Close up of a modern GPU.

Supercomputer Saga receives major GPU compute capacity boost

We are pleased to announce that Saga, on of the national supercomputers, is about to receive a significant boost in its GPU compute capacity. Scheduled to be completed by the end of September, this upgrade is positioned to bring in a whole new level of computing power and exciting possibilities for researchers in various fields. 

The upcoming enhancement will introduce eight cutting-edge GPU nodes, each featuring an AMD EPYC 7542 32-core CPU, an astounding 1024 GB of RAM, 9.6 TB of flash storage, and 4 NVIDIA A100 GPUs equipped with a remarkable 80 GB of memory. These nodes will seamlessly replace some of the existing GPU nodes on Saga, augmenting its processing capabilities to a greater level. The NVIDIA A100 GPUs are at the forefront of GPU technology, promising not only enhanced computational power but also improved efficiency and flexibility.
Currently, Saga relies on 32 NVIDIA P100 GPUs distributed across eight GPU nodes to meet researchers' computational demands. However, with the upcoming expansion, a select number of P100 GPUs will be retained within the system and serve as a development platform for users, allowing them to explore and innovate with this older generation of GPUs.

Researchers using Saga can thus prepare to gain access to state-of-the-art hardware capable of handling demanding tasks in fields such as machine learning, artificial intelligence, climate modelling, and drug discovery. This substantial increase in GPU compute capacity on Saga signifies a significant step toward continuously advancing scientific and technological research in Norway.