hpcnotes's profile picture. 25 yrs using/researching/buying #supercomputers. Now an engineering leader for #supercomputing capabilities at Microsoft. Posts on #HPC #AI #cloud #F1 #travel

Andrew Jones

@hpcnotes

25 yrs using/researching/buying #supercomputers. Now an engineering leader for #supercomputing capabilities at Microsoft. Posts on #HPC #AI #cloud #F1 #travel

Andrew Jones รีโพสต์แล้ว

1.1M tokens/sec on just one rack of GB300 GPUs in our Azure fleet. An industry record made possible by our longstanding co-innovation with NVIDIA and expertise of running AI at production scale! techcommunity.microsoft.com/blog/azurehigh…


Andrew Jones รีโพสต์แล้ว

1/ Just wrapped our earnings call and wanted to share a few highlights from the quarter. Bottom line: We’ve got great momentum across our AI platform and our family of Copilots, fueling our increasing investments in both capital and talent.


Andrew Jones รีโพสต์แล้ว

The next chapter of the Microsoft–OpenAI partnership: aka.ms/AAyczj7

Microsoft's tweet image. The next chapter of the Microsoft–OpenAI partnership: aka.ms/AAyczj7

Inspired by 18 innings? Interested in #AI, #Supercomputers, #Cloud, #HPC? Attending #SC25? Get your place at the sought-after #Microsoft + #AMD SC reception! Tue 18th, 6.30pm, Busch Stadium. Registration & approval required: aka.ms/microsoftamd-m… #HPCignites

hpcnotes's tweet image. Inspired by 18 innings? Interested in #AI, #Supercomputers, #Cloud, #HPC? Attending #SC25?

Get your place at the sought-after #Microsoft + #AMD SC reception!

Tue 18th, 6.30pm, Busch Stadium.

Registration & approval required:
aka.ms/microsoftamd-m…

#HPCignites

Andrew Jones รีโพสต์แล้ว

This is the first of several outcomes from a larger, longitudinal study we've been conducting to understand AI workloads in production. It's been fun writing code and applying fancy stats again. #HPC #AI

Inverse correlation: The bigger the frontier model training job, the less I/O you need per GPU. This is one of the counter-intuitive learnings that @glennklockwood teases out after analyzing nearly 100,000 checkpoint operations on frontier model training systems. @VAST_Data is…

JeffDenworth's tweet image. Inverse correlation: The bigger the frontier model training job, the less I/O you need per GPU. This is one of the counter-intuitive learnings that @glennklockwood teases out after analyzing nearly 100,000 checkpoint operations on frontier model training systems.

@VAST_Data is…


Imagine you could visit 5 #supercomputer centres anywhere in the world - but only 5 - to learn the most about the state of the art for #HPC, plus future trends, etc. - which 5 would you choose and why?


Is it time for a #HPC community debate about the cities that are being picked for @Supercomputing? #SC25 is really struggling for available hotels near the conference center. Plus some people are worried about the street safety reputation of STL and other recent SC cities.


Woah! How is it the middle of October already? When did that happen?


Andrew Jones รีโพสต์แล้ว

Another first for our AI fleet... a supercomputing cluster of NVIDIA GB300s with 4600+ GPUs and featuring next gen InfiniBand. First of many as we scale to hundreds of thousands of GB300s across our DCs, and rethink every layer of the stack across silicon, systems, and software…


One side effect of having lots of LED light bulbs around the house is that you forget you have the occasional bulb that is still the old filament style and also that those filament bulbs get really hot really quickly. Anyway so I have a blister on my finger now.


The HPC (+ AI Infrastructure) Leadership Training Institute this week at BSC hit registration capacity weeks ago - extra places were opened & they went fast too, and people still on a waiting list. Nice to see the guess of demand to run the first ever HPCLI in Europe was good!


Sunrise in Barcelona

hpcnotes's tweet image. Sunrise in Barcelona
hpcnotes's tweet image. Sunrise in Barcelona
hpcnotes's tweet image. Sunrise in Barcelona
hpcnotes's tweet image. Sunrise in Barcelona

Microsoft announces #Fairwater #AI #supercomputer Over 100,000 #GPUs in single system in custom liquid-cooled datacenter 10x performance of world’s fastest public supercomputer Plus CPU supercomputer with millions of cores And exabytes of storage. blogs.microsoft.com/blog/2025/09/1…


Andrew Jones รีโพสต์แล้ว

If intelligence is the log of compute… it starts with a lot of compute! And that’s why we’re scaling our GPU fleet faster than anyone else. Just last year, we added over 2 gigawatts of new capacity – roughly the output of 2 nuclear power plants. And today we’re going further,…


Microsoft announces $30B investment in UK, including 23,000 #GPUs of #supercomputers for #AI

We’re committed to creating new opportunity for people and businesses on both sides of the Atlantic, and ensuring America remains a trusted and reliable tech partner for the United Kingdom. That’s why today we announced a $30 billion investment in the UK over four years,…



Boarding @British_Airways for Seattle … oh, nope, crew got on then back off again to say the aircraft hadn’t been cleaned at all, so now we all waiting at gate for estimated 40mins delay before boarding while aircraft gets cleaned.

hpcnotes's tweet image. Boarding @British_Airways for Seattle … oh, nope, crew got on then back off again to say the aircraft hadn’t been cleaned at all, so now we all waiting at gate for estimated 40mins delay before boarding while aircraft gets cleaned.

Views of London through the clouds on way in to Heathrow earlier today.

hpcnotes's tweet image. Views of London through the clouds on way in to Heathrow earlier today.
hpcnotes's tweet image. Views of London through the clouds on way in to Heathrow earlier today.

Hearing the announcement: “boarding complete” for on time departure = good. Followed by pilot: “there is a fault with the aircraft so there will be a delay before we can go …” = less good.


Loading...

Something went wrong.


Something went wrong.