Given the success of AIDArc 2018 and the significantly
increased interest in utilizing AI to improve computer architecture in the past
year, we are thrilled to organize AIDArc 2019, held in conjunction with
Recent advancements in machine learning algorithms, fueled
by increased data availability and high-performance computing infrastructure,
have led to successful applications of machine learning (and AI in general) in
numerous disciplines and domains. Although much attention has been drawn in the
computer architecture community on accelerating machine learning, limited
research has been conducted to utilize the power of AI/ML to help architects
design better computer architectures and systems.
The AIDArc Workshop is intended to bring together
researchers, scientists and practitioners across academia and industry, to
share early discoveries, successful examples, and opinions on opportunities and
challenges regarding utilizing AI to assist computer architecture designs.
Research along this line may potentially transform the way computers are
designed and optimized. It may also lead to interesting “self-evolving
architecture”, where AI helps to speed up computers which, in turn, are used to
speed up the AI.
Topics of submitted papers include, but not limited to, the
exploration of artificial intelligence in assisting the design and optimization
Various components of computer system architecture, e.g., branch predictor,
cache, memory, I/O, interconnection networks, etc.
Various design objectives of computer system architecture, e.g., power/energy,
performance, resource, reliability, security, etc.
Different types of computer architectures and systems, e.g.,
embedded/mobile/wearable devices, CPUs, GPUs, special-purpose accelerators,
datacenters, HPCs, etc.
Interaction of computer architecture with other layers, e.g., operating
systems, compilers, circuit-level designs, etc.
Papers will be reviewed based on originality, novelty,
technical strength, presentation quality, correctness and relevance to the
workshop scope. Early but novel works on related topics are highly encouraged.
Lizhong Chen, Oregon State University,