frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

Open in hackernews

NUMA Is the New Network: Reshaping Per-Socket Microservice Placement

https://codemia.io/blog/path/NUMA-Is-the-New-Network-How-Per-Socket-Memory-Models-Are-Reshaping-Microservice-Placement
12•signa11•1h ago

Comments

stego-tech•25m ago
Solid writeup of NUMA, scheduling, and the need for pinning for folks who don’t spend a lot of time in the IT side of things (where we, unfortunately, have been wrangling with this for over a decade). The long and short of it is that if you’re building a HPC application, or are sensitive to throughput and latency on your cutting-edge/high-traffic system design, then you need to manually pin your workloads for optimal performance.

One thing the writeup didn’t seem to get into is the lack of scalability of this approach (manual pinning). As core counts and chiplets continue to explode, we still need better ways of scaling manual pinning or building more NUMA-aware OSes/applications that can auto-schedule with minimal penalties. Don’t get me wrong, it’s a lot better than ye olden days of dual core, multi-socket servers and stern warnings against fussing with NUMA schedulers from vendors if you wanted to preserve basic functionality, but it’s not a solved problem just yet.

jasonjayr•22m ago
This strikes me as something that Kubernetes could handle if it could support it. You can use affinity to ensure workloads stay together on the same machines, if K8s was NUMA aware, you could extend that affinity/anti-affinity mechanism down to the core/socket level.

EDIT: aaaand ... I commented before reading the article, which describes this very mechanism.

colechristensen•20m ago
This is one of those way down the road optimizations for folks in fairly rare scale situations in fairly rare tight loops.

Most of us are in the realm of the lowest hanging fruit being database queries that could be 100x faster and functions being called a million times a day that only need to be called twice.

stego-tech•9m ago
100% with you there. I can count one time in my entire 15 years where I had to pin a production workload for performance, and it was Hyperion.

In 99% of use cases, there’s other, easier optimizations to be had. You’ll know if you’re in the 1% workload pinning is advantageous to.

For everyone else, it’s an excellent explainer why most guides and documentation will sternly warn you against fussing with the NUMA scheduler.

jauntywundrkind•11m ago
There's a constant drum-beat of NUMA related work going by if you follow phoronix.com .

https://www.phoronix.com/news/Linux-6.17-NUMA-Locality-Rando... https://www.phoronix.com/news/Linux-6.13-Sched_Ext https://www.phoronix.com/news/DAMON-Self-Tuned-Memory-Tierin... https://www.phoronix.com/news/Linux-6.14-FUSE

There's some big work I'm missing thats more recent too, again about allocating & scheduling IIRC. Still trying to find it. The third link is in DAMON, which is trying to do a lot to optimize; good thread to tug more on!

I have this pocket belief that eventually we might see post NUMA post coherency architectures, where even a single chip acts more like multiple independent clusters, that use something more like networking (CXL or UltraEthernet or something) to allow RDMA, but without coherency.

Even today, the title here is woefully under-describing the problem. A Epyc chip is actually multiple different compute die, each with their own NUMA zone and their own L3 and other caches. For now yes each socket's memory is all via a single IO die & semi uniform, but whether that holds is in question, and even today, the multiple NUMA zones on one socket already require careful tuning for efficient workload processing.

Customizing Lisp REPLs

https://aartaka.me/customize-repl.html
1•todsacerdoti•1m ago•0 comments

Decimal FP faster than binary?

1•newbie-02•8m ago•0 comments

PyCrucible – Another PyInstaller Alternative

https://github.com/razorblade23/PyCrucible
1•razorblade23•11m ago•0 comments

We built ToolVerse: An AI Tools Discovery Platform with 500 curated apps

https://toolverse.com
1•yuanchuangAI•11m ago•1 comments

Serverless Infrastructure for AI apps – 3x perf of baseten, 1/5 the cost

https://www.hyperpodai.com
1•ollayf•12m ago•0 comments

Show HN: Fast360 – A web tool to benchmark open-source OCR models side-by-side

https://fast360.xyz
1•yanaimngvov•17m ago•1 comments

How not to check or poll URLs, as illustrated by Fediverse software

https://utcc.utoronto.ca/~cks/space/blog/web/HowNotToCheckURLs
2•zdw•27m ago•0 comments

How CATL Made Sodium-Ion Batteries 90% Cheaper

https://www.youtube.com/watch?v=Wf84NJSiAeU
2•joak•28m ago•0 comments

If AI takes most of our jobs, money as we know it will be over. What then?

https://theconversation.com/if-ai-takes-most-of-our-jobs-money-as-we-know-it-will-be-over-what-then-262338
2•Improvement•33m ago•0 comments

Injecting doubts in the CoT of reasoning models

https://github.com/martianlantern/cot-doubt-injection
1•martianlantern•34m ago•0 comments

If you love it, download it

https://erysdren.me/blog/2025-08-16/
2•Improvement•36m ago•0 comments

What If Every Speed Limit Suddenly Changed? [video]

https://www.youtube.com/watch?v=bVpzCEiT4oo
1•Klaster_1•37m ago•0 comments

What is mathematics? A classification based on universals

https://ebellani.github.io/blog/2025/what-is-mathematics-a-classification-based-on-universals/
1•b-man•38m ago•0 comments

Wplace.ai - A living social globe — meet fascinating people anywhere on Earth.

https://wplace.ai/
1•easytube•40m ago•0 comments

Ruby gems still broken after 15 years

https://felipec.wordpress.com/2025/08/17/ruby-gems-still-broken/
3•booleandilemma•41m ago•0 comments

Google admits anti-competitive conduct involving Google Search in Australia

https://www.accc.gov.au/media-release/google-admits-anti-competitive-conduct-involving-google-search-in-australia
20•Improvement•42m ago•3 comments

Show HN: ASCII Tree Editor

https://asciitree.reorx.com/
2•novoreorx•44m ago•0 comments

Vannevar Bush: As We May Think (1945) [pdf]

https://worrydream.com/refs/Bush%20-%20As%20We%20May%20Think%20(Life%20Magazine%209-10-1945).pdf
3•g42gregory•44m ago•0 comments

Modelling Exactly-Once Delivery Using TLA+

https://medium.com/@fqaiser94/exactly-once-modelling-using-tla-3c792655be44
2•jelloslate•54m ago•0 comments

Ask HN: RT video chat app with transaltion that works between the US AND China?

1•iatemykindle•1h ago•0 comments

Dedicated volunteer exposes "largest self-promotion operation Wikipedia history"

https://arstechnica.com/culture/2025/08/why-was-the-most-translated-wikipedia-article-in-the-world-about-a-lover-of-aryan-culture/
3•thread_id•1h ago•1 comments

Order vs. Chaos – 6×6 variant of tic-tac-toe

https://www.prabhakargupta.com/paper-games/order-chaos/
2•prabhakar267•1h ago•2 comments

Fk...a hacking tool for vibe coded apps

https://overeasy-landing.codapt.app/
1•bunbunpumpkin•1h ago•0 comments

Evidence for common fungal networks among plants

https://www.nature.com/articles/s42003-025-08432-x
2•PaulHoule•1h ago•0 comments

From Visual UI to High Quality Next.js Code(No Lock-In)

https://nextbunny.co
1•mvsingh•1h ago•0 comments

How to gain more from your reading

https://psyche.co/guides/how-to-gain-more-from-reading-by-taking-it-all-in-more-slowly
1•DocFeind•1h ago•0 comments

Rooted Android phones vulnerable due to Android kernel patching flaws

https://zimperium.com/blog/the-rooting-of-all-evil-security-holes-that-could-compromise-your-mobile-device
3•witnessme•1h ago•1 comments

China's Biotech Is Cheaper and Faster

https://www.nytimes.com/2025/08/17/opinion/china-biotech.html
1•KnuthIsGod•1h ago•0 comments

Text to Handwriting – Free online converter with realistic handwriting effects

https://text-to-handwriting.org
1•TinyMomentum•1h ago•1 comments

Lego Transformers Soundwave

https://www.lego.com/en-us/product/transformers-soundwave-10358
8•wslh•1h ago•2 comments