FatTreeSim

Usually, scientific applications outlive the lifespan of the High Performance Computing (HPC) systems for which they are initially developed. The innovations in the HPC systems’ hardware and parallel programming standards drive the modernization of HPC applications so that they continue being performant. While such code modernization efforts may not be challenging for HPC experts and well-funded research groups, many domain-experts and students may find it challenging to adapt their applications for the latest HPC systems due to lack of expertise, time, and funds. The challenges of such domain-experts and students can be mitigated by providing them high-level tools for code modernization and migration. A brief overview of two such high-level tools is presented in this chapter. These tools support the code modernization and migration efforts by assisting users in parallelizing their applications and porting them to HPC systems with high-bandwidth memory. The tools are named as: Interactive Parallelization Tool (IPT) and Interactive Code Adaptation Tool (ICAT). Such high-level tools not only improve the productivity of their users and the performance of the applications but they also improve the utilization of HPC resources.

A vital problem of energy management in a High Performance Computing (HPC) system comprised of remote clusters and data centers is investigated. The paper presents the results of research on energy-efficient management of workload in distributed data centers and computing clusters. The optimisation scheme for allocation of clusters to batches of computing tasks is presented and discussed. The aim is to minimize the energy used by the HPC system, assuming that the required level of quality of services is guaranteed. The results of the performance evaluation of the developed decision system through simulation are presented and discussed.

In high performance computing (HPC) systems, optical network links are often utilized for the HPC networks of these systems, but they have a relatively high rate of failure compared to their electrical counterparts. Due to the high link failure rate, evaluating the impact of these failures on HPC workloads is of particular interest. We extended the Merlin network module of the Structural Simulation Toolkit (SST) in order to evaluate the impact of link failures on applications running on HPC systems which use dragonfly network topologies.We focus on dragonfly topologies as they are frequently found in HPC systems, including NERSC Cori and Edison systems.We demonstrate our changes to SST by providing a sample of performance results and routing statistics for a dragonfly network of 8,192 nodes and three HPC workloads with 1% of optical link failures. For the three motifs under consideration, we show that the impact of link failure is largely dependent on the underlying workloads running on the system.

This paper introduces a new method to promote energy efficiency in computing systems. The rapidly growing demand for computational power in high-performance computing (HPC) systems is accompanied by a significant increase in energy consumption. This research investigates the many ways to minimise energy consumption in HPC systems without sacrificing computational performance. Building upon previous research we combined and refined ideas to develop an optimised approach. Our method utilises a tri-modular framework incorporating an Energy-Efficient Hardware Design, Resource Management, and Optimisation, and Server Virtualisation. The first module employs a method of designing smart energy-efficient hardware. This method incorporates leakage reduction techniques such as clock gating and sleep states. The second module uses a technique focused on optimising server software. This method is based on Dynamic Voltage and frequency scaling (DVFS) for power management in data centres. The third module is based on server virtualisation and employs software to create multiple virtual machines on a single physical server allowing for a significant reduction in energy and hardware costs. The methods used in these three modules are systematically integrated to produce a more efficient consumption of electricity. This will allow for the computing system to minimise energy consumption without compromising computational power.

With the growth of supercomputer's scale, the communication time during executing is increasing. This phenomenon arouses the architecture researchers' interests. In this paper, based on the fat-tree topology, which is widely used in Infiniband, we present an one-to-all broadcast communication time model. After classifying applications into two kinds, we establish the ideal model and the bandwidth-limited model on the exponential-capacity binary fat-trees for the two kinds of applications. Through analyzing the models, we get the curves which describe the relationship between the communication time and the processor number. The conclusions we get in this paper can help system designers make better system design.

The interconnection network is a key element in high-performance computing (HPC) and Datacenter (DC) systems, as it must support the communication among the endnodes, whose number constantly increases. Hence, guaranteeing a suitable network performance is crucial, as otherwise the network would become the entire system bottleneck. Network performance depends on several design issues: topology and routing, switch architecture, interconnect technology, etc. Among the available interconnect technologies, InfiniBand is a prominent one. InfiniBand components and control software allow to implement efficient topologies and routing algorithms, as well as queuing schemes that reduce the Head-of-Line (HoL) blocking effect derived from congestion situations. In this paper we present a new queuing scheme called Path2SL, that optimizes the use of the InfiniBand Virtual Lanes (VLs) to reduce HoL blocking in Fat-Tree network topologies. We have implemented PathSL in the control software of a real InfiniBand-based cluster. The experiment results obtained from real workloads run in this cluster show that Path2SL is a more efficient queuing scheme than others previously proposed to deal with HoL blocking in the analysed network configurations.

Vector databases have rapidly grown in popularity, enabling efficient similarity search over data such as text, images, and video. They now play a central role in modern AI workflows, aiding large language models by grounding model outputs in external literature through retrieval-augmented generation. Despite their importance, little is known about the performance characteristics of vector databases in high-performance computing (HPC) systems that drive large-scale science. This work presents an empirical study of distributed vector database performance on the Polaris supercomputer in the Argonne Leadership Computing Facility. We construct a realistic biological-text workload from BV-BRC and generate embeddings from the peS2o corpus using Qwen3-Embedding-4B. We select Qdrant to evaluate insertion, index construction, and query latency with up to 32 workers. Informed by practical lessons from our experience, this work takes a first step toward characterizing vector database performance on HPC platforms to guide future research and optimization.

As the size and complexity of high performance computing (HPC) systems grow in line with advancements in hardware and software technology, HPC systems increasingly suffer from performance variations due to shared resource contention as well as software- and hardware-related problems. Such performance variations can lead to failures and inefficiencies, which impact the cost and resilience of HPC systems. To minimize the impact of performance variations, one must quickly and accurately detect and diagnose the anomalies that cause the variations and take mitigating actions. However, it is difficult to identify anomalies based on the voluminous, high-dimensional, and noisy data collected by system monitoring infrastructures. This paper presents a novel machine learning based framework to automatically diagnose performance anomalies at runtime. Our framework leverages historical resource usage data to extract signatures of previously-observed anomalies. We first convert collected time series data into easy-to-compute statistical features. We then identify the features that are required to detect anomalies, and extract the signatures of these anomalies. At runtime, we use these signatures to diagnose anomalies with negligible overhead. We evaluate our framework using experiments on a real-world HPC supercomputer and demonstrate that our approach successfully identifies 98 percent of injected anomalies and consistently outperforms existing anomaly diagnosis techniques.

In addition to the performance, power consumption has become a major concern in High Performance Computing (HPC) systems. Typically the cooling system and the IT loads are the major contributors to the power bills. Understanding the power consumption at different granular levels in the HPC system is a first step to quantify the problems in HPC system. By proper monitoring and effective utilization of the cooling system, the power requirements of the HPC facility can be effectively met. In this paper we present a system where node level power measurement and WSN based rack level temperature measurement are used to provide localized control of cold air supply. A Smart Power Monitoring and Distribution Unit (PMDU) is designed and developed, to replace the existing Power Distribution Unit (PDU) in HPC. This can measure and report the power consumption, to support power profiling of large scale HPC system. This measured data are communicated to a base station via Ethernet. This base station collects all such measurements which can be used for power profiling of IT load of the HPC system. This helps to provide better insight into the power utilization pattern. Wireless Sensor Network (WSN) is used to collect exhaust and inlet air temperature of the server nodes and this information is used for directing the cold air effectively. A vent control system is designed and fabricated for intelligently directing the air flow to the server node inlet. It takes the node power and the temperature data as its inputs. This enables supplying/ redirecting more cold air towards under-cooled nodes without creating an extra load on the cooling system, thereby bringing in effective cooling at reduced power consumption. As an added advantage this could help in hot-spots mitigation.

Data centers play an important role in today's Internet development. Research to find scalable architecture and efficient routing algorithms for data center networks has gained popularity. The fat-tree architecture, which is essentially a folded version of a Clos network, has proved to be readily implementable and is scalable. In this paper, we investigate routing on a fat-tree network by deriving its global packing number and by presenting explicit algorithms for the construction of optimal, load-balanced routing solutions. Consider an optical network that employs wavelength division multiplexing in which every user node sets up a connection with every other user node. The global packing number is basically the number of wavelengths required by the network to support such a traffic load, under the restriction that each source-to-destination connection is assigned a wavelength that remains constant in the network. In mathematical terms, consider a bidirectional, simple graph, G and let N ⊆ V(G) be a set of nodes. A path system P of G with respect to N consists of |N|(|N| -1) directed paths, one path to connect each of the source-destination node pairs in N. The global packing number of a path system P, denoted by Φ(G, N, P), is the minimum integer k to guarantee the existence of a mapping φ : P → (1, 2, ..., k), such that φ(P) ≠ φ(P̅) if P and P̅ have common arc(s). The global packing number of (G, N), denoted by Φ(G, N), is defined to be the minimum Φ(G, N, P) among all possible path systems ?. In additional to wavelength division optical networks, this number also carries significance for networks employing time division multiple access. In this paper, we compute by explicit route construction the global packing number of (T <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">n</sub> , N), where T <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">n</sub> denotes the topology of the n-ary fat-tree network, and N is considered to be the set of all edge switches or the set of all supported hosts. We show that the constructed routes are load-balanced and require minimal link capacity at all network links.

Many data centers are built using a fat-tree network topology because of its high bisection bandwidth. Therefore thereis a need to develop analytical models for the energy behavior of fat-tree networks and examine strategies to reduce energy consumption. The most effective strategy is to power off entire switches, if possible. In this paper, we derive formulas for the minimum number of active switches needed in a fat-tree data center network for arbitrary types of loading. We also derive expressions for the expected traffic loss when these networks are overloaded with external (Internet) traffic. Results of detailed simulations conducted using well-known traffic models for data center networks [4] closely match our derived formulas. We show that a fat-tree network incurs significant energy cost even when very lightly loaded. In order to further reduce energy consumption, we consolidate traffic into fewer switches and derive expressions for energy cost versus load assuming traffic consolidation and show linear scaling. Finally, we observe that traffic patterns have a significant impact on energy consumption and this fact is evident in the analytical formulas.

Scientific workflows are increasingly transferring large amounts of data between high performance computing (HPC) systems. Even though these HPC systems are connected via high-speed dedicated networks and use dedicated data transfer nodes (DTNs), it is still difficult to predict the data transfer throughput because of variations in data transfer protocols, host configurations, performance of file systems, and overlapping workloads. In order to provide reliable performance prediction for better resource management and job scheduling, we need models for predicting data transfer throughput under real-world conditions. In this paper, we explore different machine learning approaches for building data-driven models to improve performance and prediction of large-scale data transfer throughput. In addition to the variables already collected by the network monitoring system, we also develop heuristics to derive additional metrics for improving the prediction accuracy. We use the prediction results to identify the importance of different network parameters in predicting the throughput for large-scale data transfers. Through extensive tests, we identify key network parameters, discover interesting variations among different HPC sites, and show that we can predict throughput with high accuracy. We also analyze our models and results to provide recommendations for improving the performance of big data transfers.

In recent years, High Performance Computing (HPC) systems have been shifting from expensive massively parallel custom architectures to clusters of commodity personal computers to take advantage of cost and performance benefits. To avoid having to restart an application in case of sudden failure, checkpointing/restart fault tolerance mechanisms are commonly implemented. One drawback to checkpointing/restart is that it creates an overhead which increases the execution time of an application. We present a theoretical analysis of our technique. The results show that the PLR checkpointing/restart can significantly improve the overall reliability of an HPC system.

Cloud computing offers new computing paradigms, capacity and flexible solutions to high performance computing (HPC) applications. For example, Hardware as a Service (HaaS) allows users to provide a large number of virtual machines (VMs) for computation-intensive applications using the HaaS model. Due to the large number of VMs and electronic components in HPC system in the cloud, any fault during the execution would result in re-running the applications, which will cost time, money and energy. In this paper we presented a proactive fault tolerance (FT) approach to HPC systems in the cloud to reduce the wall-clock execution time and dollar cost in the presence of faults. We also developed a generic FT algorithm for HPC systems in the cloud. Our algorithm does not rely on a spare node prior to prediction of a failure. We also developed a cost model for executing computation-intensive applications on HPC systems in the cloud. We analysed the dollar cost of provisioning spare nodes and checkpointing FT to assess the value of our approach. Our experimental results obtained from a real cloud execution environment show that the wall-clock execution time and cost of running computation-intensive applications in cloud can be reduced by as much as 30%. The frequency of checkpointing of computation-intensive applications can be reduced up to 50% with our FT approach for HPC in the cloud compared with current FT approaches.

Data centre networks are intended to meet the data transmission needs of the data centre’s highly networked hosts. The network architecture and routing technique can have a major impact on performance parameters such as latency. The fat-tree network is now one of the most popular data centre network architectures. In recent times, data centre network traffic has been steadily expanding. Because of the high volume of traffic, it is nearly impossible for a single server to satisfy all the client’s requests. In this paper, a bio-inspired Territorial Predator Scent Marking Algorithm (TPSMA) is proposed for load balancing in a fat-tree software-defined data centre. The bio-inspired algorithm was developed to increase network Quality of Service (QoS). The proposed technique is validated using the network emulator Mininet, with the Open Network Operating System ONOS controller. The performance study conducted showed that the throughput and network utilization were improved, hence improving the QoS.