The GPU-Cluster Oldenburg (GOLD) is a high-performance compute cluster dedicated for highly parallel and short latency computation. To realize optimal performance, GOLD is equipped with 30 state-of-the-art GPUs (14x NVIDIA Geforce Titan Black, 11x NVIDIA Geforce Titan X, 2x NVIDIA Geforce Titan X Pascal, 3x NVIDIA Tesla P100 16GB) with in total 89216 CUDA cores. These are complemented by 144 CPU cores (Intel Xeon). The currently eleven compute nodes of GOLD have each 128-256 GB on board DDR3/DDR4 memory and are Infiband connected.

The cluster was initiated by the Machine Learning group and is currently used, run, and funded by the following groups

• Machine Learning, Dept of Medical Physics and Acoustics (Prof. Dr. Jörg Lücke)
• Automatic Speech and Audio Processing, Dept of Medical Physics and Acoustics (Dr. Bernd T. Meyer)
• Computational Theoretical Physics, Institute of Physics (Prof. Dr. Alexander Hartmann)
• Acoustics, Dept of Medical Physics and Acoustics (Prof. Dr. Steven van de Par)

All groups gratefully acknowledge support of the GOLD cluster by the IT Services of the University of Oldenburg.