The SC #122
Issue #122 of The SC. Weekly supercomputing news. AMD gives us new Zen6 CPUs and MI4000 series GPUs. And a new software stack to drive them. We also got more details on Vera Rubin and some interesting information on combining AI and quantum computing (for error correction).
Finding the real supercomputing news this week and picking it out of the AI haystack was harder than usual. Which is odd because AMD released its Zen6 CPUs and MI4000 series GPUs.
In a week filled with news about increased spending from OpenAI and Google, the OpenAI and Hugging Face saga and declining OpenAI revenues, what would normally have been a fairly exciting release seems to have been almost lost.
The press release from AMD focuses on its Helios rack scale product but honestly, for those of us that aren’t hyperscalers or neoclouds (in other words most of you reading this) the more interesting part is the chips inside and putting them in your own standard 19” racks not rebuild your entire DC. Unless of course you’re about to replace your supercomputer at a national lab 😉
This is especially true because AMD are still producing GPUs with native FP64 that hasn’t been crippled. You know, stuff for doing real work 😁
I know what you’re thinking though, CUDA. You can’t use AMD. No idea how well this stuff works yet, but they also launched ROCm.ai to try and bridge that gap.
Nvidia being Nvidia though, couldn’t let AMD have the news cycle to themselves and released a few more details about Vera Rubin. We’ve been talking about that for so long, can’t we just get on with it and release it widely please! Maybe no one has any space left to install them (if Ed Zitron is correct).
Not exactly new, but Next Platform has an interesting piece this week too about using AI for quantum error correction. I’d say something snarky about this, but I will hold fire till we complete our literature review of using quantum computing in financial services (specifically for risk analytics). If that goes well, we may have something more interesting to talk about and if it doesn’t then I guess I have more ammunition for my snark cannon. Or more work to do!
In other news, we had our first full week at HMx Towers with Sahana having joined us as our new Strategy and Partnerships lead. We still have an open engineering role though, know anyone that might be interested?
All the News in Depth
Cloud and vendor releases for Supercomputing & HPC (and AI too if you change the filters)
AMD gives us new Zen6 CPUs, MI4000 series GPUs, including some that can do real FP64 work!
And HPC Wire’s take on the release
What about that CUDA moat? Yea AMD’s been working on that…
Using AI for quantum computing error correction
HMx Labs Updates
We finally hired someone for our open strategy and partnerships role!

We still have one open engineering role though

Off Topic
AI spending seems to be accelerating still...
Even Google wants to spend more since it sold more cloud
Even though AI still can't make any money, even with ads
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
