NSM’s PARAM Shakti
(Currently in Testing Phase)
Welcome to NSM PARAM RUDRA Supercomputing Facility PARAM Shakti – 3.1PF Cluster @ IITM
In 2020, under the National Supercomputing Mission (NSM), IIT Madras was selected for the installation of a 3 Petaflop supercomputing facility. With support from C-DAC Pune, the installation will be completed during 2024–2025.
The PARAM RUDRA Supercomputing Facility at IIT Madras, named as PARAM SHAKTI, marks a significant breakthrough in computing power and infrastructure.
PARAM SHAKTI Overview
PARAM Shakti is a high-performance supercomputer developed under the National Supercomputing Mission (NSM). It is built on PARAM RUDRA servers, a powerful computing platform designed to handle complex calculations and simulations at exceptional speed and scale.
The supercomputer features a heterogeneous and hybrid architecture, combining Intel Xeon 2nd Gen Cascade Lake processors with NVIDIA Ampere A100 GPU cards. The system was designed and implemented by the HPC Technologies Group at the Centre for Development of Advanced Computing (C-DAC).Key features include:
- Total number of Nodes: 622 (20 + 602)
- 10 Login Nodes and 10 Management Nodes
- 476 CPU only Nodes and 96 high-memory CPU only Nodes (768 GB each)
- 30 GPU Nodes with Tesla A100 GPUs,
- 4 PiB of storage with 40 GB/s bandwidth
- 400 TiB of high-speed storage at 100 GB/s
- achieving a peak performance of 3.1 PFLOPS
For more information about the system configuration, how to obtain an account, and how to access the Aqua Cluster, please refer to the links below:
Please click the links to learn more about each topic.
| Title/Link | Description |
|---|---|
| Introduction | A Brief Overview of the Supercomputing Facility |
| System Architecture and Configuration | Detailed Information About the PARAM SHAKTI Supercomputer |
| Getting User Accounts -IITM Users | How to Get a User Account-IITM Users |
| First things First | New User Information |
| How to access the Cluster | Information on Logging In and Login Methods |
| Transferring files | How to Transfer Files Between Your Desktop and the Cluster |
| Resource Management | Quick Guide About SLURM and Sample Job Submission Scripts |
| Loading packages through SPACK | How to Load Applications and Packages Using Spack |
| Preparing Your Own Executable | How to Load Compilers and Libraries During Program Compilation |
| Debugging Your Codes | Methods of Debugging |
| ML /DL Application Development | Overview of Available Python (Conda) ML/DL Libraries and Modules |
| Some Important Facts | About the Lustre File Sizes |
| Best Practices for HPC | Getting started Information for New Users |
| Installed Applications | Applications and Packages Available on the Cluster |
| Acknowledging the National Supercomputing Mission in Publications | How to Clite the Cluster in your Work |
| Getting Help – PARAM Rudra Support | How to Get Assistance for Issues You Encounter on the Cluster |
| User Creation Process | How to create a new User Account |
| Closing Your Account on PARAM Rudra | How to Close your User Account |
| References | Links to the reference sites |