About FRIDA

FRIDA is a Slurm cluster that was initially set up with the help of the Development of Slovene in the Digital Environment (RSDO) project in 2020 when the first NVIDIA DGX-A100 system was acquired. Later it was upgraded with a new and more capable login node, and two compute nodes with 8x 80GB/GPU GPUs each, one of them being an NVIDIA DGX-H100.

FRIDA is progressively further expanded. For example in Q2 of 2024, attention was given to future/alternative technologies, with the addition of development kit compute nodes equipped with AMD MI210 GPUs as well as NVIDIA GraceHopper Superchips. Later expansions concentrated on faster InfiniBand data interconnect, a fast and larger shared data storage based on the Weka Data Platform. The most recent additions as of Q3 2026 are a faster Ethernet fabric, four NVIDIA DGX-B200, and four NVIDIA DGX-B300. These systems are the current top-tier NVL8 systems by NVIDIA. Each NVIDIA DGX-B200 provides 8x 180GB/GPU Blackwell GPUs, and each DGX-B300 comes equipped with 8x 288GB/GPU Blackwell Ultra GPUs. In both cases the GPUs are interconnected via the 1.8TB/s 5th gen NVLink+NVSwitch. The compute fabric for the DGX-B200 nodes is based on 400G InfiniBand for a total node-to-node bandwidth of 400 GB/s, wheres the compute fabric for the DGX-B300 nodes is based on dual-rail 400 GbE for a total node-to-node bandwidth of 800 GB/s with fail-over capabilities. In both cases the storage fabric is based on 400GbE.

Research labs that own and manage their computing systems themselves, may inquire about the possibility of integrating their infrastructure into FRIDA. Cofunding of future FRIDA expansions is also possible. All inquiries should be addressed to the UL FRI Management Board, the technical details will be coordinated by the FRIDA technical committee.