faster

118 stories and discussions about faster, aggregated from every source we track.

1.

Claude Sonnet 5.5 is a clear upgrade over Claude Sonnet 5, runs 30%+ faster, and costs up to 30% less for most work.

879 points•D2OQZG8l5BI1S06•about 2 months ago•609 comments•
2.
466 points•ibobev•2 months ago•190 comments•
3.

Tailscale performance updates reduce memory use, increase throughput, and speed startup with multi-queue, writev, and netmap caching.

244 points•yarapavan•about 2 months ago•106 comments•
4.

See how Netlify rebuilt Edge Functions on MicroVMs to deliver lower latency, greater reliability, and faster log delivery.

228 points•jbott•about 2 months ago•105 comments•
5.

Inside our performance sprint: the benchmarks Claude built, the loop each Slack thread ran, and the guardrails that let us ship 3,000 changes safely.

224 points•matthieu_bl•about 2 months ago•150 comments•
6.
85 points•mooreds•2 months ago•46 comments•
7.

Four changes to the n-gram caches of llama.cpp make drafting up to 41.6x faster, load the static cache up to 23.5x faster, and lower peak memory up to 2.65x.

70 points•pptadversary•about 2 months ago•11 comments•
9.

SHA-1 with collision detection doesn't have to be so slow. How a code generator that moves the checks to SIMD can narrow the gap to plain SHA-1 to ~20–30%.

46 points•srijs•2 months ago•9 comments•
10.

NumPy on Emscripten-forge now links OpenBLAS in WebAssembly, with large speedups for matmul and linear algebra in the browser.

36 points•Matumio•2 months ago•5 comments•
11.

write Python that’s faster than C by optimizing your code, adding standard type annotations, and using Mypyc.

32 points•carlana•over 4 years ago•6 comments
12.

Private, domain-specific benchmarks in legal, tax, and finance.

28 points•leumon•2 months ago•5 comments•
14.

I use AI constantly when I code and when I write. I use it to research ideas, plan code, debug,...

23 points•mikachu•about 2 months ago•6 comments
15.

In short: add your Jira projects as Google Chrome search engines so that you can quickly find issues by number.

21 points•ricardbejarano•over 4 years ago•11 comments
16.

In the previous post, I mentioned that buildcache has some unique properties compared to ccache and sccache. One of them is its Lua plugin system, which lets you write custom wrappers for programs that aren’t compilers…

20 points•fanf•7 months ago•1 comment
18.

I benchmarked Blacksmith against github-hosted ubuntu-latest with 10 repeated trials each on an identical GitHub Actions job. Blacksmith ran 3.17x to 3.77x faster with noticeably tighter variance, but it wouldn't even queue until I moved the repo into an organization.

17 points•alexgeorgiev17•2 months ago•0 comments
19.

Speed up development with this little trick.

16 points•robertkrahn•over 4 years ago•1 comment
21.

graf is graphify, rebuilt properly in rust: 1000x faster search, 89x faster updates, and one native binary. - ctxrs/graf

12 points•ripped_britches•2 months ago•3 comments•
22.

A practical guide to canceling outdated runs, caching dependencies, reducing duplicate work, and getting faster feedback from GitHub Actions.

10 points•johnnylemonny•about 2 months ago•4 comments
23.

For years, cybersecurity had one obvious problem: Finding vulnerabilities was...

10 points•robertadam987_•2 months ago•0 comments
24.

Most of the time, you’ll hear advice that you should select as few columns as possible so that SQL Server can do less work. However, one of my clients hit a neat edge case scenario where SELECT * queries actually…

9 points•pek•almost 7 years ago•0 comments
26.
8 points•bigfishhk•about 2 months ago•9 comments•
28.

I have Claude Code operate macOS apps in the background: fill a form, pick an option from a pop-up,...

7 points•mimo-3•about 2 months ago•1 comment
29.
6 points•ovyan•2 months ago•2 comments•
30.

The engineering choices that turn joins into a strength. A deep dive with real-world case studies.

6 points•eatonphil•10 months ago•0 comments
31.

Benchmark TypeScript 7 safely, compare real project results, and find compatibility problems before changing your main branch.

5 points•johnnylemonny•about 2 months ago•0 comments
32.

Computer Use Agents like Claude in Chrome or ChatGPT relearn every site on every run. With Reduck MCP, they learn a site once, then reuse it for 2.5x faster browsing.

5 points•Labo333•about 2 months ago•1 comment•
33.

Eleven v4 and v4 Turbo: more emotional nuance, faster responses, stronger voice cloning. Live now in ElevenAgents, ElevenCreative and API.

5 points•saretup•about 2 months ago•1 comment•
35.

Codename One removes native startup waits, repeated style scans, and unnecessary JavaScript suspension. Profiles and compiler benchmarks expose costs that bundle size and frame rates miss.

5 points•codenameone•2 months ago•0 comments
36.

NumPy on Emscripten-forge now links OpenBLAS in WebAssembly, with large speedups for matmul and linear algebra in the browser.

5 points•rajtilakjee•2 months ago•0 comments
38.

A few years ago, building with AI usually meant one thing: send some data to a model, get a...

4 points•art_light•about 2 months ago•0 comments
39.

Until now, cryptographers thought factoring was the only way to break RSA. Not anymore.

4 points•jnord•about 2 months ago•0 comments•
40.

Until now, cryptographers thought factoring was the only way to break RSA. Not anymore.

4 points•el_duderino•about 2 months ago•0 comments•
42.

I had to get there eventually. I had a blog post called “I Wrote a Fast Hashtable” and another blog post called “I Wrote a Faster Hashtable.” Now I finally wrote the fastest…

4 points•calvin•over 9 years ago•0 comments
43.

Agents open trivial PRs faster than anyone can read them, so the approvals turn into rubber stamps. Build the guardrails that earn trust, write down what counts as trivial, and let those changes ship without a person…

3 points•bostonaholic•about 2 months ago•0 comments•
44.
3 points•jbredeche•about 2 months ago•2 comments•
45.

ONCE Campfire in Rust: one binary, the Rails app's data, 19–44× faster - basecamp/once-campfire-rust

3 points•ksec•about 2 months ago•0 comments•
46.
3 points•quasiben•about 2 months ago•0 comments•
47.

Agent-loop discourse sells parallel code output. For solo builders the binding constraint is usually demand, and a useful receipt needs cost, artifact, and a user-visible before/after.

3 points•conorcleary•about 2 months ago•0 comments•
48.

New estimates using the planet’s surface wrinkles show that our smallest planet is growing even smaller, at a faster clip than expected.

3 points•reaperducer•2 months ago•3 comments•
49.

I spent a weekend playing with Jev, a model that scores things instead of writing text. It became a code search tool that makes Codex, OpenCode and Claude Code faster, measured over 162 sessions.

3 points•bartlomein•2 months ago•0 comments•
50.

Cotyper: The modern devtool for vibe coders

3 points•sumitsrivastava•2 months ago•0 comments•
51.

Discover how updating non-code files can cause regressions in benchmarks. We investigate how CPU differences, cache architecture, and instruction set extensions affect performance measurements in CI.

3 points•itamarst•9 months ago•2 comments
52.

Transparent, Human-First, AI-Assisted Writing in Academia. TeXposit is the modern collaborative LaTeX, Markdown and LaTeX+Markdown Hybrid editor. TeXposit features an advanced AI writing assistant, built-in reference…

2 points•dszb•about 2 months ago•0 comments•
53.

What Matters Today is CPU and RAM Utilization (and infra costs) tl;dr: As agents take over writing code, the cost of validating generated code becomes a primary determinant of engineering throughput. Optimizing your…

2 points•roflc0ptic•about 2 months ago•0 comments•
55.

Jev is faster and cheaper, and potentially better. The big bet: software needs a model trained to make calibrated decisions, and chat models were trained to please people.

2 points•imaurer•about 2 months ago•3 comments•
56.

Qwik 2.0 RC is out: faster reactivity, smaller HTML and experimental out-of-order streaming.

2 points•maiieul•about 2 months ago•0 comments•
57.

Epoch AI measures how fast the cost of a given level of AI performance is falling across five benchmarks covering math, science and games of skill: about 47% per quarter, or 13x per year, since 2023, faster than…

2 points•giuliomagnifico•about 2 months ago•0 comments•
58.

Vendor loyalty and subscription schemes have limited appeal when prices can fall so fast.

2 points•Brajeshwar•about 2 months ago•0 comments•
59.

Large language models can work faster by dividing problems into sub-problems to be solved in parallel instead of in sequence.

2 points•ent101•about 2 months ago•0 comments•
60.

Two weeks after everyone agreed to pace the frontier, the labs shipped again and my feed filled with agent-escape chatter. I'm confused.

2 points•rmason•about 2 months ago•0 comments•

Related topics