Exactly Solving Hard Permutation Flowshop Scheduling Problems on Peta-Scale GPU-Accelerated Supercomputers

Makespan minimization in permutation flow-shop scheduling is a well-known hard combinatorial optimization problem. Among the 120 standard benchmark instances proposed by E. Taillard 1993, 23 have remained unsolved for almost three decades. In this paper, we present our attempts to solve these optimality using parallel Branch-and-Bound (BB) on GPU-accelerated Jean Zay supercomputer. We report exact solution of 11 previously problem and improved upper bounds eight instances. The problems requires both algorithmic improvements leveraging computing power peta-scale high-performance platforms. challenge consists efficiently performing depth-first traversal highly irregular fine-grained search tree distributed systems composed hundreds massively accelerator devices multicore processors. discuss design implementation permutation-based BB experimentally evaluate its performance up 384 V100 GPUs (2 million CUDA cores) 3840 CPU cores. proof largest solved instance about 64 CPU-years computation—using 256 over 4 agents, completed 13 hours, exploring [Formula: see text] nodes.