The most interesting story last week wasn’t details on a new CPU or accelerator, a new product version or cool new startup with a solution in the supercomputing space. It was just an idea, a thought experiment. Anyone working in HPC for more than a hot minute (so non of you AI-come-lately types then 😆 … I’m kidding, relax already!) knows that raw compute hasn’t been the problem for quite some time. The memory wall is all too real. Jack Dongarra even talked about it in his closing remarks at ISC 26. If modelling memory (even if that’s only in the L3 cache) as something other than a linear block can solve that, I’m game to try it out.

The DOE (USA) is in a real rush to get some real-world science use out of quantum computing and has put up a cool $215m (what’s that, half an NVL144 rack these days?) in a competition. Shame we can’t get some of that if we manage to get financial risk analytics working on quantum huh. 🤣

Nvidia had a couple of interesting announcements (and neither of them are about GPUs). First up they presented DSX, which, strictly speaking, is a little boring because it’s just about getting various data centre utilities to talk to each other so they can all operate at the right level and use less power. Overall, that’s kinda cool though, right? I have to say I’m surprised Nvidia didn’t just try and solve that by throwing a swarm of agents at it and getting them to just control all the knobs in the DC. I am quite glad they didn’t though!

The second release is also power related, but this time around integration with the grid to redirect compute to where power is available. Again, also kinda cool. If/ when it works it will be at least anyway.

We also had Cornelis and Delos Data both release products to provide an alternative to Nvidia’s now closed networking based on its Mellanox acquisition. I’ll admit, I do have my fingers crossed some of this stuff works out.

A little less notice than we normally like to provide but we have finally set a date for the last HPC Club for this year. October 14^th and it will take place in London. We should have the booking link live on Wednesday morning so keep an eye out for that if you’d like to come along, If you want one of the fancy new T shirts then you’ll definitely want to come along (and maybe try and bribe Mugheesa to make sure she orders one in the right size for you).

Have a good week all.

All the News in Depth

Cloud and vendor releases for Supercomputing & HPC (and AI too if you change the filters) including new Z4D VMs from Google.

Anyone working in HPC is no stranger to the memory wall, so this is an interesting solution / thought experiment

DOE lines up some serious cash to accelerate quantum computing adoption in real world use

Nvidia gets DC infra to communicate to help save power

Couple this with a standard for getting DCs to talk to each other and the grid…

Supercomputer interconnects other than the old Mellanox gear that Nvidia now owns start picking up

HMx Labs Updates

The next HPC Club will take place on 14 Oct in London. 

HPC Club: 14 Oct 2026

The next HPC Club will be in London on 14 Oct 2026. Mark your calendars, book your babysitter

Flux by HMx LabsHamza

And we have some new T shirts to give away too. The only way to get one is to come and snag one in person

New HPC Club T Shirts

All new HPC Club T-Shirt design just landed

Flux by HMx LabsHamza

Off Topic

Scaling LLMs. I guess we all kind of knew this instinctively but now MIT has maths that says it too.

If you’ve already seen the original video on enshittification by Norwegian Consumer Council this is a great follow up!

Just in case you’ve been living under a rock and haven’t seen the original

The modern life of a software developer getting you down? You’re not alone

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

I’ve alluded to new designs for HPC Club T shirts a few times this summer… I think I’ve teased you for long enough. Here’s the first one!

OpenAI can suck it! IYKYK 🤣

The only way to get hold of one of these is going to be by coming along to the next HPC Club event on 14 Oct or winning one on the random occasions I run silly competitions.

HPC Club. 14 Oct 2026 @ 17:00 in London

A bit less notice than we try and give people normally I’m afraid but hopefully still plenty of time for you to mark your calendars, book the babysitter or get your night out pass stamped.

We’ll be talking about software defined supercomputing, and in particular about being able run multi cloud compute clusters that you can create and destroy on a whim. Why you might want to and how you can.

Venue details and sign up link coming soon.

Your weekly supercomputing and HPC news update, and its short and bitter this week. You could argue this is because I was otherwise occupied last week and spent a large proportion of the weekend behind the wheel so didn’t have time to do it justice.And you might have a point. I prefer to think it’s because there wasn’t much that was genuinely interesting so I’m keeping it short rather than wasting your time with the regular news cycle everyone else already covered 😁

First up, unless you’ve been under a rock all week (and I have but this still made enough noise that I couldn’t miss it), it seems we now have a solution to the Navier Stokes problem. I won’t get into the furore around who actually solved it (see the Off Topic section for more if you do care), what I care about is, will this mean I can run less compute for my CFD? Does my weather forecast, my turbine blade or F1 car rear wing design suddenly get cheaper? As best as I can tell (and I could well be wrong here) the answer seems to be no. (Ok it makes some of the troubleshooting easier when it all blows up in your face because you know the math isn’t the problem anymore). Which makes this all a big meh really.

The only other interesting thing was a press release from d-Matrix around its up and coming (yet to be taped out) Ai inference accelerator. Yea yea, this again confirms we live in a heterogenous silicon world, but I suspect you’re tired of me beating that particular drum. No, what’s more interesting about this, is that it will live inside an Nvidia NVL144 rack. Yep, an Nvidia rack with someone else’s accelerator inside it. Things that you go hmmm…

Not a massive update from HMx last week either, but if you like parallels between AI and other things, I wrote a take comparing AI to digital photography.

May your FLOPs be error free this week. Adios.

All the News in Depth

Cloud and vendor releases for Supercomputing & HPC (and AI too if you change the filters)

Nvidia racks with 3^rd party accelerators in them? Yep that’s a thing now.

You proved Navier Stokes and won a million bucks. Whoopee for you. What does it mean for those us actually trying to model fluid flow???

HMx Labs Updates

A few thoughts on the impact of AI on software engineering and a slightly different comparison to the usual narrative

Vibe Coding & Photography

Are LLMs to software engineering what digital cameras were to photography?

Flux by HMx LabsHamza

Off Topic

I don’t want to get into it, but Primeagen (and Fireship and others) did

He also has an opinion on the end of programming. Wouldn’t normally have included this one but since I wrote about this topic too…

For completeness some of the input that inspired the comparison to photography

https://www.youtube.com/watch?v=C1ZHQyjrVMs

https://www.youtube.com/watch?v=pVMM23kUVH8

https://www.youtube.com/watch?v=qdG7kygCdsU

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

We often compare creating software and the arrival of LLMs with the horse and carriage and the arrival of the car. I think that’s a flawed analogy. And I think I have a better one.

I’d like to claim some form of superintelligence here but the truth of it is that it’s not even my own idea but a bit of an amalgam of a few ideas from people in the photography and software communities. I think LLMs might be for software creation what digital cameras have been for photography.

Film photography is like creating software without LLMs. Taking pictures with old film cameras required you to understand how to load and process the film (or use the services of someone else who that for you). You had to understand how to expose the film correctly without any live feedback on a little screen and you wouldn’t know if you’d got it right for hours at best and possibly days to weeks at worst once the film was developed. A significant part of taking a photograph involved technical skills around the operation of a camera, a light meter and dealing with film. There were entire professions that did this and never pressed the shutter button to take a picture. All this before we even start thinking about what the picture is even of.

Without getting into a complete history of the evolution of digital photography, I think we can agree that modern digital cameras (think of the phone you’re reading this on or in your pocket) make it such that getting those basics wrong is close to impossible. You’d have to try pretty hard to take a picture that is incorrectly exposed and you’d know about it before you pressed the button.

We have more photographs (real ones, not even AI generated ones) in existence today than we ever had. Many more of the photographs we take are of no interest to anyone but ourselves. Yet we still pay people to be photographers. It’s still a viable career.

The work of a photographer shifted. A photographer still understands how to correctly expose a photograph (or when to deliberately not) but the amount of effort in that is far lower than it used to be. Instead, the work shifted to the content of the photograph. How its framed, the composition, the lighting, the subject and most importantly, the story it tells. This was always important. Great photographs even from the film era still needed this.

I think creating (and I used the word creating very deliberately) software is in the same boat. LLMs (and vibe coding and tools such as Lovable in particular) have taken away much of the skill that was needed in creating software. In much the same way anyone can take a competent picture with their phone today, anyone can create software using these tools. In much the same way digital cameras have shifted where the effort of a photographer goes, LLMs will (have?) shift the effort of software creators. Just because you can take a picture on your phone doesn’t mean anyone else wants to see it. Just because you create a new game using Codex and Astra doesn’t mean anyone else will want to play it. And that might be just fine.

Photography today is in an interesting place. There’s a revival of interest in film photography. People are paying inordinate prices for old digital cameras to obtain a “filmic” look to their photographs. Modern cameras (and your phone) come with built in modes to make your photographs look like they’re old.

I’m sure part of this is just the cyclic nature of fashion and trends but a part of it, at least from the photographers, is chasing the process. A desire to make photography slower, more deliberate. Removing the work that went into taking a photograph took away something from the final output too.

I think we’re seeing the same with software. The slow and difficult process of creating software wasn’t just a barrier but promoted deliberation, more refinement before ever typing the code. Sometimes the limiting factor isn’t the ability to write the code but to refine the idea.

The photo? I’ll let you guess how it was taken and how old it is.

The most classical supercomputing aligned news (even if it is for an AI system technically) from last week must be the selection by EuroHPC of an AMD based supercomputer for the replacement of LUMI. AMD has dominated the Top500 for some time now and seems to be cementing that lead even for AI capable systems. At least for those that are owned or operated by classical HPC/ supercomputing institutions. Maybe FP64 is kinda important after all huh. Who would have guessed 😁 Especially if you’re running converged classical and AI workloads.

CERN moved to Debian (from RHEL)! Before we start the distro wars again (probably coupled with AI ideologies) the details here are important and make this a fairly boring and eminently practical and sensible decision. The move is specifically for its computes that act as controllers. The upgrade cycle is around 15 years. 3 year LTS releases can stay at home! Combine this with RHEL9 dropping support for anything below x86-64-v2 and RHEL10 x86-64-v3. Many of the control boards at CERN don’t meet that requirement and it's far simpler to just use a distro that will support them rather than upgrade a lot of very bespoke hardware. Put your pitchforks away please.

In other Linux related news, Amazon Linux 2027 preview was released last week and early testing of the 7.3 kernel show improved performance on AMD EPYC CPUs for certain types of workloads. So that’s good news.

IBM also released its latest quantum chip, Nighthawk. Lots of impressive sounding improvements there but still relatively little in terms of what I can actually do with it back herein the real world.  We haven’t forgotten about this either and are still plugging away on our little side quest to see how viable it is to run financial risk analytics (MC based at least) on a quantum machine.

Azure delivered another little mini gem in the form of a public preview of its multicloud interconnect. No doubt driven by regulatory concerns around concentration risk and the subsequent demands from its customers, but it does mean we are slowly creeping into a a multicloud world.  Just need to do something about those pesky egress costs 😉

Have a good week y’all.

All the News in Depth

Cloud and vendor releases for Supercomputing & HPC (and AI too if you change the filters)

EuroHPC Goes AMD for its AI Supercomputer. AMD Press Release.

CERN moves to Debian! Or watch the presentation here.

IBM releases Nighthawk, its latest Quantum computer. HPC Wire has a more in depth explanation.

Multicloud compute is slowly getting more and more real (Azure)

HMx Labs Updates

If you didn’t guess already, HPC NErD was a CERN last week.

I’ll have to go digging in my camera SD cards to see if I can find another little puzzle for later this week.

Off Topic

Nothing to see here, just capitalism doing what capitalism done when no one is around to regulate. Gotta keep that bubble inflated.

Well of course I was going to include this one… its finance folks talking tech (or about that at least)

Fireship somehow manages to sound anti AI while still making the AI thing sound pretty far out. I mean, I thought it was pretty cool.

Fortunately, Cal Newport has a pretty down to earth take on things which had me back to thinking, meh, it’s just LLM stuff.

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

Hot Chips. The silicon conference as opposed to the calorie laden but rather tasty potato based food. That was the was the main source of supercomputing news last week. We naturally had the AI/ GPU related updates from the usual suspects but we also had insights into CPUs from Intel, Fujitsu and ARM.

Let’s talk Intel first, although it’s possibly the least exciting update its probably also the most relevant one for most of us. Xeon7, code named Diamon Rapids, will give us more cores than originally anticipate. A full 256 of them in fact. But that’s yer lot in terms of threads or logical CPUs too I’m afraid. For some incomprehensible reason, Intel, in their infinite wisdom decided that Hyperthreading (or SMT to give it its unbranded name) isn’t needed anymore. Oops. Slated for a 2027 release but if you want to wait for something with SMT you’ll need to hold out for Coral Rapids.

Sticking with CPUs and supercomputers for a bit longer, Fujitsu provided more details on its 144 core Monaka ARM v9 architecture chips. The previous generations powered Fugaku but with ARM’s increased popularity we might see these more widely used? They’re built on a 2nm process and have 2 SVE2 256bit execution units so could be pretty handy for your SIMD workloads. So long as you port them to ARM.

Talking of ARM, the company rather than the ISA this time, they also released more details on their confusingly named new CPU, AGI. As the name implies ARM seems to be chasing AI money here claiming that the CPU is targeting agentic workloads. Honestly, I still have no idea exactly what demands an agentic workload has on a CPU. Surely this is IO constrained anyway? What makes agentic special here?

Since we’re firmly into AI territory now anyway, Nvidia also provided more details on their NVL72 rack scale GPU solution. I had a funny thought here. As we see more rack scale deployment,  are we about to see what happened in the consumer space with devices becoming smaller, more power efficient but also less repairable and upgradeable in the data centre space? DIMM failure? Time for a new rack. Hmm…

And of course, OpenAI gave us details of Jalapeno, their new inference engine. I’ll let you read the link below for the details on that one.

In more fun news, AWS EC2 was 20 years old last week. Happy Birthday! And Azure will give us 248 and 372 vCPU sized machines to celebrate 😁. But Microsoft giveth with one hand and taketh away with the other. HPC Pack, their faithful but aged scheduler is being retired.

Things from us at HMx have been pretty quiet still last week… I haven’t made any progress at all on my piece about compute markets I’m afraid so this week you’ll just have to have more fun guessing where HPC NErD is again.

Have a good week y’all.

All the News in Depth

Cloud and vendor releases for Supercomputing & HPC (and AI too if you change the filters)

Intel Diamond Rapids… no Hyperthreading but you do get 256 cores.

ARM shares details on AGI, its first foray into the CPU market

Fujitsu’s new ARM CPU

A closer look at Jalapeno, OpenAI’s new inference engine

More details on the NVL72

Happy Birthday EC2

HMx Labs Updates

If you didn’t guess (might have been a bit hard) last week HPC NErD was at Neuchatel at a temporary exhibit on the measurement of time. A new HPC Club T shirt if you can guess where he is this time.

Off Topic

How’s your token spend going? Getting much in return?

What is AI coding doing to you?

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

There were a few interesting updates in the world of supercomputing last week, nothing life changing but certainly some things of note to keep an eye on. 

First up we get the new Linux 7.2 kernel release which is worth keeping eye on for supercomputing folk because it includes within its many updates the ability to use cache (we’re talking CPU caches here) aware scheduling. i.e. you have the ability to run code on cores where the local cache already has the data you need. For multi socket and NUMA systems (even on a single machine, but even more so at supercomputer scale) this gets pretty useful. It might be a little while before these changes trickle down to your favourite distro but at least you know they’re on the way giving you time to start updating your code to take advantage of it.

Microsoft Azure has finally seen sense and is allowing you to choose how you use the compute you rent. You can now enable and disable cores on a VM as you see fit as well as enable and disable SMT/ Hyperthreading. And not too soon! Whilst disabling SMT where it was already enabled was previously possible on Azure, it did require using a fairly poorly documented API which itself was only enabled once you raised a support request for it. I don’t understand the fascination CSPs seem to have with wanting to dictate how we use their machines. Price it how you like but let me do what I want with it. Come on AWS, let’s see the same change from you now. Let me enable SMT on your [CM]7a VMs.

Not new, but HPC Wire provided a fairly nice write up of scale up, vs out vs across networking (for AI) as practised at Meta. There’s been a bit of chatter around this over the last week, so it seems like a timely piece. To my FSI friends who are reading this and thinking, what’s new here we’ve been doing this for years (just not at quite the same scale) then understand, this why you need to start talking a bit more about what you do in the supercomputing space!

We also had a nitrous oxide fuelled, turbo charged version of Cerberus’ wafer scale inference engine drop last week. CXL made a bit of progress into something that’s edges from lab to reality and AMD dropped a bunch of pull requests to get its Ultra Accelerator Link open standard GPU networking protocol into the Linux kernel.

Lastly, if you’ve been missing the dulcet tones of Boof breaking down and explaining HPC concepts and haven’t yet found his new YouTube channel which is now unencumbered from the oversight of AWS give it a look.

And that’s yer lot. Have a good week all.

All the News in Depth

Cloud and vendor releases for Supercomputing & HPC (and AI too if you change the filters)

Linux 7.2 gets cut and includes cache aware scheduling. Tasty.

Microsoft provides better support in Azure for controlling SMT and active cores.

Nothing new, per se, here but it’s not a bad summary of scale up, out and across networking which seems to be getting a bit of attention lately. 

AMD pushes PRs to integrate UALink into Linux

Cerberas overclocks its AI inference

Want to use CXL? Maybe it's getting closer to reality with Marvel and Synopsys releases.

I know this is old, but I somehow forgot to include it my updates so far! If you managed to miss Boof’s new YouTube channel it’s time to hit Subscribe on it

HMx Labs Updates

Not much from us last week… I’m in summer mode so here’s a picture instead. Free HPC Club T shirt if you can guess where HPC NErD is or what's in the background.

Off Topic

Deepseek now provides a coding harness based on a plugin model.

More anecdata supporting AI generated anything not being all that awesome after all

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

You’d be forgiven for thinking that if you exclude all the mainstream AI news, especially everything that’s more about financing and power grids that nothing much of note happened last week. Even my AI research bots turned up little to write about. A little old fashioned manual digging in release notes did turn up a couple of interesting stories though.

We have word from Microsoft that that they will not only be releasing an update to their own AI chips (Maia) but also rolling out a surprisingly aggressive number of them. Treat that statement with a pinch of salt but it does rather confirm the very heterogenous nature of compute. Yes, present tense. Now.

Nvidia gave us an open source LLM router which I suppose could be interesting in some domains of HPC to try and control cost where we have converged LLM inference and classical compute. I’m not sure what the workflow for that looks like though. I guess you need the routing decision before the scheduling decision else you can’t allocate resources correctly. I wonder how long it is before we see an LLM router integrated into a HPC scheduler. For LLM inference only workloads this seems like a bit of a no brainer surely? 

How is it possible that even with AI genies writing code at a rate that eclipses my fat fingers on a keyboard that I still have more possible ideas than engineering capacity. 

Riddle me that batman. 

Talking of converged AI and HPC compute, Google gives us AlphaFold for HPC as an addition to their catalogue of offerings. Looks like an out of the box way to run AlphaFold on Google Batch to reduce your compute wall time.

AWS also gave us Kubernetes style health checks for the application layer in EC2 (see the release notes in Noteworthy). Please let this be another nail in the coffin of HPC on K8.

Lastly, we welcomed Eduardo to the team last week and I had a bit of a rant about synthetic and natural intelligence.

All the News in Depth

Cloud and vendor releases for Supercomputing & HPC (and AI too if you change the filters)

Microsoft is still investing in its own Nvidia alternatives for AI focused compute with the next generation of Maia processor due for release imminently.

Nvidia provides an open source LLM router

Google adds to their converged AI/HPC capability with AlphaEvolve for HPC

HMx Labs Updates

Not sure if this is a rant, man shouting at clouds or just incoherent mutterings of a mad man. I can say none of it was written, edited or illustrated by AI. I think I had a few interesting things to say but do feel free to disagree

AI: Context, Taste & Inspiration

Somewhere between a rant and some random thoughts on AI.

Flux by HMx LabsHamza

Also, welcome to Eduardo!

Eduardo Joins HMx Labs

Welcome to Eduardo Visinheski who joins us at HMx Labs as our newest HPC Engineer.

Flux by HMx LabsHamza

Off Topic

AI progress is accelerating on an exponential curve right? Right? Maybe not?

Is the Dunning Kruger effect just nonsense? I don’t know, but I can say I agree with Sabine’s last point, that data cloud doesn’t look like it has any kind of correlation at all!

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

Time to go off topic a little and stray into a mass market subject. AI, context, taste and inspiration. I’m not sure I have anything coherent to say but I had a few thoughts knocking around in my head that I thought I’d share. Consider yourself forewarned.

First up, context. Frontier models (hate that term) have a context window of 1 million tokens. That sounds pretty impressive. It’s not. Ok I guess compared to what was available a year ago, it is. Compared to the hyped claims of AGI its pathetic. Plenty of people have already explained that it’s too small a context window to hold even a moderately sized real codebase. I don’t want to talk about that.

A decent code harness can mostly work around that limitation by grepping through your files and reading the bits it thinks match. Honestly, that’s often good enough and if you don’t know the code its certainly faster than attempting to do something similar manually. 

Know what it can’t do? Hold 20+ years of experience in its head. Or even the 2+ years of a junior developer. It doesn’t know your company’s infrastructure restrictions or regulatory requirements. Things even your most junior developer picked up in the first week and stuck with.

This often gets banded about as “taste”. Another stupid word that’s being used to hide the fact that it’s just more context. Context that the model doesn’t have access to. Context that the model can never hope to hold in its context window. Taste is a stupid word. It’s just knowledge. But if we call it that, it belittles the AGI narrative. 

Taste is just context your model can’t access. 

Want to stay in a job? Make sure you always have more context than the AI.

Still think that AI can be creative? Only if your idea of creativity is reversion to the mean. Let me try and illustrate this with a bit of anecdata and I won’t use code for these. Earlier this week I wrote a post introducing our newest HPC engineer. When I sat down to write that post I had the first three lines only. This was all I could come up with:

Just a quick note to welcome our latest HPC engineer, Eduardo to the team. Welcome Eduardo! That’s all folks!

I sat there with just that on the screen in front of me for the longest time. Then I cracked. I ask ChatGPT 5.6 High. Yea. I wasn’t about to mess around. Throw the most capable expensive model at it right away. I told it to write the post in my own style, and this was what it came up with:

Very pleased to welcome Eduardo Cook Visinheski to HMx Labs as our newest HPC Engineer.

Eduardo joins us to help work on the increasingly difficult problems that sit somewhere between HPC, cloud infrastructure and “why is this behaving differently on that machine?”

Exactly the sort of problems we like.

Welcome to the team, Eduardo. Looking forward to building some interesting things together.

🤮

If I ever produce anything quite so bland please take me out back and give me a good kicking till I see sense. I thoroughly admonished the poor 4 bit matrix math addled simpleton and off it went toread more of my previous ruminations and managed to spit out the following:

“HMx Labs has a new HPC Engineer.

Welcome Eduardo Cook Visinheski 👋

We’re building a team around a fairly simple premise: there are a lot of interesting problems in HPC that still aren’t solved particularly well.

Cloud. Scheduling. Performance. Portability. Heterogeneous compute. And computers occasionally producing different answers because you looked at them funny.

Eduardo has decided to join us in trying to fix some of them.

Excellent decision. Possibly.

Welcome to HMx!” 

Better. Still not me. And critically, still garbage. I’d never put my name to that.

Here’s where I will give it some credit. After having been stuck on a largely blank page for an hour at this point, the above did manage to trigger me to write the following paragraph.

“Eh. You’re still here? Ok fine. Eduardo’s made the mistake of letting himself be duped into solving the kinds of problems in supercomputing that everyone else would rather pretend don’t exist. The ability not just to just bet on the price of compute but to actually settle the trade with delivery. Figuring out what a next generation scheduler that can deal with HTC levels of throughput but blend it with topological awareness for AI inference as part of the workload. Transforming switching schedulers from a multi-million pound multi-year change programme to a deployment and the flick of a switch. You know, simple stuff 😆”

I think it’s better. Perhaps you disagree.

Let me give you another example. I like taking photos. I seemed to have not taken many lately and the last real camera I bought was in 2012.  A rather large professional grade DSLR and I had been happily using it since then. The mirrorless revolution had largely passed me by. Using ChatGPT and/or Claude to quickly research cameras, aggregate real user feedback (and not just shill reviews) was actually pretty great. Though I still ended up reading far too many reddit threads. You know what sucked though? The little bots ability to actually recommend something.

Despite having a huge amount of context in a back and forth that spanned months, without fail the recommendations were always close to what most “The Best Camera” web searches would produce. Which in turn was mostly human and AI generated sponsored slop lists with little to no substance.

I wanted a “proper” camera, with manual controls, a decent sized sensor and enjoyable to use. But it had to be cheap enough to not worry about throwing into bags and taking down mountain bike trails, snowboarding and using on the beach. 

The AI recommendation? Olympus Tough TG-4. Fair enough. Except I’d hate using that. It doesn’t meet any of my requirements. Even with further prompting and having read, by this point, literally thousands of Reddit threads exploring cameras that align to I like in a camera, it was incapable of recommending something that actually fit what I wanted. 

Because what I wanted was out of the ordinary. What I wanted was bit left field. An old, but therefore cheap enough to not care, Sony NEX5. Or a discontinued series of camera’s by Nikon, the 1. The AW1 more specifically as it is waterproof and impact resistant. Not much chance of AI coming up with either though.

This pattern repeated with multiple use cases, so it wasn’t a one off either.

Don’t trust the genie to be creative for you. But it might spark an idea in your own generative intelligence neurones.

Which brings me to my last point. I don’t care if you’re coding, writing LinkedIn posts or composing symphonies. You need to train that creative muscle. Heck even actual body builders understand this concept. It’s called “greasing the groove”. The act of repeatedly exercising certain parts of your brain leads to strengthened neural pathways. It doesn’t matter if you’re doing pull ups or writing HPC code. If you’re not actively exercising your brain to do it, someone else who is, will take your job. 😁

Just a quick note to welcome our latest HPC engineer, Eduardo to the team.

Welcome Eduardo!

That’s all folks! 

Eh. You’re still here? Ok fine. Eduardo’s made the mistake of letting himself be duped into solving the kinds of problems in supercomputing that everyone else would rather pretend don’t exist. The ability not just to just bet on the price of compute but to actually settle the trade with delivery. Figuring out what a next generation scheduler that can deal with HTC levels of throughput but blend it with topological awareness for AI inference as part of the workload. Transforming switching schedulers from a multi-million pound multi-year change programme to a deployment and the flick of a switch. You know, simple stuff 😆

Hopefully he doesn’t run away now that he knows what he’s in for.

And no, the photo is NOT AI. I took it! I’m looking at you Omar 😡

Quite an interesting week for supercomputing news, albeit a little AMD dominant though of course it’s hard to talk about anything GPU these days without Nvidia being mentioned in the same breath.

Remember Talaas? They had a fair bit of fanfare earlier this year with AI inference token throughput rates that were off the charts. AMD just bought them. So now Nvidia has Groq, AMD has Talaas. As I’ve said in the past, when a use case becomes sufficiently prevalent it ends up being baked into the silicon. LLM inference will be no different and that’s across consumer, edge and data centre. Or at least I think so.

When we released our scheduler selection tool last year, I didn’t think we’d need to make many changes. I mean, the HPC scheduler market seemed pretty mature, and companies weren’t exactly falling over themselves to create new ones now were they. Yea how wrong was I. AMD just dropped one too. ROCm Spur. Obviously, its targeting GPU scheduling as the primary use case, is Slurm compatible and of course written in Rust to keep up with the cool kids. Having said all that, I’m not sure what meaningful addition this brings to the table. Guess we need to update the scheduler selector tool again!

Want to do something with your GPUs that isn’t just 4 bit cat picture generation and needs real precision. AMD has always been supportive of this use case with Nvidia pushing emulation using the Ozaki scheme instead and still hoping to win the top trumps game. Looks like AMD isn’t messing about. Their native hardware FP64 support in MI430X GPUs weighs in faster than Nvidia’s Rubin GPUs even including Ozaki emulation. Crikey. I do wonder how much longer that CUDA moat will hold out.

The scientific community seems to agree with the new DOE supercomputers being AMD based.

Let’s talk CPUs for a minute. Nividia’s new Vera chips seem pretty great, just not quite as great as their marketing material would have you believe. I’d quite like to get my hands on one to see how it stacks up for financial risk analytics workloads rather than the nonsense of Agentic AI that people seem to think needs an insane amount of compute.

All the News in Depth

Cloud and vendor releases for Supercomputing & HPC (and AI too if you change the filters)

Vera is a pretty good CPU; shame they are marketing it as though it isn’t and resorting to fairly petty things.

AMD buys Talaas the AI inference ASIC company. Reuters give you the quick take but if you want meandering tale the ends with surprisingly little detail about Talaas but gives you lots of context then the Next Platform article might be more your speed.

AMD is also rattling Nvidia for real FP64 performance. It was never a question if you needed real non emulated FP64 but even if you can accept Ozaki emulated FP64 it looks like AMD still comes out ahead!

AMD has even dropped their own, Slurm compatible, scheduler. Honestly I’m not sure who this is targeted at or how it moves the game forward if at all.

HMx Labs Updates

We’re starting to look at what version 2 of the COREx benchmark should look like. Want to provide some input?

Benchmarks Should Represent Your Workload: COREX v2

This is your reminder that a benchmark is pointless if it does not represent your own workload. Period.

Flux by HMx LabsHamza

We’d also like to know which venue you prefer for HPC Club

HPC Club: Which Venue Did You Prefer?

What’s more important to you, a casual setting for the conversations or a better environment for the presentation?

Flux by HMx LabsHamza

Off Topic

Most open source projects you’ve heard of probably aren’t developed by lone developers tolling away under the midnight oil out of the goodness of their hear (though for sure that exists). No most of them are probably paid handsomely by big tech. Ever wondered why? This is a pretty solid take which I mostly agree with.

This was quite a good watch.. and I certainly agree that if you take away the noise then there really has been very little of note in the AI space lately.

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

Some of the big chip manufacturers are making lots of noise about how great their silicon is on various benchmarks and attempting to bludgeon you into belief with chart crime and long words. Fortunately, we have some actual brainiacs calling them out on it too (not me).

So, this is just your friendly reminder that benchmarks are meaningless unless they genuinely reflect your own workload. It really is as simple as that.

Which brings me to COREx, our little financial risk analytics benchmark. The same absolutely applies to this too. Which is why we’re working on version 2. 

Financial risk analytics is not a homogenous blob. When I first created COREx, the next best thing was running the unit tests for QuantLib and this was certainly an improvement. 

But we can do better. 

Specifically, I feel the difference between real time risk and end of day risk needs to be better captured as does the difference between market risk and portfolio risk.

That’s just me though, so before we run headlong into this, I’d love to get some input from you too. 

I have a question for the HPC Club regulars amongst you. 

Especially if you came to the last event at The Crypt and have been to previous events at Notes Café. Simple question. Which venue did you prefer?

We broke with tradition a little for the last event to allow us to have a remote presenter (I hope you’ll agree it was worth it), but if we stick to our general rule of in person only for the next event, where would you prefer we hold it?

Personally, I think The Crypt was amazing for the presentation. Having decent audio, a big screen and seating setup for a presentation made a huge difference. But I also like the really casual nature and easy conversations that happen when we hold it at Notes.

Which did you prefer?

AI infrastructure made lots of noise, mostly I guess because its earnings season, IBM and partners claimed quantum advantage and the EU wants to join the AI party.

Let’s talk quantum advantage first though because I’m sure you had enough other outlets pushing AI news down your throat last week. IBM and partners, announced quantum advantage for several use cases with IBM’s 156 qubit Heron. The claims here are fairly impressive with validation of the quantum compute using other trapped ion systems from Quantinuum and the classical compute comparison done using Fugaku which held the number one spot on the Top500 in 2020 and is still ranked at number 9.

HPC Wire seem to concur that we’re close to utility, if not there already, for quantum computing too.

I’d love to provide some kind of independent commentary on the above, but this is really hard to do. Unlike, say AI, where it’s pretty easy to validate its capabilities for yourself. the same cannot be said for quantum computing. Even if you had access to a real quantum computer. The amount of work, and level of both domain knowledge and math required simply places it out of reach for most people. I would like to answer this question though for a use case that we work with extensively. Financial risk analytics. So, we’re working on that, stay tuned for more.

Its earnings season on Wall Street and of course because the biggest companies work in AI infrastructure which is the most super of supercomputing you’re probably impacted if you have to purchase any kind of compute let alone a new supercomputer. 

My condolences if you’re in that position. 

The news is not easy to decipher. On the one hand, we have worsening credit spreads on corporate bonds for AI infrastructure companies and Oracle being downgraded to just above junk bond status. On the other hand, we have stellar earnings and increased investment into AI infrastructure from the hyperscalers.

In the midst of this, the EU has decided to enter the fray with EuroHPC putting out RFPs for AI gigafactories, or in normal parlance, new supercomputers.

Have a good week all.

All the News in Depth

Cloud and vendor releases for Supercomputing & HPC (and AI too if you change the filters)

IBM and partners claim quantum advantage

HPC Wire also seems to think real use of quantum may be closer than we think

Intel is increasingly looking like its back from its near death experience even if its CPUs don’t quite hold up to its x86 competitors

CoreWeave financing looks increasingly shaky

Oracle downgraded to BBB rating

More circular financing? Nvidia backs OpenAI’s to the tune of $250B

And the EU wans in on the action too

HMx Labs Updates

Is finding developers who understand memory management getting too hard? I’m not just asking as a theoretical conversation starter but because we need people who can and there are still lots of systems in financial services that need these skills

HPC Software Engineer Wanted

We’re still looking for a software engineer who can write code that manages its own memory… Know anyone?

Flux by HMx LabsHamza

Want to accelerate your HPC network (people you know rather than packets on the wire)? For less than the price of a sushi dinner? Read this

Want to Learn HPC and Build an Amazing Network?

The HPC Leadersip Institute, 2026 Edition is taking place at the Barcelona Supercomputing Centre again this year in September.

Flux by HMx LabsHamza

Off Topic

I like Alberta, she’s funny. Makes a valid point here too, that I’ve also made in the past is a much less funny way

The whole AI is a scary cyber hacker story from the last week or so is getting out of hand, here’s a good sane take on it all from Cal Newport. Search for it on your podcast app of choice or put up with the ads on YouTube. “The Real Story Behind OpenAI’s Rogue Model”

Codeberg approves the change to ban AI Slop code. I’m wondering if we need to move our private repos off GitHub at least and wondering about Codeberg or internally hosting Forgejo.

Lastly, the family and I are hoping to get away to the alps again for a spot of downhill biking so I’ve been watching the incredibly talented Danny MacAskill with them.

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