Compute Cluster
The Department of Statistics at UW is dedicated to providing a computing environment that nurtures innovation and research excellence. Central to this commitment is our compute cluster, designed to support sophisticated statistical analyses and computational research essential for faculty and graduate students.
Streamlined Resource Management with Slurm
Our cluster harnesses the power of Slurm, an industry-standard, open-source workload manager. Slurm optimizes the efficiency of our computing resources by intelligently queuing jobs and managing workloads. This powerful system ensures that computational tasks, whether they are large-scale simulations or intricate data analyses, are executed as promptly as possible, maximizing productivity and research output.
Customized Computing with a Flexible Module System
To further enhance the research experience, we employ a dynamic module system that allows users to easily customize their computing environment. This system simplifies access to a wide array of applications and software versions, enabling researchers to select and load the specific tools they need to develop cutting-edge statistical models. The versatility of this system means that our researchers can always stay at the forefront of statistical software advancements, ensuring that their work remains on the cutting edge.
Empowering Statistical Discovery
Our computational infrastructure, equipped with Slurm and a flexible module system, is more than just a collection of high-performance hardware. It is a platform that empowers our statisticians to explore new theories, apply complex models, and contribute to the statistical community through innovative discoveries.
Our department's computer cluster is comprised of high-performance computing nodes that are designed to tackle large-scale computational tasks efficiently. The cluster features advanced Intel® Xeon® processors, which include the Skylake Gold 5220R and 6230R models as well as the Haswell E5-2680 v3, all renowned for their robust performance and reliability.
The nodes are equipped with an range of 24 to 52 CPU cores and memory capacities from 188 GB to 376 GB RAM per node, for a total of over 3300 CPU cores and 23TB of RAM, ensuring that complex data analyses and statistical simulations can be executed swiftly and effectively.
A full list of the nodes and technical specs is available here:
Compute Node Technical Information
Specialized Resources for GPU-Accelerated Computing
Understanding the growing need for GPU-accelerated computing, we have integrated specialized nodes that feature the powerful Intel® Xeon® Gold 5122 CPUs, complemented by NVIDIA Tesla GPUs. These resources are specifically optimized for tasks that require intense computational power, such as machine learning algorithms and large-scale simulations.
General Statistics Computing Resources
- High Performance Research Computing Resources:
- General compute cluster
- OS: Rocky Linux 9
- Slurm Workload Manager Queuing System
- 82 nodes
- 3352 Cores
- 4 NVIDIA Tesla V100 GPUs
- 23 TB of aggregate memory
- 25Gb/10Gb interconnect
- General compute cluster
- General Computing Resources:
- R Studio servers
- Provide access to R via a web browser
- http://rstudio.stat.washington.edu
- Windows Terminal Services
- Provide centralized remote access to Windows desktops and software
- ts1.stat.washington.edu, 24 GB Ram, 8 Cores @ 2.0 GHz (Intel L5335)
- ts2.stat.washington.edu, 16 GB Ram, 8 Cores
- Linux general purpose computing servers
- madrid.stat.washington.edu
- Web servers
- Provide departmental and individual web sites
- http://www.stat.washington.edu
- Printing services
- printer.stat.washington.edu
- R Studio servers