The Param Vikram-1000 Super Computer was established as a cutting-edge High-Performance Computing (HPC) Cluster. This remarkable computing facility caters to the advanced computational needs of the scientific and technical fraternity at PRL.
The Param Vikram-1000 boasts an impressive configuration, offering a theoretical peak performance (Rpeak) of 1395.63 TeraFlop/s (TF) and a Maximal LINPACK performance (Rmax) of 956.34 TF. It consists of a centralized infrastructure comprising 108 computing nodes, which collectively house 7296 CPU cores and 276,480 GPU cores. Additionally, the system is equipped with 74TB of RAM and a high-performance Lustre parallel filesystem with a capacity of 1 PB.
To support image analysis and facilitate the execution of scientific tools such as Mathematica, a dedicated Visualization Node is available. This specialized node features two 32-core Intel Icelake CPUs operating at a clock speed of 2.60 GHz, along with two Nvidia A40 GPU Cards.
The implementation of the Param Vikram-1000 has leveraged open-source tools and libraries including Lustre, OpenHPC, Slurm, Ganglia, and Intel oneAPI, ensuring robust and efficient computing.