frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

Just brute force your embeddings

https://softwaredoug.com/blog/2026/07/29/just-brute-force-embeddings
8•JohnBerryman•17h ago

Comments

emschwartz•1d ago
This works especially well if your embedding model was trained to perform well with quantized embeddings. Binary + hamming distance = incredibly fast.

This post is from 2024 but I wrote about using this technique in https://emschwartz.me/binary-vector-embeddings-are-so-cool/

softwaredoug•1d ago
Hamming w/xor+popcount is the only thing I can make numpy do faster than float32 dot products :)

int8s, float16s are all fairly slow. I suppose it’s because BLAS does float32/64 very fast.

emschwartz•1d ago
Yeah that makes sense. I took the optimizations of my hamming distance library to a bit of an insane level. I wrote this about the first round https://emschwartz.me/unnecessary-optimization-in-rust-hammi... The second round brought the best result down to 0.8 nanoseconds per comparison on x86 with SIMD (for a batch of 1000).

Separately, I’m a huge fan of your writing about search! It’s been very helpful while I’ve been building https://scour.ing.

softwaredoug•1d ago
Thank you! <3
aitchnyu•1d ago
Umm, is pgvector relevant to this usecase?
Ellis_dev•17h ago
The 1M-document numbers are a great reality check. I’d happily start with the NumPy version and only reach for a vector DB once it actually hurts.
firasd•26m ago
Yes! I'm working on an MCP server called Liveclip (not released yet) and we just store embeddings as text in SQLite.. it works over ~40MB of text, 1450 SCOTUS opinions

start=$(date +%s); curl -s -X POST "https://[urlredacted].workers.dev/mcp" -H "Content-Type: application/json" -H "Accept: application/json, text/event-stream" -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"table_search_similar","arguments":{"key":"scotus_2010s","col":"M","query":"endangered fish"}}}' 2>&1 | grep '^data:' | sed 's/^data: //' | jq; end=$(date +%s); echo "elapsed: $((end-start))s" { "result": { "content": [], "structuredContent": { "has_more": true, "headers": { "A": "author_name", "B": "category", "C": "per_curiam", "D": "case_name", "E": "date_filed", "F": "federal_cite_one", "G": "absolute_url", "H": "year_filed", "I": "scdb_id", "J": "scdb_decision_direction", "K": "scdb_votes_majority", "L": "scdb_votes_minority", "M": "text" }, "offset": 0, "ok": true, "rows": [ { "A": "Justice Thomas", "B": "dissenting", "C": "False", "D": "Florida v. Georgia", "E": "2018-06-27", "F": "", "G": "https://www.courtlistener.com/opinion/4511641/florida-v-geor...", "H": "2018", "I": "", "J": "", "K": "", "L": "", "M": "Florida asks this Court to cap Georgia’s use of water in\nthe Apalachicola-Chattahoochee-Flint River Basin (Basin).\nFlorida claims that such a cap would allow additional\nwater to flow into the Apalachicola River and Bay, which\nwould benefit Florida by alleviating certain ecological\nharms. To prevail under our precedents, Florida must\npresent clear and convincing evidence that its proposed\ncap will benefit Florida more than it harms Georgia. See\nColorado v. New Mexico, 459 U.S. 176, 187 (1982) (Colo-\nrado I ). The Special Master applied this balance-of-harms\nstandard and, after presiding over a 1-month trial involv­\ning 40 witnesses and more than 2,000 exhibits, found that\nFlorida had not met its burden. Because that finding\nis well supported by the evidence, I would have over-\nruled Florida’s objections to the Special Master’s Report\n(Report) and denied Florida’s request for relief. I respectfully\ndissent.\n I\n The Court’s recitation of the facts focuses o", "_match": true, "_row": 1189 },

elapsed: 12s (on my laptop)

blt•14m ago
Nothing new under the sun, the same thing has been happening since the "big data" keyword became popular in the 2010s.
behnamoh•7m ago
I 'member this is what Karpathy used in his restaurant menu app, and John Carmack said this is the way. I also agree that embeddings are overkill for the majority of tasks.

Tailscale didn't stop the Hugging Face intrusion

https://tailscale.com/blog/hugging-face-intrusion
257•bluehatbrit•2h ago•104 comments

Elevators

https://john.fun/elevators
724•Jrh0203•6h ago•186 comments

qm

https://github.com/yc-software/qm
302•tosh•3h ago•69 comments

Twenty-five years ago it was cryptography, today it's model weights

https://weeraman.com/because-we-can/
36•aweeraman•3d ago•6 comments

Golang proposal: container/: generic collection types

https://github.com/golang/go/issues/80590
82•jabits•3h ago•40 comments

Severance

https://lcamtuf.substack.com/p/severance
148•surprisetalk•4h ago•39 comments

Progressive Web Components

https://arielsalminen.com/2026/progressive-web-components/
24•hosteur•11h ago•3 comments

Loops (YC W22) Is Hiring a Product Educator

https://www.ycombinator.com/companies/loops/jobs/zqUnwqB-product-educator-technical-content-creator
1•chrisfrantz•1h ago

Demystifying DRAM Read Disturbance: RowHammer and RowPress Phenomena

https://arxiv.org/abs/2607.28233
10•Jimmc414•1h ago•5 comments

Big Food vs. the People

https://www.lighthousereports.com/investigation/big-food-vs-the-people/
164•jruohonen•5h ago•106 comments

June in Servo: real world compat, media queries, SharedWorker, and more

https://servo.org/blog/2026/07/31/june-in-servo/
59•iamnothere•3h ago•16 comments

Getting 25 Gbps Thunderbolt Ethernet on My Mac Studio

https://www.jeffgeerling.com/blog/2026/getting-25g-ethernet-mac-thunderbolt/
105•speckx•5h ago•67 comments

Let's make the worst Htmx

https://zserge.com/posts/worst-htmx-ever/
30•RebelPotato•16h ago•8 comments

DeepSeek V4 Flash 0731 Intelligence, Performance and Price Analysis

https://artificialanalysis.ai/models/deepseek-v4-flash
497•theanonymousone•14h ago•275 comments

DeepSeek-V4-Flash Update

https://api-docs.deepseek.com/updates/
657•dnhkng•15h ago•313 comments

Termixer (TUI DJ Mixer)

https://github.com/l00sed/termixer
37•l00sed•3h ago•27 comments

The most official water costs $120k a gallon

https://signoregalilei.com/2026/07/26/the-most-official-water-costs-120000-a-gallon/
100•surprisetalk•7h ago•81 comments

Authorize, don't authenticate

https://blog.marcua.net/2026/07/31/authorize-dont-authenticate.html
22•marcua•7h ago•4 comments

Using the railway network as a flatbed scanner [video]

https://media.ccc.de/v/emf2026-74-1-using-the-railway-network-as-a-flatbed-scanner
36•Jimmc414•3h ago•14 comments

Run Kimi K3 using 29 GB of RAM at 0.50 tok/s

https://github.com/sqliteai/waste
107•marcobambini•7h ago•44 comments

Everyone is building LLM routers, we deprecated ours

https://manifest.build/blog/why-we-deprecated-our-llm-router/
67•brunaxLorax•3h ago•36 comments

Predictive Speculative KV Replication for Bursty LLM Inference

https://jwlabs.vercel.app/post/biting-the-bullet
9•shreybirmiwal•2h ago•0 comments

Dubious research tied to Red Bull has shaped energy drink policy

https://www.theexamination.org/articles/red-bull-funded-research-energy-drinks-alcohol
82•Jimmc414•6h ago•140 comments

Algorithms on billion-scale graph using 10GB RAM: I love DataFusion

https://semyonsinchenko.github.io/ssinchenko/post/datafusion-graphs-cc-2/
82•speckx•6h ago•30 comments

How JPEG works: Interactively explore JPEG's lossy compression methods

https://cgjennings.ca/articles/jpeg-compression/
75•at1as•4d ago•7 comments

Show HN: How to build and self-host a code review agent

https://www.trytilde.ai/blog/how-to-build-code-review-agent
9•solsol94•1h ago•2 comments

Orca-Bench: How Ready Are Language Model Agents for Oncall?

https://arxiv.org/abs/2607.28545
20•yruzin•3h ago•5 comments

Just brute force your embeddings

https://softwaredoug.com/blog/2026/07/29/just-brute-force-embeddings
8•JohnBerryman•17h ago•9 comments

A simple clustering algorithm for lists

https://cassidoo.co/post/clustering-tiles/
16•nreece•3d ago•2 comments

A GTK4 SSH-askpass in Zig

https://xn--gckvb8fzb.com/a-gtk4-ssh-askpass-in-zig/
58•surprisetalk•6h ago•18 comments