ThinkParQ: BeeGFS v7.3.0 Parallel File System With Nvidia Magnum IO GPUDirect Storage Support

ThinkParQ GmbH, the company behind the parallel file system BeeGFS, announced BeeGFS v7.3.0.

BeeGFS v7.3.0 includes Nvidia’s Magnum IO GPUDirect Storage (GDS) and support for client side multirail RDMA networking, supporting Arm architectures, along with Linux 5.10 support.

As more and more HPC driven workloads are moving towards GPU-accelerated architectures, the support of GDS will increase the ability and performance of BeeGFS. Multirail network configurations will allow the BeeGFS client to be configured to use multiple RDMA capable network interfaces on the same network, whilst automatically balancing the traffic across all those interfaces.

 “Our engineering team has been working closely with NetApp’s team on many development projects to further the innovation of BeeGFS, and together we collaborated with  Nvidia to certify BeeGFS with Nvidia DGX A100 systems for machine learning and deep learning,” said Frank Herold, CEO. “We’re excited that GDS now supports BeeGFS, and look forward to performance results from our customers and the BeeGFS community.”

Along with GDS support, BeeGFS v7.3.0 also includes the advancement of Arm architecture support. With Arm architecture advancing from traditional mobile processors and becoming more prominent in the server market, ThinkParQ saw more demand and request for Arm support from the field and decided that BeeGFS will support Arm in v7.3.0 onwards.

The long-awaited Linux 5.10 kernel support has also been implemented, laying the groundwork to support more modern I/O frameworks in the kernel.

“Whilst Linux 5.10 has been out for some time and other filesystems are also challenged with supporting it and newer kernels due to the underlying changes, we have now completely reworked our I/O path in the client module and are able to support Linux 5.10 and future kernel releases,” said Philipp Falk, head, engineering.

Ressource:
Blog: BeeGFS now supports NVIDIA Magnum IO GPUDirect Storage. So, what’s the big deal?