do you need secure fast compute? | hotaisle.xyz/
- Tweets 23,073
- Following 4,649
- Followers 6,151
- Likes 28,360
We’ve built a platform that lets you provision high-performance $AMD MI300x compute almost instantly, on demand and without contracts. Once you can do that, starting and stopping access to GPUs becomes trivial.
Our CLI is now installable via apt on Ubuntu, which makes it easy to integrate Hot Aisle directly into CI pipelines or agentic workflows.
The flow looks like this:
Install the CLI tool
Start a VM with cloud-init to auto-install your software
Run your workload
Stop the VM
No manual steps. Fully automated. Pay for what you use.
github.com/hotaisle/hotaisle…
9
11
92
229,380
Leave it to TensorWave to try to sign a deal with a company with zero revenue and a $486m loss that is outsourcing all the work to CBRE.
Year to date, FRMI stock has fallen 33%, in a time when everything is massively bullish. ¯\_(ツ)_/¯
stocktwits.com/news-articles…
2
254
"get out of the contract" -- if there is another taker
Why have a contract in the first place?
We're announcing Elastic InfiniBand Partitions in SFC.
You can now get RDMA connected nodes reserved, change plans, and sell your whole contract on the fly. That’s not “sell to spot”, that’s “sell the whole contract”.
Launch the 30-day hero run on Monday, hit the snag on Tuesday, get out of the contract on Wednesday.
Running a PaaS and your customers need their own partitions? Create thousands of partitions, one for each customer. Lots of providers give IB, but all on the same partition, which means it’s not secure enough for production.
sfcompute.com/news/elastic-i…
1
431
More details on the Fermi / TensorWave debacle. The signing of the contract is delayed because TW can't get the backing for it.
Traditionally, $AMD has been the one to back TW's deals, and it looks like that is no longer happening.
Not positive for TW, at all.
financefeeds.com/fermi-tenso…
346
The better solution is to just get yourself removed from their rankings and ignore them entirely.
Kent says $IREN doesn’t currently have active managed clusters. Its contracted Slurm/K8s deployments are still under development.
@SemiAnalysis_ tested a B300 cluster accessed through Together and running at IREN’s Prince George site, presenting that experience like a review of IREN’s managed-cluster product before that product is even live is the issue everyone is arguing about.
1
4
952
The crane is rated to 5 tons. Maximum safe limit would be 3.75 (75%), and they are doing 8,000 lbs, or 4 tons.
Double wide... 8,000 lb rack... lifted onto a vibration test. Hear how our team is testing and validating AMD Helios for real-world scenarios.
🎥 bit.ly/4yZdNLe
3
61
8,535
I’m exploring adding storage in our data center so VM and bare-metal customers can keep their data between runs. The backend is already built, we save VM data to an S3-compatible API. What we’re missing is the physical storage.
Once that’s in place, you could pick up almost immediately where you left off. The experience would feel much like running locally: power off when you’re done, power on when you’re ready, and keep going.
You’d pay for what you use, pay a much smaller amount when you don't need it, without the upfront cost of buying hardware that’s much slower than our supercomputers.
4
1
19
1,230
Everyone is mad that IREN got a sh\t review, and nobody is mad that we were removed without comment.
If IREN had just been quietly removed, that would have been a better outcome than the shitty review. Plus, IREN could have gone out bigger, with calling them grifters, like I did.
That's the hilarious part about all of this.
At the end of the day, the best participation is no participation.
3
20
1,766
PSA to NeoClouds: the smart move is to just get yourself removed from their ratings by incessantly pointing out what grifters they are.
The concept that they are driving customers to or from your business, is wrong. We're full and have hundreds of happy customers, and we're no longer on the list.
Problem solved.
1
3
39
2,256
