Adam Thompson

This is not the Coyote server. Actual photos soon!

New Milestone: Our first validated Coyote server!

Packed into a 4U chassis:

- 192 GB of VRAM – 4x RTX 6000 Ada
- 16 TB NVMe storage
- 800 Gbps of networking

Powering the next-generation signal processing at the Allen Telescope Array. Designed for science, engineered for more!

Your section on "People Matter" has been living in my brain for days. Early career employees are often told to find their "niche", but when that niche turns into a passion, it can unfortunately limit career growth in those organizations built with, for, and by survivors.

Not a bad exhibition floor idea. Put on a VR headset and push a red button. I'm in!

DOE Announces New Supercomputer Powered by Dell and NVIDIA to Speed Scientific Discovery - NERSC: National Energy Research Scientific Computing Center During a visit to Lawrence Berkeley National Laboratory (Berkeley Lab), Secretary of Energy Chris Wright today announced a new contract with Dell Technologies to develop NERSC-10, the next flagship su...

Emerging workflows for discovery require deep integration between compute, software and service models, fabrics, storage, and external integration. Very happy to be working with both Dell and NERSC on this project to accelerate complex science.
www.nersc.gov/news-and-eve...

YouTube video by Arts & Crafts Maggie Rogers & Sylvan Esso - Anthems For A Seventeen Year-Old Girl

Two of my favorite bands playing one of my favorite bands' song. I think I may explode.

youtu.be/4IuyMBxPlLI

Ultracool microscopy yields a sharper look at proteins After years of frustration, researchers learn how to sharpen molecular images with liquid helium

"Not only will it provide sharper details of proteins and other molecules imaged with cryo-EM," says
@berkeleylab.lbl.gov 's Peter Denes, "but it should enable microscopists to image even proteins roughly half the size of what is currently possible."

Now people have to re-calibrate ratings per reviewer and medium. We know that a 4.4 Uber rating is bad, but a 4.4 restaurant is probably pretty decent.

Under the Adam Rating System™️, you merely have to think "would I evangelize this experience to a friend". If you're "meh", it's just 3 stars!

It really annoys me to read a restaurant, book, or album review and -- seemingly with no basis -- see a rating of 7.3. For one, 1-10 is way too large of a range. Two -- given 10 points of range, what separates a 7.5 from a 7.3?

We're all pedants about something, and mine is with ratings.

The Adam Rating System

⭐- Bad - would actively dissuade people
⭐⭐- Okay - would caution against if asked
⭐⭐⭐- Good - standard experience [default!]
⭐⭐⭐⭐- Great - would recommend if asked
⭐⭐⭐⭐⭐- Excellent - would evangelize to people

Today is a good day.

Used NVIDIA’s Global Free Days this quarter to celebrate our 10 year wedding anniversary in beautiful San Miguel de Allende, Mexico!

Everyone is singing the birthday song way too slowly. It’s a celebration, not a dirge!

If you're at #AGU24 this week, consider attending "Automated Seismic Detection for Planetary Missions" on 12/11 at 10:20a to learn how we can conduct better far-edge missions by detecting good data at the source and not spending energy to transmit noise to the Earth!

agu.confex.com/agu/agu24/me...

I have been playing the Wingspan board game with my wife over the past couple evenings, and it's such a beautiful game and playing experience. I love how each game unique, how it's competitive -- but not cut-throat -- and most of all, it's fun to learn about birds!

Sure would be nice to be able to upload a photo of yourself and see all the matches automatically! If only there were AI companies that could help ;).

400GbE streaming from FPGA to GPU in Radio Astronomy! arxiv.org/pdf/2411.15630

It's been 2 years since many of us have connected, so here are some updates!

#cuSignal has been deprecated as an NVIDIA product and merged into #CuPy. We closed out with over 400k Conda downloads!

I now lead product for Edge HPC, where we think about connecting GPU compute to large instruments.

I'm biased, but I thought #SC24 was particularly special as it featured a lot of new folks to HPC (myself included!). While I've been on the "HPC Orbit" for years, the emergence of converged edge, AI, and simulation workflows - led by DOE initiatives like IRI and HPDF - has me fully committed.

So great to interact with you again, Jeremy! Love that the community is re-uniting. ❤️‍🩹

Yep! It was on the backside, which ended up being a really great spot because it was between NVIDIA and AMD. Lots of foot traffic! ☺️

Sharing our #SC24 demo built in partnership with the @setiinstitute.bsky.social. 100Gbps of streaming #RF data is RDMA’d from RFSoC to CX7 + RTX A6000 where it is then processed + AI inferenced in real-time. AI for real-time instruments is here!

👋🏻 Bluesky. So great to see friendly faces from #SDR and #HPC ex-Twitter here! These worlds have united for me, and now I’m deeply thinking about data streaming to HPC. Had a great time at #SC24 last week talking, learning, and sharing!