The continuation of the heated discussion in global markets in 2017, coupled with the active involvement of various technology companies, artificial intelligence (AI) and autopilot / automotive technology, is expected to continue to be the most important factor in the semiconductor industry in 2018 Two 'A' applications with potential for growth.
The year 2017 is regarded by the technology industry as 'the first year of artificial intelligence development,' and the year 2018 is expected to be a year from artificial intelligence into the real world from the laboratory to the technology available to the general public in daily life. Such as AI on Chip. For example, Google developed a proprietary tensor processor (TPU) for its own server center, Project Brainwave, a low-latency deep learning cloud platform built by Microsoft from Intel, Intel) Nervana chips developed specifically for machine learning; Nvidia graphics processors (GPUs) and artificial intelligence chips continue to evolve.
There are also Qualcomm's neural network dedicated chips; TrueNorth, the first brain chip developed by IBM, and Spiking, a neuron adaptive processor developed for unsupervised learning of brain chips, as well as startups Cerebras and Groq Launches artificial intelligence chips in 2018. These types of artificial intelligence chips will be stationed in various types of devices that are closer to the public and work with high-performance computing (HPC) chips that perform big data analysis in cloud data centers to create new users for consumers Experience; and because AI chips and HPC chips are required to perform more complex computing functions and store large amounts of data, the circuit design is large, the structure will be more complex, chip testing is expected to be an integral part.
As the automotive electronics market, which has become the most powerful growth driver for all types of semiconductor applications in recent years, has become even more industrial focus due to the rise of advanced driver assistance system (ADAS) and AI-driven autopilot technology. ADAS / Autopilot The ultimate goal is to use electronic technology to help drivers eliminate any blind spots and human error and the possibility of human error, or even directly replace the use of machine-driven man in the expectation of achieving zero accident traffic safety.
In order to achieve the goal of fully autonomous driving, high-performance AI chips capable of handling large and complex data volumes generated by devices such as automobile sensors in time are the keys, while safety certification and quality management are the keys to the correct operation of the chip. Therefore, the safety standards that vehicle chips need to follow in design are also stricter than those of any other consumer electronic chip designs. Functional Safety of ISO 26262 Automotive Electronic / Motor System for Passenger Vehicles under 3.5 Tons Standard that covers the entire 'safety lifecycle' of a vehicle from design, production, operation and maintenance to final scrap recycling and can effectively enhance the reliability of the vehicle's system safety functions. It is supported by Tier 1 Automotive component suppliers attach great importance to the standard.
In addition, the Tier 1 automotive electronics supply chain needs to meet the reliability standards of AEC-Q100 (IC chip), 101 (split components), 200 (passive components), and zero-failure supply chain quality management standards ISO / TS 16949 specification; ISO 7637-2 / 16750-2 power load test is also essential when the ADAS / autopilot vehicle chip with AI function, the depot on chip reliability requirements will be higher, so the test technology will play a more important role This will be the key to the safe driving of a self-driving vehicle, and it will mean that automotive semiconductor testing is entering a completely new era.