Memory usage can be broadly simplified into two values: Virtual memory (VMEM) which a program believes it has, and Physical memory (also known as "Resident Set Size" or RSS for short), which is the actual amount of memory it uses.
NVIDIA recently announced the GH200 Grace Hopper Superchip which is a combined CPU+GPU with high memory bandwidth, designed for AI workloads. These will also feature in the forthcoming Isambard AI National supercomputer. We were offered the chance to pick up a couple of these new servers for a very attractive launch price.
The CPU is a 72-core ARM-based Grace processor, which is connected to an H100 GPU via the NVIDIA chip-2-chip interconnect, which delivers 7x the bandwidth of PCIe Gen5, commonly found in our other GPU nodes. This effectively allows the GPU to seamlessly access the system memory. This datasheet contains further details.
Since this new chip offers a lot of potential for accelerating AI workloads, particularly for workloads requiring large amounts of GPU RAM or involving a lot of memory copying between the host and the GPU, we've been running a few tests to see how this compares with the alternatives.
Compression tools can significantly reduce the amount of disk space consumed by your data. In this article, we will look at the effectiveness of some compression tools on real-world data sets, make some recommendations, and perhaps persuade you that compression is worth the effort.
We have recently procured 120TB of NVMe based SSD storage from E8 Storage for the Apocrita HPC Cluster. The plan is to deploy this to replace our oldest and slowest provision of scratch storage. We have been performing extensive testing on this new storage as we expect it to offer new possibilities and advantages within the cluster.
Singularity (now Apptainer) is a container solution designed for HPC. Due to the secure and simple design, it can be easily used to provide applications for use with HPC clusters where other containers, such as Docker would not be suitable.