Link to paper

The full paper is available here.

You can also find the paper on PapersWithCode here.

Abstract

Common myths and legends exist among members of the high-performance computing community.
These myths are based on evidence or argumentation, not scientific facts.
These myths can be used to discuss possible new directions for research and industry investment.

Human society has myths and legends, including in the HPC community
HPC drives powerful computers and technologies forward
AI language models cannot create or share fictional content
12 myths about HPC are discussed
Myth 1: Quantum Computing will take over HPC
Myth 2: Everything will be Deep Learning
Myth 3: Extreme specialization as seen in smartphones will push supercomputers beyond Moore’s Law
AI is now in the palm of everyone’s hand
GPUs have been successful in HPC
Multitudes of hardware customization per each facet of the workload is not likely to be successful
Amdahl’s Law limits potential speedup
Weak scaling is important for modern supercomputers
Load balancing is difficult with heterogeneous accelerators
Dark silicon is becoming increasingly difficult to utilize
Plethora of accelerators is only beneficial for small programs on small machines

Debates myths in HPC community
Many myths form core of thinking
Some points may settle in the future, others may not
Serious treatment needed to guide future directions for research, industry and government investment