...
About HPC Cluster |
About HPC Cluster
The new HPC cluster at C2B2 is now called "Ganesha.", This HPC cluster s a Linux-based (Rocky9.4) compute cluster consisting of 62 Dell Server, 2 head nodes, and a virtualized pool of login (submit) nodes, 8 Weka storage nodes. The nodes fit in a dense configuration in 9 high-density racks and are cooled by dedicated rack refrigeration systems., is designed with the goals of running compute intensive AI workloads.
The clusters comprise:
20 compute nodes, each with 192 core processors and 768 GB of memory.
2 nodes with 192 cores and 1.5 TB of memory.
40 GPU node featuring 2 NVIDIA L40s GPU cards 192 cores processors and 768 GB memory1 GPU node with a .
One NVIDIA Superchip GH200, with 72-core ARM architectureCPU, 1 H100 GPU. Due to tight design, and 570 GB of memory
...
a very high bandwidth between CPU and GPU allows GPU to use of 480 GB CPU memory, apart from GPU’s on-chip 96 GB memory. Although CPU memory is not as fast as GPU’s, a decent 900 GB/s memory bandwidth allows the Large Language Model (LLM) application to access all the 576 GB memory effectively.
The primary network for all the compute nodes and Weka storage is HDR100, the 100 Gbps low-latency Infiniband fabric. Since each node has 192 CPU cores, the need to go across nodes over network is greatly reduced. If a large MPI application still needs to use multiple nodes, the low-latency Infiniband fabric greatly removes network bottlenecks. Each node also has a 25 Gbps IP network for applications that must use IP network. Additionally, a set of login nodes running on Proxmox virtualization provide a pool of virtual login nodes for user access to this and other systems. Like our previous clusters, this cluster the system is controlled by SLURM. The login nodes are the primary gateways to the HPC cluster. They are not expected to run heavy duty interactive work, for which users can always start an interactive shell session on a compute node that has more compute resources than login nodes. See below.
Storage for the cluster is provided exclusively by our Weka parallel filesystem with over 1 PB of total capacity.For assistance with cluster-related issues, please email dsbit-of a large bank of all-NVMe very fast drives. WekaFS further boosts performance by distributing the load to 8 servers in parallel. Applications can use GPU-Direct RDMA technology, which bypasses Kernel-based network stack to access data from Weka directly via Mellanox ConnectX-6 NICs. See https://docs.nvidia.com/cuda/gpudirect-rdma/ for more details about this technology.
If you're experiencing issues with the cluster, please reach out to dsbit_help@cumc.columbia.edu, including for support. To facilitate a quick and precise response, be sure to include the following details in your messageemail:
Your Columbia University Network ID (UNI)
Job ID numbers , (if your inquiry pertains issue is related to a specific job issue
This information will help ensure a prompt and accurate response to your cluster-related questions.
Getting Started
Job Examples
Research Products
Available software
Storage
Submitting Jobs
Technical Information
...
)
| Info |
|---|
This HPC cluster exclusively accepts MC credentials for authentication. However, to access the cluster, you also need an active HPC account with C2B2. If you don't have an account, please reach out to dsbit_help@cumc.columbia.edu to request one. |
Getting Access
In order to get access to this HPC cluster, every research group needs to establish a PI Account using an MoU-SLA agreement that can be downloaded DSBIT-MOU-SLA.pdf This document provides further details about modalities, rights & responsibilities, and charges etc.
Logging In
You will need to use SSH in order to access the cluster. Windows users can use PuTTY or Cygwin or MobaXterm. MacOS users can use the built-in Terminal application.
Users log in to the cluster's login node at hpc.c2b2.columbia.edu using MC credentials
$ ssh <UNI>@hpc.c2b2.columbia.edu |
Interactive login to Compute Node
All users will access the HPC resources via a login node. These nodes are meant for basic tasks like editing files or creating new directories, but not for heavy workloads. If you need to perform certain heavy duty tasks in an interactive mode, you must open an interactive shell session on a compute node using SLURM’s srun command, like an example below. See the SLURM User Guide link on the side navigation to learn more about SLURM.
srun --pty -t 1:00:00 /bin/bash |
Interactive login on GPU node
srun -p gpu --gres=gpu:L40S:1 --mem=8G --pty /bin/bash |
Interactive login on GPU node With memory and time limit
srun -n 1 --time=01:00:00 -p gpu --gres=gpu:L40S:1 --mem=10G --pty /bin/bash |