A NATIONAL FACILITY OPERATED BY NARIT
Chalawan runs CPU and GPU workloads through Slurm, on a shared Lustre file system and an InfiniBand interconnect. Researchers in Thailand apply for compute time through an open call for proposals.
01 · COMPUTE
chalawan_cpu19 Intel Xeon nodes with up to 28 cores each, for MPI codes and large simulations.
chalawan_gpu3 nodes, each with 4× NVIDIA V100 on NVLink, for deep learning and GPU-accelerated science.
a 329 TB Lustre file system at /lustre, mounted on every node, plus 32 TB of NFS at /data.
02 · PLATFORM
FDR InfiniBand at 56 Gbps keeps multi-node MPI jobs fast.
Environment modules for GCC, Intel compilers, Open MPI and MVAPICH2.
Slurm partitions and published queue policies give every project a fair turn.
The HPC team helps you port, scale and troubleshoot your code.