frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

Running a One Trillion-Parameter LLM Locally on AMD Ryzen AI Max+ Cluster

https://www.amd.com/en/developer/resources/technical-articles/2026/how-to-run-a-one-trillion-parameter-llm-locally-an-amd.html
24•mindcrime•1h ago

Comments

verdverm•1h ago
The setup was around $10k, but maybe more now with mem/ssd prices.

This is a good list, I like my Beelink a lot, my Minisforum likes to turn itself off every couple of weeks, not sure why yet.

https://www.techradar.com/pro/there-are-15-amd-ryzen-ai-max-...

---

Performance is pretty bad (<10/tps) and context is quite limited. Still good to see progress

Prompt Size (tokens) | TFT (s) - Flash Attention Disabled | TFT (s) - Flash Attention Enabled

4096 | 53.7s | 39.7s

8192 | Out Of Memory (OOM) | 90.5s

16384 | Out Of Memory (OOM) | 239.1s

ibeckermayer•1h ago
Cool that it's possible but basically unusable performance characteristics. For an 8192 token prompt they report a ~1.5 minute time-to-first-token and then 8.30tk/s from there. For context ChatGPT is typically <<1s ttft and ~50tk/s.
elcritch•40m ago
That’s pretty awesome!

Though only 5gig Ethernet? Can’t they do usb-c / thunderbolt 40 Gb/s connections like Macs?

burnt-resistor•21m ago
Framework has gone fully in the tank of Apple consumerization route of unrepairability and unupgradeability with a nonstandard machine, soldered-on RAM, and no meaningful PCIe slots. There's only the superficial appearance of longevity and future-proofness when it's really yet another silo. There's no way to add an IB, FC, or 100/400 GbE NICs to these machines. 5 GbE is a joke. Non-ECC RAM is a joke.
tills13•15m ago
I set up ollama today and can barely run a 3b parameter model before the lag makes it unbearable.

How much is one of these gonna run me?

A State Government Tried to Regulate Linux; It Went How You'd Expect

https://www.youtube.com/watch?v=mQLdDR-hJpc
1•cable2600•1m ago•0 comments

I built AI agents that do the grunt work solo founders hate

2•Seleci•8m ago•0 comments

TorchLean: Formalizing Neural Networks in Lean

https://leandojo.org/torchlean.html
2•matt_d•8m ago•0 comments

Hackers Expose the Surveillance Stack Hiding Inside "Age Verification"

https://www.techdirt.com/2026/02/25/hackers-expose-the-massive-surveillance-stack-hiding-inside-y...
1•nobody9999•10m ago•1 comments

Japanese firm Space One plans to launch Kairos No.3 rocket on Sunday

https://www3.nhk.or.jp/nhkworld/en/news/20260301_01/
2•HardwareLust•12m ago•1 comments

Show HN: Sailor.ai – source-backed personalized outbound emails

https://trysailor.ai/
1•bill_waybird•12m ago•1 comments

Show HN: Brand Analytics for AI Search Engines (Beta)

https://explore.somantra.ai/dashboard/141d19d6-1ee7-4a25-81cf-411e6792e286/Australia
1•prasaar•14m ago•0 comments

Show HN: Parallax – Ansible Without Python

https://parallax.digitalxero.dev/
1•DjGilcrease•15m ago•0 comments

Skills.sh Ecosystem Dashboard

https://skills-dashboard.olshansky.info/
1•Olshansky•18m ago•0 comments

Show HN: A visual sitemap editor that forces you to design structure before UI

3•epic_ai•23m ago•2 comments

Show HN: Memctl v0.1.0 Open source shared persistent memory for AI coding agents

https://memctl.com
3•meszmate•24m ago•0 comments

HeadElf-Mvidia: Executive Intelligence Template

https://github.com/pauljbernard/HeadElf-MVIDIA
3•paulbernard•26m ago•2 comments

Agents are not thinking: Science of agent behavior

https://technoyoda.github.io/agent-science.html
3•chse_cake•30m ago•0 comments

Sam Altman Answers Questions on X.com About Pentagon Deal, Threats to Anthropic

https://news.slashdot.org/story/26/03/01/0233230/sam-altman-answers-questions-on-xcom-about-penta...
1•MilnerRoute•31m ago•0 comments

Church of the SubGenius

https://en.wikipedia.org/wiki/Church_of_the_SubGenius
1•thomassmith65•32m ago•0 comments

Show HN: MCP server that strips injection vectors from LLM input

https://github.com/timstarkk/mcp-safe-fetch
1•timstark•33m ago•0 comments

Shipping Traffic Through Strait of Hormuz Plummets After Attacks on Iran

https://www.nytimes.com/2026/02/28/world/middleeast/strait-of-hormuz-ship-traffic.html
1•ParentiSoundSys•36m ago•0 comments

Show HN: ChoresMates – Splitwise, but for Household Chores

https://apps.apple.com/us/app/choresmates/id6757452488
1•bittujoju•39m ago•0 comments

Iran's Supreme Leader Ali Khamenei Killed

https://www.reuters.com/world/middle-east/irans-supreme-leader-ali-khamenei-killed-senior-israeli...
7•codethief•46m ago•2 comments

Ask HN: What did you find out or explore today?

4•blahaj•55m ago•4 comments

Ercot Max Solar Record 31 GW

https://www.gridstatus.io/records/ercot?record=Maximum%20Solar
1•chris222•56m ago•1 comments

Dating Apps, Data Structures, and Dopamine

https://www.errorcodezero.dev/blog/dating-apps-dsa-and-dopamine/
3•errorcodezero•56m ago•1 comments

The Science of Detecting LLM-Generated Text

https://dl.acm.org/doi/10.1145/3624725
1•vinhnx•1h ago•0 comments

Niche Developer Tooling for WordPress

https://coderjerk.com/blog/icenberg
1•ddevine•1h ago•0 comments

In puzzling outbreak, officials look to cold beer, gross ice, and ChatGPT

https://arstechnica.com/health/2026/02/did-chatgpt-help-health-officials-solve-a-weird-outbreak-m...
1•Bender•1h ago•0 comments

The Double Standard of Carbon: Why we grant souls to meat but not silicon

https://natansessays.com/posts/the-myth-of-carbon-narcissism/
1•JhonOliver•1h ago•4 comments

History Rhymes: Large Language Models Off to a Bad Start?

https://michaeljburry.substack.com/p/history-rhymes-large-language-models
1•drob518•1h ago•0 comments

What Was Software Programmer Contribution in the Human Technology Timeline?

https://medium.com/@ggonweb/what-was-the-software-programmer-generations-contribution-in-the-huma...
1•ggonweb•1h ago•1 comments

Stem cells provide a potent treatment for frailty

https://www.nature.com/articles/d41586-026-00584-y
1•bilsbie•1h ago•0 comments

Strike in the Middle East use Anthropic even after the Trump ban

https://www.wsj.com/livecoverage/iran-strikes-2026/card/u-s-strikes-in-middle-east-use-anthropic-...
3•johncole•1h ago•0 comments