will only continue.However, with the development of technology, the hardware challenges faced by ADAS designers are becoming more and more complicated.In this article, we will introduce ADAS's hardware requirements, how to fill these gaps in FPGA, and why EFPGA IP will become the next ADAS hardware trend.

Adas's hardware requirements

The development of ADAS in modern cars has brought some severe challenges to the bottom hardware.The most important goal is to ensure the safety of vehicle occupants in key tasks such as ADAS.This goal requires accurate, reliable and real -time work (that is, low latency).At the same time, these systems need to run under the low power consumption to maintain the battery life and extend the battery life of Electric Vehicle.

Figure: ADAS has caused huge pressure on the calculation hardware

Balanced these needs are a huge challenge for ADAS, because systems often depend on a large amount of data and computational dense tasks, such as machine learning algorithms.Therefore, ADAS hardware must obtain data, collect data through sensors, run machine learning algorithms, and then take action -all need to run in real time, and the lowest power consumption budget -this is obviously not easy.

Car designers need a flexible and scalable hardware platform to meet the rapid development specifications, which further exacerbates this challenge.

Hardware acceleration and FPGA

In order to solve the severe challenges faced by ADAS hardware, designers are using special hardware accelerators to improve performance, not traditional based onCPUThe architecture.From more conventional computing resources (such as CPU orGPU) Transfer, provides better performance and energy efficiency for specific applications through dedicated hardware accelerators.Based on this range, selection is usually between FPGA and ASIC. FPGA provides great flexibility, and ASIC provides higher performance.

A key feature of FPGA is that FPGA can provide high -level parallelism, and at the same time can still be programmed for specific workloads.The results show that FPGA provides significant value in terms of workload acceleration, especially in the case of performance and delay.In addition, compared with more conventional CPUs and GPU -based systems, FPGAs can provide the performance of each watt ZUI for the workload that needs to be accelerated, thereby helping the system balance between performance and power efficiency.

It is also important that due to the programming characteristics of FPGA, it is more advantageous than ASIC in terms of scalability.This adaptability is particularly important in the rapidly changing machine learning system such as ADAS.In addition, ASIC's specifications must be defined for several years in advance, and FPGAs can be updated and re -programmed in one minute.This function enables FPGA -based ADAS systems to provide ASIC's scalability and versatility that ASIC cannot achieve at all.

Due to these reasons and more reasons, more than 250 million FPGAs have been equipped in the car, of which more than 75 million are used for ADAS applications.

Industry Trend: EFPGA for ADAS

Although the hardware acceleration function is powerful, devices such as FPGA and ASICs are usually unable to run alone.Today's ADAS solution needs to integrate the hardware accelerator with the CPU. The purpose is to handle many universal and tissue tasks at the system level.For this reason, the heterogeneous computing platform (such as heterogeneous SOC) has become one of the most common platforms in acceleration and ADAS platforms.

Figure: EFPGA technology like Achronix's SpeedCore IP can be tightly coupled with CPU resources

Therefore, we believe that the next main trend of ADAS calculations will be嵌入式The rise of FPGA (ie EFPGA) IP in custom ASIC.With the help of EFPGA IP, designers can use the advantages of FPGA technology, and at the same time, their hardware accelerates tightly with other ASIC subsystems (such as CPU and I / O interfaces).By embedding FPGA and CPU into a custom SOC, EFPGA IP can significantly save costs, power consumption and space compared with the separated FPGA solution.

Specifically, our evaluation shows that compared with FPGA -based independent systems, EFPGA IP integration can help designers save 90 % of costs, reduce power consumption by 75 %, delay 100 times, and increase interface bandwidth by 10 times.Therefore, ADAS will gradually adopt heterogeneous solutions based on EFPGA technology in the near future.

Achronix's industry role

Today Achronix is a company that produces independent FPGA chips and EFPGA IP solutions at the same time.Thanks to flagship products such as SpeedCore EFPGA IP, so far, more than 15 million EFPGA -based products have been authorized in many industries.