1 Introduction The perpetual demand for computational power in scientific computing incessantly propels high-performance computing(HPC)systems toward Zettascale computation and even beyond[1,2].Concurrently,the ascens...1 Introduction The perpetual demand for computational power in scientific computing incessantly propels high-performance computing(HPC)systems toward Zettascale computation and even beyond[1,2].Concurrently,the ascension of artificial intelligence(AI)has engendered a marked surge in computational requisites,doubling the required computation performance by every 3.4 months[3].The collective pursuit of ultra-high computational capability has positioned AI and scientific computing as the preeminent twin drivers of HPC.展开更多
基金supported by the Natural Science Foundation of Hunan Province(No.2022JJ10066)the National Natural Science Foundation of China(Grant No.62272477).
文摘1 Introduction The perpetual demand for computational power in scientific computing incessantly propels high-performance computing(HPC)systems toward Zettascale computation and even beyond[1,2].Concurrently,the ascension of artificial intelligence(AI)has engendered a marked surge in computational requisites,doubling the required computation performance by every 3.4 months[3].The collective pursuit of ultra-high computational capability has positioned AI and scientific computing as the preeminent twin drivers of HPC.
