The Next Generation of HPC Cognitive Workflows on Mesos
summary
Describes building high-performance cognitive computing workloads on Apache Mesos that combine interactive stateless services and compute- and data-intensive stateful jobs (including Spark-based engines). Covers integration of Mesos with batch schedulers such as LSF, Slurm, and PBS, and how Mesos can support technologies required for HPC cognitive systems: GPUs, high-performance file systems, MPI, task–CPU affinity mapping, low-latency operation, and priority-based job scheduling.
AI-written summary · about