Resources

Palmetto 2

Clemson University’s HPC environment is called Palmetto 2. Palmetto 2 was released in May of 2024 and has over 1,200 compute nodes of various configurations, totaling over 54,000 CPU cores more than 1,500 GPU accelerators. The latest additions to the cluster include: 98 nodes with 192 cores of AMD EPYC Genoa for a total of 18,816 cores, new NDR InfiniBand networking infrastructure, and specialized AI/ML nodes with 8-way NVLINK H100 or H200 GPUs for a total of 80 Hopper GPUs. Nodes are interconnected with an Infiniband network fabric with 56, 100, 200 and 400 Gbit connections, with an additional 10 or 25 Gbit ethernet connection. Palmetto 2 is a Top500 system, ranked #393 in the June 2025 list with a performance of 3.01PF.

Indigo Data Lake

Clemson also operates the Indigo Data lake, a high-performance, all-flash filesystem by VAST Data. Indigo has an effective capacity of 5.5 PB with data deduplication and data reduction. This filesystem is connected to the Palmetto 2 HPC System with 16x HDR100 Infiniband network connections with a total of 1.6Tbit of aggregate network capacity. Connectivity to this filesystem is also provided directly to campus users via SMB network shares and has 400Gbit of aggregate Ethernet network capacity.

Secretariat (CU Institute for Human Genetics HPC Cluster)

Secretariat is the CU Institute for Human Genetics' high-performance computing environment at the Greenwood campus. The cluster comprises 33 compute nodes across four categories—General Purpose, Big Memory, Interactive, and GPU—providing a combined total of 2,168 CPU cores and 30.1 TB of memory. Four GPU nodes support advanced AI and deep learning workloads, including one NVIDIA DGX system with 8× A100 80 GB GPUs, two R760XA nodes each with a H100 NVL 188 GB (equivalent to 2× H100s), and one R760 node with 2× L40S 48 GB GPUs. Secretariat delivers a combined theoretical peak performance of approximately 343 TFLOPS (FP64) across its CPU and GPU architectures. Data storage is provided through three DELL EMC PowerVault 40/5084 arrays, offering a combined 1.25 PB of RAID6 (double-parity) storage served via dedicated NFS hosts. High-speed I/O is supported by three scratch nodes providing 214 TB of solid-state storage, while secure real-time offsite replication and long-term archival through magnetic tape and AWS Deep Glacier ensure data integrity and persistence. Networking is powered by Mellanox InfiniBand (100–400 Gb/s) with a 1 Gb Ethernet backup fabric.

Learn more about Palmetto 2 and HPC at Clemson >>

Hyperion Phase III

Hyperion Phase III is the University of South Carolina’s flagship HPC environment. Intended for large parallel jobs, Hyperion has 356 compute, GPU, and 2TB Big Memory nodes, providing 16,616 CPU cores and 1.5 PFlops peak performance, all with EDR Infiniband (100 Gb/s) interconnects and 1.4 PB of GPFS scratch storage.

Theia

Theia is a new NSF funded AI-focused HPC environment with 28 112-core CPU nodes, 10 quad A100/H100 GPU nodes for a total of 3808 Cores and 675 TFLOPS combined CPU and GPU peak performance all with 1.5 PB GPFS scratch storage.

Learn more about USC’s HPC resources >>

HPC Resources

SRNL employs high-performance computing, advanced statistics, artificial intelligence, and traditional experiments to advance multidisciplinary applied sciences supporting critical DOE and NNSA missions. This includes advancements in nuclear materials, tritium, and isotope sciences vital for weapons production and global security. Our work directly enhances environmental remediation, waste management, and critical infrastructure resilience, ultimately strengthening national security.

The SRNL HPC cluster has a heterogenous compute architecture outfitted with powerful compute nodes to support cutting edge scientific research. It presently contains around 2000 processor cores and 12 TB of RAM, backed by 800 TB of storage. Specialized GPU nodes offer just over 7 petaflops of tensor performance for LLM (Large Language Model) and GPU-accelerated scientific workloads.

Learn more about SRNL’s HPC resources >>

Cyberinfrastructure Resources

Coastal Carolina University’s cyberinfrastructure supports a broad range of research applications, from computational modeling to data-intensive analysis. Across its three main partitions, the system comprises 17 nodes providing a total of 524 physical CPU cores, 6.5 TB of RAM, and 11 GPUs with a combined 252 GB of VRAM.

Waccamaw Partition: Four nodes with Intel Xeon Gold 6330 CPUs (two sockets per node, 28 cores per socket) and 256 GB of RAM per node, totaling 1,024 GB across the partition. Each node includes one NVIDIA A30 GPU with 24 GB of memory.

Peedee Partition (Large-Memory Nodes): Six nodes, each featuring dual Intel Xeon 3.00 GHz CPUs with 32 physical cores per node and 768 GB of ECC-Registered RAM per node. Altogether, the partition offers 192 physical CPU cores and 4,608 GB of RAM, providing a high-memory environment well-suited for data-intensive workloads.

Santee Partition: Seven nodes, each with dual Intel Xeon 3.30 GHz CPUs and 24 physical cores per node, along with 128 GB of ECC-Registered RAM per node. The partition provides 168 physical CPU cores and 896 GB of RAM. Each node includes an NVIDIA Tesla K20 GPU for general-purpose GPU (GPGPU) computing.

Learn more about Coastal Carolina's HPC resources >>