AMD-powered LUMI supercomputer enables TurkuNLP to rapidly develop novel AI models

TurkuNLP, the natural language processing (NLP) research group within the Department of Computer Science at the University of Turku in Finland, has announced that it has access to the world’s first AMD-powered high performance computer, called the LUMI. The new computer enables the research group to expedite the development and optimization of novel AI models.

The LUMI supercomputer is powered by AMD EPYC processors, with each node equipped with two AMD EPYC 7742 64-core processors and an NVIDIA RTX 6000 GPU. The processor cores offer a combined throughput of more than 800GFLOP/s, providing ten times more performance than the existing cluster used by the research group. The NVIDIA GPUs are also designed to accelerate the development of artificial intelligence researchers, providing more efficient training and inferencing at a faster rate.

Funded by the European High Performance Computing Joint Undertaking (“Eur oHPC”), the LUMI supercomputer is owned jointly by CSC – IT Center for Science in Finland, the Institute for Advanced Simulation at the Forschungszentrum Jülich in Germany and the Barcelona Supercomputing Center in Spain.

The LUMI supercomputer will greatly benefit the research at TurkuNLP by expediting the development of novel AI models. In particular, it will enable researchers to experiment quickly and efficiently with complex deep learning and machine translation models, which could lead to breakthroughs in natural language processing.

Furthermore, the researchers at TurkuNLP will benefit from the newest technologies provided by AMD, including the PCIe 4.0 standard, which offers higher throughput for memory, storage and networking systems. This will enable researchers to increase the speed at which data can be processed while improving performance on memory-intensive workloads.

By providing them with the world’s first AMD-powered HPC system, the LUMI supercomputer will give TurkuNLP researchers a powerful platform for accelerating research in natural language processing-related AI models. The new system will be instrumental in helping advance the research group’s mission to develop innovative AI models for natural language processing applications.