#hpc — Public Fediverse posts

Last few places open for 2026 HPC/AI Leadership Training Institute at Barcelona Supercomputer Centre, Sept 7-10

https://www.bsc.es/news/events/hpc-leadership-institute-edition-2026/agenda

Everything (almost 😁) you need to know to effectively plan, buy & manage #HPC/#AI infra services. Plus networking with peers doing the same.

Looking for an enterprise Linux system for server, cloud, AI, and embedded systems? Learn what's new in OpenAtom openEuler 24.03 LTS SP4
https://www.linux-magazine.com/Online/Features/An-Enterprise-Distro-for-Server-Cloud-and-AI?utm_source=mlm
#openEuler #EnterpriseLinux #OpenAtom #server #cloud #AI #EmbeddedSystems #SuperPoD #HPC

Oof: https://pepy.tech/search?q=snakemake-executor-plugin-slurm

That would be > 500 k via #PyPI and > 100 k via #Conda (got the number with `condastats`)

That, of course, includes all downloads over all versions and including CI usage.

However, there are ~5,000 downloads per relevant version update for Conda alone. This fills me with pride and gives some ammunition for the "for admins" part of my #Snakemake #HPC tutorial. 😉

#SLURM

Whenever I learn about a new statistical method:

Do I know what I'm doing?
- Highly debatable

Do I hope for the best?
- 100%

It's not helping that I need to run things on an #HPC, so my failing forward is hogging resources.

I sincerely hope #Reviewer2 is not reading my toots, but at least the results look plausible.

#AcademicLife #PostdocLife #AcademicChatter #PhDLife #OverlyHonestMethods

#Linux #HPC #bash
I see why programmers get a shot of dopamine after reducing overall LoC. I just shaved off ~200 lines of code by reducing redundant calls in an epilog script.

🧪 Almost 20 years ago, I realized I didn’t just want to build supercomputers. I wanted to use them to solve problems that mattered. Great to talk with TheScientist about that journey and our work using HARVEY to build vascular digital twins. www.the-scientist.com/digital-twin...

#DigitalTwins #HPC

Digital Twins Modeling Blood F...

Free for participants from European countries: Multi-GPU Programming Bootcamp. 29 to 30 Sept 2026, online.

Scale #CUDA and #MPI codes to multiple #GPUs and nodes, use #Nsight Systems, #NCCL and #NVSHMEM, with hands-on access to an A100 cluster.

Info: events.asc.ac.at/event/358/

#HPC

General #HPC people question: what is your site policy for users on using coding agents on the cluster via VSCode or similar?

"It's their stuff, they can do whatever."? "Begone, foul shades!"? Something in between?

Today I was knee-deep in #Python dependency hell.

At some point I used my browser to access the #HPC, to emulate #KDE, to launch another browser with #Jupyter.

It still did not work.
Current solution is to try making a minimal bundle and run on #GoogleCollab.

Luckily it's only a tiny piece of the pipeline, that has to run interactively.

I can't wait to get back to my lovely #rstats

Observations on different social media platforms for #HPC / #supercomputing / #AIinfra engagements - numbers of views vs followers, subjective engagement rates, ...

https://lnkd.in/p/edSb626s

Everything is GREAT, now we just need a customer!

BitCoin miner "BitFarms" completes data center transitions to now take on AI and HPC (high performance computing) workloads.

The data center developer currently has no operating bitcoin business in the United States and no signed AI tenant - but is sitting on ~$819M of total liquidity. https://unchainedcrypto.com/keel-completes-shutdown-of-all-u-s-bitcoin-mining-as-revenue-halves-and-operating-loss-hits-141-million/ #BitCoin #Crypto #BitCoinMiner #BitFarms #Keel #DataCenters #AI #AIDataCenters #HPC #BlockChain

There we go. Another rack anointed in blood. May Khorne be appeased.

Guess I'm literally giving my blood for this job that isn't even mine anymore (well, soon)

#HPC

So günstig kann #Schnellladen DIREKT AN DER #AUTOBAHN sein – trotzdem fehlt etwas!

Kann Schnellladen wirklich schon ab 25 Cent pro kWh funktionieren?
Der Vispiron Ladepark in Schäftlarn zeigt, dass intelligente Ideen den Preis beim Laden von Elektroautos deutlich senken können. Das Konzept "Preis nach Sonnenstand" wurde weiterentwickelt: mehr #HPC-Ladepunkte, eine deutlich bessere Lage direkt an der Autobahn und jetzt sogar Stellplätze für Fahrzeuge...

#Elektrobohne

https://youtube.com/watch?v=tic-0sFOURI

In SuperMUC-NG 2 am LRZ arbeiten #GPU von Intel. In einem Bootcamp können Forschende der #lifescience #biologie Bioinformatik und Materialwissenschaften das System am 17. September kennenlernen:
• Umgang mit dem #HPC -System
• skalierbare #Workflows aufbauen,
• mit Forschenden vernetzen.
Bis zum 31. August läuft die Bewerbungsfrist für das Bootcamp. es ist für Masterstudierende, Doktorand:innen und Postdocs gemacht_ https://app1.edoobox.com/en/LRZ/Bootcamps%20and%20Hackathons/Workshop.ed.8d3ac6031182_10473187807.Life%20and%20Material%20Science%20Bootcamp%20on%20SuperMUC-NG%20Phase%202%20apply%20via%20e-mail

#ausbildung #studium #forschung #supercomputing

"Anyone can build a fast CPU. The trick is to build a fast system." - Seymour Cray, 1964

"Anyone can build a fast system. The trick is to build a usable, reliable, and affordable fast system!" - me

2026 #AI update - "and get it into production quickly" :-)

#hpc #supercomputer #supercomputing

I’m not telling you, I’m asking you (yes, you)!

1. What do you predict will be the biggest single #HPC or #supercomputing story at #SC26?

2. What will be the main theme or buzzword of the overall week?

Join us in Vienna for IWOMP 2026!

Join us October 7–9, 2026 for the 22nd annual IWOMP event in Austria. IWOMP is the annual workshop dedicated to the advancement of all aspects of parallel programming with OpenMP. It is the premier forum to present issues, trends, recent research ideas, and results related to parallel programming.

August 14th is the last day with the early-bird pricing -- full pricing starts on August 15

https://www.iwomp.org/registration/
#OpenMP #concurrence #parallelism #HPC

New blog post: https://blogs.fediscience.org/rupture-de-catenaire/2026/08/09/the-2nd-snakemake-hackathon/

My reflections on the latest #SnakemakeHackathon2026 . A little late, but nevertheless.

If you want to comment, please do so under the article. Only there comments will be persistent.

#Reproduciblecomputing #HPC #Hackthon

New #HPC Job! 👉️: Title: Head of IT Infrastructure & Scientific Computing (m/f/d)
Location: Vienna, Austria
Employer: CeMM Research Center for Molecular Medicine of the Austrian Academy of Sciences
Remote: Remote friendly
https://cemm.at/open-positions/0lbkdw77m8yp36xygqbyrocoual841c

Happy to host EuroMPI & IWOMP 2026 at TU Wien!

#HPC #MPI #OpenMP

@[email protected] @[email protected]
@[email protected]
@[email protected]
@[email protected]

RE: https://bsky.app/profile/did:plc:shms6dtucthbotaccudl6qun/post/3msixksmiu224

#HPC #Linux
Continuing on my previous post, using https://github.com/UoB-HPC/BabelStream allows one to quickly compile a single CUDA binary (there are other formats, too) to additional testing against a GPU. Again, this isn't meant for benchmark results for a whitepaper, but rather for systems people to perform health checks. My experience with NVIDIA's Grace offerings: GraceGrace, GraceHopper, and GraceBlackwell (NVL72 & NVL4) does warrant these checks. So, between the gpu-burn test for quickly allocating all of the GPU memory, and the cuda-stream, we have all the health checks we need to: (a) bring a node back into rotation, and (b) check the node during epilog cycles.

#HPC #Linux
For any systems people that manage GPU's, this software suite still builds without issue on Rocky 10.x with Grace Blackwells: https://github.com/wilicc/gpu-burn/blob/master/README.md

TL;DR - this CUDA software is a quick way to test NVIDIA GPU's with basic health checks. You can run on single GPU's or all at once. While it is not meant to be all encompassing, it's a great way to add a health check to the hardware.