Quantum #102

Issue #102 of the weekly HPC newsletter by HMx Labs. More details and benchmarks on the HPC8a and H4D VMs. The GPU/Accelerator space looks more crowded and time to check your cloud continuity planning.

Quantum #102

Yesterday was International Women’s Day so I think its only right that I start out this week’s HPC newsletter with a quick plug for WHPC.

Not a great deal that I would classify as new this week but we got some more detail and benchmarks on AWS’ HPC8a and Google made some noise about the release of its H4D last week after its rather quiet release the week before.

I know I vaguely promised some benchmarks of our own last week (using COREx our financial risk benchmark) but I’m kind of artificially holding them up because I’d like to share them using our new user interface that sits on top of our automated benchmarking system. It’s been dev complete for over a month now, but I seem to keep failing at actually putting it live due to a lack of time! It’s coming soon, I swear. Along with not only the benchmarks of the two new HPC VMs but the ability to see and compare any cloud VM and some bare metal hardware too. All on your own. No more waiting for me to publish snippets of data.

In other news Broadcom also wants to challenge Nvidia’s dominance in the AI/HPC accelerator space and announced forecasts for $100 billion. I’ll believe that when I see it. But between that, the plethora of inferencing hardware startups, AMD’s GPUs getting a bit more traction and Apple quietly making a play for the on-device consumer space I think it’s far from clear what the future will look like. I really hope that it will be heterogenous as that’s going to be better for market dynamics, but I think we still lack the software foundations to make that work.

Last week also served as your reminder that if you need resilience for your cloud HPC workloads you need to be multi region. If you still think real HPC can’t be done on the cloud let alone on the cloud in multiple regions, give me a shout.


In The News

Updates from the big three clouds on all things HPC.

HPC Cloud Updates WE 08 Mar 2026
Updates to AWS, Azure & GCP in the last week relevant for HPC practitioners. More details on the H4D and HPC8a this week.

Broadcom pushes for more of the data centre accelerator market

Broadcom May Become The Biggest Counterbalance To Nvidia
Here’s a funny thought. If Broadcom had not bought legacy software maker Computer Associates in Nove…

Can your cloud HPC run in multiple regions?

Amazon says it could take at least a day to restore data centers hit by ‘objects’ in the UAE
Amazon Web Services said on Sunday that connectivity from one of its UAE facilities went down after “objects” hit it and sparked a fire.

From HMx Labs

HPC Maturity and AI
HPC capability and maturity is about so much more than size
HAL Update: Cache Control Plane Complete
The control plane is dev complete. Allegedly. But the only thing that matters is production.

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