In brief: Nvidia GPUs have become the cornerstone of the booming AI business, but the company is struggling to keep up with industry demand. ASIC chips, which provide a customized silicon design to hardware-accelerate specific computing workloads, could soon become the next big thing in the field.
According to recent job listings posted online, Meta wants to hire highly skilled hardware engineers to develop a new generation of application-specific integrated circuit (ASIC) designs. ASIC chips are customized at the hardware level for a particular computing use, which means they are not intended for general-purpose use like traditional CPUs. But ASICs are also extremely efficient at what they are designed to do, like crunching machine learning algorithms for AI workloads and chatbot services.
Meta's latest offerings spotted by The Register are for jobs in Bangalore, India, and Sunnyvale, California. In India, the Facebook parent is seeking an ASIC engineer to build hardware accelerators for its data centers. The new employee will become an integral part of the ASIC team working on a state-of-the art chip architecture to accelerate AI/machine learning algorithms, the job offer says.
Meta wants to build "green" data center accelerators despite the notorious hunger for energy shown by AI applications. The new ASIC engineer will need to work on both a new advanced hardware architecture and algorithms to power and test those new chips. Responsibilities of the role include development of performance and functional models to validate the new architecture, creation of machine learning kernels to analyze the ASIC chips, and optimization of the architecture for ML workloads.
The Facebook company is looking for someone with a bachelor's in computer science or engineering at minimum, with more than 10 years of experience and knowledge of computer architecture concepts such as processors, memory systems and on-chip interconnection networks. The candidate will also need a significant experience in low-level, object-oriented programming languages like C++ (and C).
Some of Meta's job offerings related to ASIC architectures were first posted in December 2023, The Register remarks, but they were reposted two weeks ago. The company previously said that it wants to develop its own "Inference accelerators," which would be brought online in 2024.
According to Nvidia CEO Jensen Huang, GPUs have played a pivotal role in improving Meta's "recommendation engines" and the company's overall revenues. But Nvidia has also said that it will likely be unable to provide enough GPU accelerators for anyone working in the AI business in the upcoming future. Meta is seemingly working on the elusive concept of artificial general intelligence (AGI) too, which is an incredibly hard task on the latest GPU tech but could greatly benefit from specialized ASIC silicon designs.
https://www.techspot.com/news/102023-meta-seeking-chip-engineers-develop-new-ai-ml.html
