Slurm vs Kubernetes which is better for workload management on accelerated computing.

If anyone had compared Slurm and K8s for GPU optimized management, please provide the recommendations.