Modern multi GPU HPC systems expose substantial computational capacity, yet inefficient GPU allocation often leads to wasted energy and underutilization. In practice, GPU applications exhibit heterogeneous and nonlinear scaling, making it inefficient to always use all available GPUs. We present EcoSched, an online scheduler that jointly optimizes GPU count selection and application coscheduling to improve workload level efficiency on multi GPU systems. EcoSched uses lightweight runtime profiling to estimate relative performance across GPU counts, applies a score based policy to balance energy efficiency and idle resources, and incorporates NUMA aware placement to mitigate interference. We implement EcoSched on heterogeneous CPU GPU platforms and evaluate it with diverse workloads on H100, A100, and V100 systems. EcoSched achieves up to 14.8% energy savings, 30.1% makespan improvement, and 40.4% EDP reduction over baseline schedulers, with modest performance overhead. These results show that jointly selecting GPU counts and coscheduling actions is essential for efficient multi GPU workload execution.

翻译：现代多GPU高性能计算系统具备强大的计算能力，但低效的GPU分配常导致能量浪费与资源利用率不足。实际应用中，GPU工作负载呈现异质非线性扩展特性，使得始终使用全部可用GPU效率低下。我们提出EcoSched——一种在线调度器，通过联合优化GPU数量选择与应用程序协同调度，提升多GPU系统的工作负载级效率。EcoSched采用轻量级运行时性能分析，估算不同GPU数量下的相对性能；应用基于评分策略平衡能效与闲置资源；并引入NUMA感知的放置策略以缓解干扰。我们在异构CPU-GPU平台上实现了EcoSched，并在H100、A100及V100系统上使用多样化工作负载进行验证。与基准调度器相比，EcoSched最高可节省14.8%能耗、缩短30.1%执行时间、降低40.4%能量延迟积，且额外性能开销可控。实验表明，联合优化GPU数量选择与协同调度行为是实现高效多GPU工作负载执行的关键。