اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدBidding policies for market-based HPC workflow scheduling
102
0
0.0
(
0
)
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
This paper considers the scheduling of jobs on distributed, heterogeneous High Performance Computing (HPC) clusters. Market-based approaches are known to be efficient for allocating limited resources to those that are most prepared to pay. This context is applicable to an HPC or cloud computing scenario where the platform is overloaded. In this paper, jobs are composed of dependent tasks. Each job has a non-increasing time-value curve associated with it. Jobs are submitted to and scheduled by a market-clearing centralised auctioneer. This paper compares the performance of several policies for generating task bids. The aim investigated here is to maximise the value for the platform provider while minimising the number of jobs that do not complete (or starve). It is found that the Projected Value Remaining bidding policy gives the highest level of value under a typical overload situation, and gives the lowest number of starved tasks across the space of utilisation examined. It does this by attempting to capture the urgency of tasks in the queue. At high levels of overload, some alternative algorithms produce slightly higher value, but at the cost of a hugely higher number of starved workflows.
قيم البحث
اقرأ أيضاً
Traditionally, on-demand, rigid, and malleable applications have been scheduled and executed on separate systems. The ever-growing workload demands and rapidly developing HPC infrastructure trigger the interest of converging these applications on a s
ingle HPC system. Although allocating the hybrid workloads within one system could potentially improve system efficiency, it is difficult to balance the tradeoff between the responsiveness of on-demand requests, the incentive for malleable jobs, and the performance of rigid applications. In this study, we present several scheduling mechanisms to address the issues involved in co-scheduling on-demand, rigid, and malleable jobs on a single HPC system. We extensively evaluate and compare their performance under various configurations and workloads. Our experimental results show that our proposed mechanisms are capable of serving on-demand workloads with minimal delay, offering incentives for declaring malleability, and improving system performance.
High-performance computing (HPC) is undergoing significant changes. Next generation HPC systems are equipped with diverse global and local resources, such as I/O burst buffer resources, memory resources (e.g., on-chip and off-chip RAM, external RAM/N
VRA), network resources, and possibly other resources. Job schedulers play a crucial role in efficient use of resources. However, traditional job schedulers are single-objective and fail to efficient use of other resources. In this paper, we propose ROME, a novel multi-dimensional job scheduling framework to explore potential tradeoffs among multiple resources and provides balanced scheduling decision. Our design leverages genetic algorithm as the multi-dimensional optimization engine to generate fast scheduling decision and to support effective resource utilization.
Workflow decision making is critical to performing many practical workflow applications. Scheduling in edge-cloud environments can address the high complexity of workflow applications, while decreasing the data transmission delay between the cloud an
d end devices. However, due to the heterogeneous resources in edge-cloud environments and the complicated data dependencies between the tasks in a workflow, significant challenges for workflow scheduling remain, including the selection of an optimal tasks-servers solution from the possible numerous combinations. Existing studies are mainly done subject to rigorous conditions without fluctuations, ignoring the fact that workflow scheduling is typically present in uncertain environments. In this study, we focus on reducing the execution cost of workflow applications mainly caused by task computation and data transmission, while satisfying the workflow deadline in uncertain edge-cloud environments. The Triangular Fuzzy Numbers (TFNs) are adopted to represent the task processing time and data transferring time. A cost-driven fuzzy scheduling strategy based on an Adaptive Discrete Particle Swarm Optimization (ADPSO) algorithm is proposed, which employs the operators of Genetic Algorithm (GA). This strategy introduces the randomly two-point crossover operator, neighborhood mutation operator, and adaptive multipoint mutation operator of GA to effectively avoid converging on local optima. The experimental results show that our strategy can effectively reduce the workflow execution cost in uncertain edge-cloud environments, compared with other benchmark solutions.
Edge computing has been developed to utilize multiple tiers of resources for privacy, cost and Quality of Service (QoS) reasons. Edge workloads have the characteristics of data-driven and latency-sensitive. Because of this, edge systems have develope
d to be both heterogeneous and distributed. The unique characteristics of edge workloads and edge systems have motivated EdgeBench, a workflow-based benchmark aims to provide the ability to explore the full design space of edge workloads and edge systems. EdgeBench is both customizable and representative. It allows users to customize the workflow logic of edge workloads, the data storage backends, and the distribution of the individual workflow stages to different computing tiers. To illustrate the usability of EdgeBench, we also implements two representative edge workflows, a video analytics workflow and an IoT hub workflow that represents two distinct but common edge workloads. Both workflows are evaluated using the workflow-level and function-level metrics reported by EdgeBench to illustrate both the performance bottlenecks of the edge systems and the edge workloads.
Designing efficient and scalable sparse linear algebra kernels on modern multi-GPU based HPC systems is a daunting task due to significant irregular memory references and workload imbalance across the GPUs. This is particularly the case for Sparse Tr
iangular Solver (SpTRSV) which introduces additional two-dimensional computation dependencies among subsequent computation steps. Dependency information is exchanged and shared among GPUs, thus warrant for efficient memory allocation, data partitioning, and workload distribution as well as fine-grained communication and synchronization support. In this work, we demonstrate that directly adopting unified memory can adversely affect the performance of SpTRSV on multi-GPU architectures, despite linking via fast interconnect like NVLinks and NVSwitches. Alternatively, we employ the latest NVSHMEM technology based on Partitioned Global Address Space programming model to enable efficient fine-grained communication and drastic synchronization overhead reduction. Furthermore, to handle workload imbalance, we propose a malleable task-pool execution model which can further enhance the utilization of GPUs. By applying these techniques, our experiments on the NVIDIA multi-GPU supernode V100-DGX-1 and DGX-2 systems demonstrate that our design can achieve on average 3.53x (up to 9.86x) speedup on a DGX-1 system and 3.66x (up to 9.64x) speedup on a DGX-2 system with 4-GPUs over the Unified-Memory design. The comprehensive sensitivity and scalability studies also show that the proposed zero-copy SpTRSV is able to fully utilize the computing and communication resources of the multi-GPU system.
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
