Acerca de

February 27, 2025

Add-on workshops will be available as part of the 18th annual Energy HPC Conference. Each workshop will take place at Rice University's BRC on Thursday, February 27, 2025 - they will occur simultaneously, so only one workshop can be chosen per registration. 

​

 

Best Practices in HPC Systems Management 

2nd Floor, Room 280 | 8:00 am - 4:00 pm

​​

Speakers: Practitioners and Experts from Industry, Academia, and National Labs

​

Organizer: ​Keith Gray (TotalEnergies)

​

Schedule:

• 8:00 - 8:30 am: Check-in + Breakfast 

• 8:30 - 8:35 am: Welcome + Introductions 

  • ​Keith Gray, TotalEnergies

• 8:35 - 9:35 am: TACC Update 

  • ​Tommy Minyard, TACC

• 9:35 - 10:40 am: Facilities Trends and Best Practices

  • Tommy Minyard, TACC

  • Kent Blancett, bp

  • Jeremy Singer, ExxonMobil

  • Bill Guyton, Shell

  • Donny Cooper, TotalEnergies

  • Wade Vinson, NVIDIA (Host)

• 10:40 - 11:00 am: Break 

• 11:00 - 12:00 am: Systems Management

  • Ashwini Rathnakara, Shell - ​Experiences with Singularity, Enroot, and Podman

  • Russell Jones, TotalEnergies - Migrating from xCAT to Warewulf

  • Jonathon Anderson, CIQ - Surveying the state of CentOS migration

  • In this systems-management focused session we will hear practical experiences from the field. First, Ashwini from Shell will present her experiences with a broad range of user-facing container solutions; then Russell from TotalEnergies will present his experience migrating their backend cluster management system from xCAT to Warewulf. Jonathon will end the session by leading a discussion about the current state of migration away from EOL versions of CentOS and, if time allows, may provide an overview of what to expect in the soon-to-be-released Warewulf v4.6.

• 12:00 - 1:00 pm: Lunch 

• 1:00 - 2:00 pm: RoCE in the Blue corner!

  • ​Shawn Hall, Jump Trading

  • Donny Cooper, TotalEnergies

  • ​We will discuss advanced networking technologies, such as RoCE (RDMA over Converged Ethernet).  We will share some initial results of benchmarking TCP, RoCE and InfiniBand to see the progress of RoCE technology, including and application scaling study of a CFD application, which is used for high-fidelity wind turbine wake modeling.

• 2:00 - 2:20 pm: Memory Technologies Update

  • Eric L Pope, HPE 

  • PJ Waskiewicz, Jump Trading 

• 2:20 - 3:00 pm: Towards an Architecture for Cloud-Native HPC

  • Alex Loddoch, Chevron

  • Alex Morris, Microsoft

  • Many HPC workloads common in the Energy industry (and beyond) have characteristics that would make them a good fit for cloud computing. However, in order to realize the benefits of elasticity and scale that cloud computing offers, we need to use a different set of tools for orchestrating and running these workloads. Queuing systems, parallel POSIX filesystems and MPI do work in a cloud environment but they are not an ideal fit, are not natively supported or even inhibit effective cloud deployments. Many POCs, demonstrations and projects have been conducted that demonstrate “cloud native” approaches. Most of these however seem to be use-case specific one-offs that need to be rebuilt for each application or cloud environment, limiting progress for the overall community. This Birds-of-feather intends to bring together users, developers and (cloud)technology providers to describe and ultimately develop a common and widely applicable framework that makes deploying and running pleasingly parallel HPC workloads in different cloud environments possible and indeed pleasing.​

• 3:00 - 3:15 am: Break 

• 3:15 - 4:00 pm: Queueing System Migration Experiences

  • ​Kent Blancett, bp

  • Tim Osborne, ORNL

• 4:00 pm: Close

​​​​

​​​​​​​