Quantum #65
Issue #65 of the weekly HPC newsletter by HMx Labs. Supercomputing Strategy Group Launch, controversy in IO rankings and some info from ISC workshops.
Well the most important thing last week was clearly HPC Club, not that I’m biased at all 😁
Some great conversations around the future of scheduling in HPC and a fantastic panel discussion with guests coming in from Paris and New York to make it happen. Love it.
The success of HPC Club has given me a sense of hope. For several years now I’ve watched as even behemoths like Amazon have not meaningfully moved the needle in HPC. My apologies if that offends, that is not my intent, and I am certainly not saying there has been no progress. Just that it falls short of what I want to see and what I’m seeing quietly demanded by (and not delivered) by clients and others in the industry.
Part of that I think is because these are problems that can no longer be solved by one organisation. Even when that company is at the scale of Amazon or Microsoft. Certainly not with AI is sucking all the air (and money) out of the room.
Which is what led me to our next little experiment. Supercomputing Strategy Group.
Much like HPC Club the intention is to bring people and companies together. The big difference is that it isn’t just to talk. To be involved you must be willing to help move the needle in some form. Whether that is through expertise, influence or something else. This isn’t just social. Its work.
I believe we have enough initial traction to make this viable. I have some ideas of my own, both on how this should work, and what I’d like to see in future in HPC. But what I want to do most of all, is to establish common ground and work from there. Together.
Want to shape the future of HPC? You know where I am.
In The News
Updates from the big three clouds on all things HPC.
So how big is the HPC market now? And how much of that is AI? Intersect 360 has some numbers for you
Doug Eadline
Lots of controversy last week around who has the fastest filesystem with DDN trash talking Weka who were also on stage at ISC claiming to be the fastest. Sit back, grab the popcorn and watch… anyone serious about IO performance stopped using filesystem years ago in my world. The whole debate around having the fastest horse when the rest of us moved on to cars is rather amusing.
DDN Team
From HMx Labs
Both articles this week are related (in some way) the workshops that I attended at ISC last week. The first of these is about Flux Framework, a workload scheduler used at 6 of the 10 largest supercomputers in the world including El Capitan at the number one spot:
The second is a look at the pains of parallel programming and in somewhat of a continuation of the theme from last week of comparing HPC in FSI with classical HPC
Know someone else who might like to read this newsletter? Forward this on to them or even better, ask them to sign up here: https://cloudhpc.news
