HGPU groupMixed-precision numerics in scientific applications: survey and perspectives
hpc.social admins
- 188 Followers
- 60 Following
- 683 Posts
Mastodon administration for mast.hpc.social. See more information at https://hpc.social and help sponsor this instance and related projects at https://github.com/sponsors/hpc-social?frequency=one-time
AmartyaMy work for the month is to optimise the matrix adressing scheme of OpenFOAM to reduce cache miss. Initial idea is to replace the LDU Matrix addressing scheme with Diagnol matrix addressing scheme using multiple arrays for structures meshes. Will check it out on a simple 2D Poisson equation using Gauss-Seidel solver to check performance benefits, and will eventually proceed to introduce a new solver as a plugin if benefits are noticeable.
Wish me luck.
vsochWe are excited to announce support for Flux for #Kubeflow v2.2 to enable AI/ML workloads paired with #HPC simulation in #Kubernetes! 🥳
https://bsky.app/profile/vsoch.bsky.social/post/3mhssklh5xk2q
See the full post above to learn more, or jump into the demo! https://youtu.be/NbP0NdSDwog?si=DLHkdtYVnWa5lobg
v (@vsoch.bsky.social)
We are excited to announce support for Flux for #Kubeflow v2.2 to enable AI/ML workloads paired with #HPC simulation. Flux adds a ZeroMQ bootstrap, support for #PMIx, more flavors of #MPI, and bypasses potential etcd and kube-sched bottlenecks. We are excited to bring this to the larger community! 🥳 [contains quote post or other embedded content]
Benjamin Carr, Ph.D. 👨🏻💻🧬Price of #Nvidia's #VeraRubin #NVL72 racks skyrockets to $8.8M apiece, but #server makers' margins will be tight — Nvidia is moving closer to shipping entire full-scale systems
#Blackwell #NVL72 #rackscale system costs $2.8 – $3.4M for an #AI training and #HPC NVL72 #GB200 and $6M to $6.5M for an AI inference NVL72 #GB300
Vera Rubin NVL72 #VR200 systems are currently quoted at $5M - $7M per unit.
Nvidia has never confirmed the list prices of its NVL72 or #NVL144 products.
https://www.tomshardware.com/tech-industry/artificial-intelligence/price-of-nvidias-vera-rubin-nvl72-racks-skyrockets-to-as-much-as-usd8-8-million-apiece-but-server-makers-margins-will-be-tight-nvidia-is-moving-closer-to-shipping-entire-full-scale-systems
#Blackwell #NVL72 #rackscale system costs $2.8 – $3.4M for an #AI training and #HPC NVL72 #GB200 and $6M to $6.5M for an AI inference NVL72 #GB300
Vera Rubin NVL72 #VR200 systems are currently quoted at $5M - $7M per unit.
Nvidia has never confirmed the list prices of its NVL72 or #NVL144 products.
https://www.tomshardware.com/tech-industry/artificial-intelligence/price-of-nvidias-vera-rubin-nvl72-racks-skyrockets-to-as-much-as-usd8-8-million-apiece-but-server-makers-margins-will-be-tight-nvidia-is-moving-closer-to-shipping-entire-full-scale-systems
ADMIN magazineThe EuroHPC Joint Undertaking has launched the HPCTRAIN project, a program to help trainees gain practical experience with HPC systems and technologies
https://www.admin-magazine.com/News/EuroHPC-JU-Establishes-HPC-Training-Program?utm_source=mam
#HPC #EuroHPCJU #training #HPCTRAIN #supercomputing
https://www.admin-magazine.com/News/EuroHPC-JU-Establishes-HPC-Training-Program?utm_source=mam
#HPC #EuroHPCJU #training #HPCTRAIN #supercomputing
Kenneth Hoste
Lustre Users Group 2026
April 27 – 29, 2026
Indianapolis, IN
Make plans to be with us in Indianapolis, IN for the conference for all things related to Lustre shared parallel storage. Opening Reception the evening of April 27th. The conference presentations are April 28th through April 29th, 2026 to learn about new features/improvements in Lustre including the most-recent release, Lustre 2.17 with Hybrid IO, Dynamic LNet NID Configuration and Nodemap enhancements.
OpenMP ARBDeadline extended until March 16th!
- Final weekend to submit your feedback -
Can you spare a minute for a very short survey? If you have ever used our OpenMP API Examples book, we are asking for your feedback in how we can improve it. The survey is short and quick.
Survey: https://link.openmp.org/4
(We will *not* add you to our contact list or sell your information)
#openmp #parallel #programming #HPC
Johannes KösterToday, the #SnakemakeHackathon2026 at the TU Munich ended with the release of Snakemake 9.17! I want to thank all participants and my co-organizers! You have been incredibly dedicated and we improved a ton of things throughout the ecosystem. While 9.17 alone is already an impressive release, more releases will come in the next days and weeks, as the remaining pull requests that have been started during the hackathon are finalized. https://snakemake.github.io