The solution is organized into multiple compute clusters. For general compute and advanced training, a node with an Intel® Xeon® 8558U CPU and integrated with an NVIDIA® H100 NVL GPU to accelerate AI/ML/DL workloads and HPC simulation. For interactive 3D visualization, rendering, and hybrid AI/VFX workloads, an optimized cluster of rack‑mounted HP Z4 G5 workstations with Intel® Xeon® W‑series CPUs, 1.57 TB of aggregate RAM, and 18 NVIDIA RTX Ada GPUs (models 4500/4000/2000) totaling 380 GB of VRAM. The complex features high‑performance shared all‑flash storage based on the Synology FS6400, providing 168 TB raw for scratch and project repositories, with a 10 GbE interface. Low‑latency interconnect is provided by a 10/25/40 Gb L3 Ethernet fabric built on FS S5860‑48XMG switches, optimized for training traffic and graphics streaming.

Manufacturer: HP
Location: Pamplona