Overview
CIERA has exclusive access to three high-performance computer clusters (CIERA, Grail, and Trident). These clusters are used not only for running simulations and performing data analysis, but also serve as a training ground for students in high performance computing and computational research. In total, the three clusters provide access to 5996 compute cores.
CIERA Cluster
The CIERA cluster is available for any CIERA researcher for astronomy-related research. The cluster consists of 23 CPU nodes with dual 26-core CPUs (Intel Xeon Gold 6230 CPU at 2.10GHz) for a total of 52 cores per node. All compute nodes contain 192 GB of error-correcting DDR4 RAM.
The CIERA Cluster also includes four specialty nodes:
- A graphics processing unit (GPU) node containing an NVIDIA 40 GB Tesla A100 GPU. This node contains two 26-core Intel Xeon Gold 6230 CPUs running at 2.10 GHz.
- A GPU node containing two NVIDIA L40S GPUs. Each L40S GPU contains 48 GB of GDDR6 VRAM. This node contains two 64-core Intel Xeon Platinum 8592+ CPUs running at 1.9 GHz.
- A high-memory node with 1 TB of error correcting DDR4 RAM. This node contains two 26-core Intel Xeon Gold 6230 CPUs at 2.10 GHz.
- A high-memory node with 2 TB of error correcting DDR5 RAM. This node containts two 64-core Intel Xeon Platinum 8592+ CPUs running at 1.9 GHz.
The nodes are interconnected with InfiniBand HDR. The CIERA Cluster contains 120 TB of high-speed disk storage directly connected to the cluster in a RAID configuration.
All CIERA members can apply for 5 TB of individual scratch space on Northwestern’s Quest cluster.
Grail Cluster
The NSF-funded Grail cluster is available for use by CIERA researchers working on gravitational wave related research (NSF PHY-2406802). Grail provides 18 nodes for general-purpose high-performance computation. Each node contains dual 64-core CPUs (Intel Xeon Platinum 8592+ at 1.9 GHz), for a total of 128 cores per node. All compute nodes contain 512 GB of error correcting DDR5 RAM.
Grail contains two types of specialty nodes
- Three nodes containing four NVIDIA A30 GPUs, for a total of 12 NVIDIA A30 GPUs. Each A30 GPU contains 24 GB of HBM2 memory. Each of these nodes contain two 32-core Intel Xeon Gold 6338 CPUs running at 2.0 GHz.
- A GPU node containing four NVIDIA L40S GPUs. Each L40S GPU contains 48 GB of GDDR6 memory. This node contains two 32-core Intel Xeon Gold 6338 CPUs running at 2.0 GHz.
The nodes are interconnected with InfiniBand HDR. Grail includes 193 TB of high-speed disk storage directly connected to the cluster in a RAID configuration.
Trident Cluster
The Trident cluster is supported by the Gordon and Betty More Foundation (GBMF8477). The cluster is available for CIERA researchers working on stellar population synthesis. Trident provides 54 nodes for general-purpose high-performance computation. The cluster provides 54 CPU nodes with dual 26-core CPUs (Intel Xeon Gold 6230 CPU at 2.10 GHz), for a total of 52 cores per node. All compute nodes contain 192 GB of error correcting DDR4 RAM.
The nodes are interconnected with InfiniBand FDR. Trident includes 100 TB of high-speed disk storage directly connected to the cluster in a RAID configuration.
95Standard Compute Nodes
6924CPU Cores
217.4CPU Processing Power (TFLOPS)
6GPU Nodes
19GPUs
692.7GPU Processing Power (TFLOPS)
Quest
Northwestern’s Research Computing Services (RCS) group houses and maintains the CIERA, Grail and Trident clusters alongside its Quest High-Performance computer. Quest currently compromises 1009 compute nodes with 49,680 computing cores. Quest also has 55 GPU nodes and 28 high-memory nodes. All nodes currently run Red Hat Enterprise Linux 7.9. Quest is supported by UPS for emergency power. Additional HPC resources, including high-memory nodes, are available through Quest.
A number of CIERA research groups have private allocations on Quest.
Support
- A dedicated Computational Specialist is on staff to assist CIERA researchers. The Computational Specialist offers trainings and tutorials, walk-in consultations, and expert guidance for HPC code development and optimization, and more.
- The Computational Specialist holds workshops roughly twice a month during the academic year to support students and researchers. These workshops can be attended both in person and through Zoom, and are recorded.
- CIERA researchers looking for details on how to access and use the Grail and CIERA clusters should refer to the CIERA internal Guide page (Northwestern NetID required).
- Video tutorials from Northwestern IT Research Computing Services are available on YouTube (visit playlist).
Funding for the Grail computer cluster was provided by the National Science Foundation’s Major Research Instrumentation (MRI) program, grant PHY-1726951.
Additional Resources
Additional resources are available to certain research groups in CIERA. These include computing resources supported by the National Science Foundation and the U.S. National Laboratories.