Dan Ernst

It's in a museum, so it must be true. #HPC

Ian Buck, VP of Hyperscale and HPC Computing at NVIDIA, sat with @dariogila, Undersecretary for Science at @ENERGY, for a conversation on the future of scientific discovery. As the Undersecretary announced first-round DOE funding to support the Genesis Mission, Ian underscored what it takes to bring that vision to life: advanced computing infrastructure, strong public-private partnership, and close collaboration across national labs, industry, and universities. NVIDIA is excited to be an industry partner of the DOE Genesis Mission.

@tomascooking @tunguz GB200 has full rate FP64 GB300 is 1/64

@tunguz We’re going to look at it for our next. Our NVL 72 GB200 is just coming online this month. I hope we get the FP64 section of the die back on Rubin. The 1/64 ratio isn’t my favorite.

“Vera Rubin is in full production.”

Oh hey #SC25. We have a surprise for you...

To be chosen to participate in the design of the next iteration is a massive honor for us. Through this project, we target a scientific discovery ecosystem based on full integration of simulation and AI. We hope this will benefit not only Japan, but also researchers worldwide.

Those who have been around #HPC a while know the historical significance (and probably the names) of Japan's flagship supercomputers. They always push the boundaries of technology, but in ways that are driven uniquely by the needs of Japan. blogs.nvidia.com/blog/fugakunex…

理化学研究所、富士通およびNVIDIAとの国際連携による「富岳NEXT」開発体制を始動 riken.jp/pr/news/2025/2…

🗣️📢 We're thrilled to work with The U.S. Department of @ENERGY on the flagship NERSC-10 supercomputer! Powered by our most advanced liquid-cooled server tech & @nvidia's next-generation accelerators - the "Doudna" system fuses #HPC & #AI to empower complex, integrated research workflows within a unified AI-optimized environment. 📰del.ly/6013N4XRT

@HatemLtaief @HPC_Guru @GeorgHager If you look at DFMA flops (and not GEMM/Tensor Core), the memory bandwidth for NV GPUs has actually grown relative to FLOPs in every generation since Volta. That's the ratio that matters for most simulations.

@HPC_Guru @ernstdj @GeorgHager 2!!! Even worse with a wider gap for #GPUs. Credits: graph from Thomas Gruber (#likwid)

Which of these is more pertinent today? 1. The most efficient FLOP is the one you didn't use 2. The most efficient BYTE is the one you didn't transfer #HPC #AI @HatemLtaief @ernstdj @GeorgHager

Generating cat videos is nice, but what if you could tackle real scientific problems with the same methods? 🧪🌌 Introducing The Well: 16 datasets (15TB) for Machine Learning, from astrophysics to fluid dynamics and biology. 🐙: github.com/PolymathicAI/t… 📜: openreview.net/pdf?id=00Sx577…

@HPC_Guru @thedeadline @HPCwire I miss the Scinet plaintext password wall of shame. 😂

RIP Prof. John D. Little. He of Little law; the most ubiquitous name in computer performance analysis (even though he knew nothing about computer systems) en.wikipedia.org/wiki/John_Litt…

@glennklockwood @RHAlexander I was going to help until you revealed your purpose 😁 There are a couple of centers looking at it, but I don't think any have them yet.

@RHAlexander That just means you can't buy one on a whim. More generally, I am wondering with whom I can commiserate about GB200 🙂

Outside of one-off demo racks, do any scientific computing HPC centers have plans to buy GB200 NVL36/NVL72?

.@xAI's Colossus in Memphis, the world's largest AI supercomputer with 100,000 NVIDIA Hopper GPUs, achieves new heights with NVIDIA Spectrum-X Ethernet. A testament to NVIDIA's dedication to #AI progress. Read more: nvda.ws/4fnuuWG

I'll be sitting on a @Supercomputing panel discussing the impact of low precision datatypes on HPC codes. Mark your calendars for Nov 19th, 3:30 ET and see you in Atlanta! CC: @ernstdj

With this and the physics prize, the Nobel committee recognizes we are in a new era of 'AI for Science'

Totally should be changing my LinkedIn to say "Applied Physicist" this morning.

It's aliiiiive

Microsoft Azure is the 1st cloud running @nvidia's Blackwell system with GB200-powered AI servers. We're optimizing at every layer to power the world's most advanced AI models, leveraging Infiniband networking and innovative closed loop liquid cooling. Learn more at MS Ignite.

ETH Zürich’s new Alps #supercomputer, powered by 10K+ Nvidia Grace Hopper chips, is set to accelerate breakthroughs in #AI for materials science, universe mapping, and healthcare. The future of scientific discovery is here: ow.ly/rnG050TEBA7

@simonmcs So happy this machine had such a great run!

After 6 years of service the Isambard 2 supercomputer is finally being retired! Starting in May 2018, Isambard was the world's first production Arm-based supercomputer, employing ThunderX2 CPUs. Isambard 3 takes over today using NVIDIA Grace Arm-based CPUs provided by HPE