AIRR compute opportunity: AI for Science | Bristol Centre for Supercomputing (BriCS) Dreaming of more time on Isambard-AI? Eligible projects can now apply for between 250,000 and 1,000,000 GPU hours. This research opportunity seeks projects aligned with AI for Science priority areas...

You can now apply for up to a million GPU hours on Isambard-AI thanks to a new access route from DSIT and UKRI. Apply now! #AI #HPC
www.linkedin.com/posts/brics-...

Isambard-AI: A New National Treasure YouTube video by BriCS

📽️ Isambard-AI. The UK's fastest supercomputer. Built for breakthroughs, powered by people.

WATCH: youtu.be/HjwH9DdIjRs?...

#supercomputing #ai #bristol

GitHub - ecmwf-ifs/openifs: ECMWF OpenIFS model ECMWF OpenIFS model. Contribute to ecmwf-ifs/openifs development by creating an account on GitHub.

ECMWF's OpenIFS weather forecasting model is now fully open source on GitHub.

OpenIFS is a version of the same system used to produce our operational forecasts. After 15 years of licensed access, it's now genuinely open - publicly available with an Apache 2 licence.

github.com/ecmwf-ifs/op...

Benchmarking is core to High-Performance Computing, assessing hardware performance and software capabilities. Here we present a list of over 100 HPC benchmarks and benchmark suites, introduced in the accompanying paper.

"Benchmarking is core to High-Performance Computing...
Here we present a list of over 100 HPC benchmarks and benchmark suites, introduced in the accompanying paper."

- fzj-jsc.github.io/benchmark-su...
- github.com/FZJ-JSC/benc...
- arxiv.org/abs/2509.08347

The Puzzle of Low Data Center Utilization Rates Contrary to popular belief, data centers aren't maxed out around the clock - and this has real consequences for power system planning

@tnorris.bsky.social on why narratives about data center power demand are making it appear more constant & closer to full capacity than it actually is. Getting this right is key to heading off overbuild of generation and grid infrastructure. Big implications for climate, electricity rates and more.

ASC Public / Training Materials / LLMs On Supercomputers · GitLab TUgitLab

Looks like the course material is also all open. Good examples to show how all the pieces can fit together in practice: gitlab.tuwien.ac.at/vsc-public/t...

C++ std::simd (ok, std::datapar) is cool. Portable, order-of-magnitude speedups (w/ modern CPUs and the right, flop intense algorithms) 🚀 #HPC #SIMD
github.com/BLAST-Impact...

AIRR advanced supercomputers for the UK

🚀 ISAMBARD-AI, the UK’s new AI Research Resource, is now live!

Two access opportunities are open for researchers to tap into world-class supercomputing.

Apply now: www.gov.uk/government/p...

#AI #Supercomputing #AIRR #UKResearch

AIRR advanced supercomputers for the UK

After the official switch on of Isambard-AI last week, you can now apply for time on this amazing resource! Open for researchers and companies in the UK, find out how to apply to the UK's newest and fastest supercomputer at the following link: www.gov.uk/government/p...

Cant wait to try this out. Huge achievement!

Are GitHub stars something of a silly metric?
In my opinion, "yes."

Would I like you to star our project anyway?
Also "yes". :)

Blog Follow VAST Data’s blogs to stay up to date with our innovations across big data and AI.

The good people at @vastdata.bsky.social were kind enough to post an announcement (and Q&A about #AI) about me joining the company. Nothing too deep, but a peek at whats to come! www.vastdata.com/blog

Shorter summary: Everybody wants GPU access and computer scientists are trying to shame academic institutions into buying more. https://arxiv.org/abs/2506.19019

To my mind it’s a positive feature and not a bug that we have programs like https://nairrpilot.org/ to soak this interest up. Scaling […]

At the panel on energy efficiency beyond hardware, the topic of “why can’t we do what cell phones do?” came up. Amazed that #HPC people don’t understand that the latency requirements of a human-driven device are way easier than RDMA and memcpy.

#ISC25

Top500 website being updated messafe.

Latest #Top500 should be available soon.

UK's Isambard-AI super powers up as government goes AI crazy Brace yourselves Britain, PM Keir Starmer's challenged his teams: 'show me how they can use AI' Britain's beefiest supercomputer, Isambard-AI, is set to become fully operational this summer, as the government steps up its strategy to push AI everywhere as a driver for economic recovery.…

UK's Isambard-AI super powers up as government goes AI crazy

Found the “Welcome Tour” on Switch 2 a great way of introducing the system and goes into a lot of detail. I especially like the hidden tips like ways to get completion if you dont have an accessory or option. Now to make one for Supercomputers. #switch2

Son found the joy of new console. Download of 2 hours for included Mario Kart World and took about an hour to fully unbox and setup (inc. transfer). Welcome Tour was easy to obtain. I did offer to preconfigure but he wanted to do it. Does look good, still to try more of the upgraded games. #switch2

Abouzied M. A. Nasar, Benedict D. Rogers, Georgios Fourtakas, Scott T. Kay, Matthieu Schaller: Task-parallelism in SWIFT for heterogeneous compute architectures https://arxiv.org/abs/2505.14538 https://arxiv.org/pdf/2505.14538 https://arxiv.org/html/2505.14538

Statue of Isambard Kingdom Brunel at London Paddington with a small Brunel lego figure on it.

Heading to New York for #cug2025 - had opportunity for recently acquired lego Brunel to meet bigger version of itself.

🇨🇭 The #CSCS team is heading to #CUG2025 in New York!
Join us May 4–10 for BoFs, tutorials, talks & best papers.
We kick off Sunday with Benjamin Cumming on CPE in a Container. Let’s talk #HPC!

Brit universities told to keep up the world-class research with less cash Government boasts of £14B in R&D spending, but grant body takes £300M hit Despite ambitions to position itself as a science and tech superpower, the UK has cut the budget for the government body responsible for university research funding.…

Brit universities told to keep up the world-class research with less cash

Nvidia’s AI suite may get a whole lot pricier, thanks to Jensen’s GPU math mistake The old naming convention didn't just 'screw up' the NVLink nomenclature - it left money on the table Comment  At its GPU Technology Conference last month, Nvidia broke with convention by shifting its definition of what counts as a GPU.…

Nvidia’s AI suite may get a whole lot pricier, thanks to Jensen’s GPU math mistake

I'm looking forward to my last talk "ARMing GPUs: On the Memory Subsystem of Grace Hopper GH200" at #SCAsia2025 at the ARM for #HPC workshop tomorrow.

Learn about the quite complex and powerful memory performance at 4:15pm in room O7 🚀

Paper: buff.ly/rP7i0Nr
Workshop: buff.ly/H18dPGj

This is about to begin! Recording will be available later.

Recommend checking this out!

Vanguards of HPC-AI: Spack Builder Todd Gamblin of LLNL on the Why’s and How’s of Change in HPC - High-Performance Computing News Analysis | insideHPC In our continuing series on current and future leaders of HPC-AI, Lawrence Livermore National Lab's Todd Gamblin has a well-deserved reputation in the HPC software community as a passionate engineer w...

Congratulations, @tgamblin.bsky.social!

LLNL’s Todd Gamblin named an HPC-AI Vanguard by insideHPC

insidehpc.com/2025/02/vang...

#HPC #AI

I wrote a little text to accompany this slide. It sums up the hour-long presentation in just a few words.

www.glennklockwood.com/garden/diffe...

#HPC #AI

Photo by Chris Close shows woman looking into microscope

We provide computational training to UK researchers as part of the ARCHER2 UK national supercomputing service.

EPCC's Evgenij Belikov writes about a course he recently delivered at Cardiff University, using materials developed by The Carpentries: https://buff.ly/3EAXXzv

This is a call out to bioinformatics developers that we are working to get all of NF-Core pipelines (and by extension most of Bioconda) to support Arm.

Spot price for c8g.16xlarge (Graviton4) and it is below 60¢/hr. Or < 1¢ per physical CPU if your application supports Arm ...