Deep Bayou, the latest addtion to LSU HPC's pool of resources, is a 13-node GPU cluster funded by the NSF CC* award #2020446. Each Deep Bayou compute node is equipped with at least two NVIDIA V100S GPUs and a 1.5 TB NVMe solid state drive. Two of the compute nodes are equipped with four NVIDIA V100S GPUs connected by NVLink, which boosts the data exchange performance between GPUs. These features make Deep Bayou a powerful platform for deep learning and AI workloads.

The compute time on Deep Bayou is allocated separately from Super Mike 2 and SuperMIC. Given its limited number of nodes and hardware features, when requesting an allocation on Deep Bayou, PIs should describe how their intended workloads will use the GPU devices.

Deep Bayou went into general production in March 2021 and is available to all LSU HPC users.

• Common Features
  • RedHat Enterprise Linux 7 Operating System
  • 100 Gb/sec (HDR100) InfiniBand
• 11 GPU Compute Nodes, each with:
  • Two 24-core Intel Cascade Lake (Intel® Xeon® Gold 6248R Processor) CPUs.
  • 192 GB memory
  • Two NVIDIA V100S GPUs (each with 32 GB memory)
  • 1.5TB NVMe SSD drive
• 2 GPU NVLink Compute Nodes, each with:
  • Two 24-core Intel Cascade Lake (Intel® Xeon® Gold 6240R Processor) CPUs.
  • 384 GB memory
  • Four NVIDIA V100S GPUs (each with 32 GB memory, SXM2 with NVLINK)
  • 1.5TB NVMe SSD drive
• 1 Login Nodes, with:
  • Two 24-core Intel Cascade Lake (Intel® Xeon® Gold 6226R Processor) CPUs.
  • 192 GB Ram
• Cluster Storage
  • 840 TB Lustre file system (shared with the SuperMIC cluster)

Last modified: March 17 2021 13:19:32.