Bbtree: Fastest Ordered Map(bbtree.xiangpeng.systems)
130 points by aoli-al | 16 comments
Blog post introducing bbtree, a Rust B-tree ordered map claiming 5× speed over std BTreeMap, with benchmarks and integrations into LSM-trees and Turso.
#performance on Hacker News
77 stories tagged performance, newest first.
Faster than earcut: triangulating 3.7M map polygons in 489 ms(globus.software)
213 points by molind | 40 comments
Blog post from Globus Software introducing earstrip, an open-source C++/WebAssembly polygon triangulator, and detailing step-by-step optimizations that beat earcut on map polygons.
Rails Is a JIT Compiler(railsatscale.com)
5 points by hellerve | 0 comments
Rails at Scale blog post on speeding up Rails enum methods by generating code with module_eval instead of define_method and public_send, with YJIT/ZJIT benchmarks.
C vs. C#: racing VLC's Gaussian blur(martinfinkel.com)
118 points by martz | 40 comments
Blog post benchmarking VLC's Gaussian blur filter in C against a C# port via .NET Native AOT, across line-by-line, algorithmically optimized, and AVX2 SIMD versions.
One Hash Bloom Filters(calabro.io)
125 points by jcalabro | 13 comments
Blog post by Jim Calabro explaining bloom filters and a technique for implementing them using a single hash function, with context from Bluesky's backend.
Why Is Speculative Decoding Fast?(jbarrow.ai)
41 points by gmays | 5 comments
Blog post by Joe Barrow explaining that speculative decoding speeds up LLM inference by changing the kind of GPU work rather than reducing FLOPs.
Maxwell: A new kind of model that sits between code and hardware(heymaxwell.ai)
112 points by laginimaineb | 49 comments
Landing page for Maxwell by doubleAI, a 'Code Execution Model' that optimizes running code for hardware and data with formally proven equivalence, offering early access.
Steam Client is now faster than Epic's new vibe coded launcher(respawnfirst.com)
290 points by starkparker | 292 comments
RespawnFirst gaming news article claiming the Steam client now outperforms Epic Games' new launcher, which it describes as vibe coded.
Vectorized CLZ and CTZ(purplesyringa.moe)
47 points by zX41ZdbW | 5 comments
Blog post on purplesyringa explaining how to implement vectorized count-leading-zeros and count-trailing-zeros using floating-point tricks.
Safe SIMD in Rust, even on the inside(shnatsel.github.io)
54 points by ravenical | 3 comments
Blog post by Sergey Davidoff on making SIMD abstractions in Rust safe internally, avoiding unsafe raw intrinsics.
Scaling and benchmarking a critical message bus using a new indexing strategy(blog.janestreet.com)
45 points by eatonphil | 3 comments
Jane Street blog post on scaling and benchmarking a critical internal message bus using a new indexing strategy.
How Fast is Python 3.15?(blog.miguelgrinberg.com)
59 points by Qem | 39 comments
Miguel Grinberg's blog post benchmarking Python 3.15 (rc3) performance against earlier interpreters back to Python 3.10.
Benchmark in Milliseconds(matklad.github.io)
143 points by surprisetalk | 47 comments
A blog post by matklad arguing that micro benchmarks should be sized to run around 300 milliseconds, explaining the reasons for this rule of thumb.
A 40ms Go garbage collector pause caused by swap(frn.sh)
91 points by shellpipe | 78 comments
A blog post explaining how running swap in production caused a 40ms Go garbage collector pause due to cgroup-level page eviction.
Gallery of Processor Cache Effects (2010)(igoro.com)
42 points by porridgeraisin | 5 comments
Igor Ostrovsky's 2010 blog post demonstrating processor cache effects with code examples; the capture shows a hosting 'Resource Limit Is Reached' error.
Why don't more developers “use the platform”?(nolanlawson.com)
304 points by vinhnx | 318 comments
Nolan Lawson blog post examining why web developers often don't "use the platform" and rely on JavaScript instead of built-in browser features.
ParadeDB Search Performance Improvements(paradedb.com)
54 points by craigkerstiens | 9 comments
ParadeDB blog post describing search performance improvements in its Postgres-based search extension.
Eight Bytes Are a Number(blog.sebastiansastre.co)
6 points by sebastianconcpt | 2 comments
Blog post by Sebastian Sastre arguing that reading numbers or types from bytes shouldn't require allocating a separate buffer, since eight bytes already are a number.
Beating the compiler (2024)(mattkeeter.com)
57 points by andsoitis | 38 comments
Matt Keeter's blog post on writing an interpreter for the Uxn stack-based CPU in hand-written assembly to outperform compiler-generated code.
The state of SIMD in Rust in 2026(shnatsel.github.io)
3 points by birdculture | 0 comments
Blog post by Sergey Davidoff surveying the state of SIMD libraries and support in Rust as of 2026, including Fearless SIMD, std::simd, wide, pulp, and macerator.
![Screenshot of youtube.com: Functional Mechanical Sympathy [video]](https://hn.urlbox.com/shots/renders/2026/9/28/036b20e0-4ba6-4b11-80fd-623756e40df3.jpg)
Functional Mechanical Sympathy [video](youtube.com)
60 points by surprisetalk | 11 comments
YouTube video of Richard Feldman's FUNARCH'26 keynote 'Functional Mechanical Sympathy', posted by ACM SIGPLAN.
Writing Efficient C++ Code (2013)(asawicki.info)
164 points by ibobev | 127 comments
Adam Sawicki's article, originally published in Programista magazine in 2013, on practices for writing efficient, high-performance C++ code.
The state of SIMD in Rust in 2026(shnatsel.github.io)
200 points by verdagon | 51 comments
Blog post by Sergey Davidoff surveying the state of SIMD libraries in Rust in 2026, including Fearless SIMD, std::simd, wide, pulp and macerator.
Platform-independent SIMD in Go(go.dev)
382 points by yurivish | 138 comments
Go Blog post describing experimental SIMD APIs in Go 1.26 and 1.27 for amd64, arm64 NEON, and wasm, and work toward platform-independent SIMD.
Making Tailscale Faster(tailscale.com)
209 points by yarapavan | 81 comments
Tailscale product blog post announcing work to make Tailscale faster, by Kabir Sikand and Kevin Purdy.
Writing Rust code that's fast by asking agents to make the code faster(minimaxir.com)
85 points by mooreds | 46 comments
Max Woolf blog post on using AI coding agents to iteratively optimize Rust code until it outperforms state-of-the-art libraries.
Fearless SIMD v1.0(linebender.org)
307 points by verdagon | 48 comments
Linebender blog post announcing the v1.0 release of fearless_simd, a Rust library for safe portable SIMD, and reflecting on its goals and design.
Performance Improvements in .NET 11(devblogs.microsoft.com)
330 points by soheilpro | 100 comments
Stephen Toub's annual Microsoft .NET Blog post detailing the performance improvements in .NET 11 across the runtime and libraries.
GigaToken: ~1000x faster Language model tokenization(github.com)
585 points by syrusakbary | 116 comments
GitHub repository for gigatoken, an open-source library for fast language model tokenization at GB/s speeds, MIT licensed.
Microsoft new Outlook takes 10 seconds to do what Outlook Classic does instantly(windowslatest.com)
603 points by Adam-Hincu | 395 comments
Windows Latest article reporting that Microsoft's new Outlook for Windows is far slower than Outlook Classic, taking about 10 seconds for actions like opening mail from notifications.
The gold standard of optimization: A look under the hood of RollerCoaster Tycoon(larstofus.com)
557 points by mariuz | 153 comments
Blog post examining the technical optimizations behind RollerCoaster Tycoon (1999), based on the author's appearance on the German gaming podcast Stay Forever.
Python 3.15's JIT is now back on track(fidget-spinner.github.io)
463 points by guidoiaquinti | 279 comments
Ken Jin's blog post reporting that CPython 3.15's JIT compiler is back on track, with benchmark charts comparing JIT and interpreter performance.
Python: The Optimization Ladder(cemrehancavdar.com)
322 points by Twirrim | 115 comments
A blog post benchmarking many ways to speed up Python, including Cython, Numba, PyPy, mypyc, Mojo and Rust, on Benchmarks Game problems and a JSON pipeline.
Anthropic Cowork feature creates 10GB VM bundle on macOS without warning(github.com)
335 points by mystcb | 166 comments
A GitHub issue on anthropics/claude-code reporting that the Cowork feature creates a 10GB VM bundle that makes Claude Desktop slow on macOS.
Command-line Tools can be 235x Faster than your Hadoop Cluster (2014)(adamdrake.com)
373 points by tosh | 251 comments
A 2014 blog post by Adam Drake showing that shell pipelines on a laptop processed 1.75GB of chess game data about 235x faster than a Hadoop EMR cluster.
A 40-line fix eliminated a 400x performance gap(questdb.com)
348 points by bluestreak | 75 comments
QuestDB engineering blog post explaining how a 40-line OpenJDK change to getting current thread user CPU time on Linux removed a 400x performance gap.
Can Bundler be as fast as uv?(tenderlovemaking.com)
336 points by ibobev | 96 comments
Aaron Patterson's blog post examining whether Ruby's Bundler can match uv's speed, reviewing uv's techniques and Bundler's bottlenecks.
Python numbers every programmer should know(mkennedy.codes)
382 points by WoodenChair | 157 comments
Michael Kennedy's blog post listing Python performance and memory numbers, such as list operations, file I/O, object sizes, and web framework speed.
How uv got so fast(nesbitt.io)
1179 points by zdw | 405 comments
Blog post by Andrew Nesbitt examining the design decisions and packaging standards that make the uv Python package installer much faster than pip.
Python 3.15’s interpreter for Windows x86-64 should hopefully be 15% faster(fidget-spinner.github.io)
389 points by lumpa | 132 comments
Blog post by CPython developer Ken Jin reporting that the tail-calling interpreter gives ~15% speedup on Windows x86-64 MSVC and ~5% on macOS AArch64 for Python 3.15.
Leaving serverless led to performance improvement and a simplified architecture(unkey.com)
441 points by vednig | 240 comments
Unkey blog post about moving away from serverless, which improved performance and simplified their architecture; the capture shows only a Vercel browser check.
Python 3.14 is here. How fast is it?(blog.miguelgrinberg.com)
695 points by pjmlp | 505 comments
Miguel Grinberg's blog post benchmarking the newly released Python 3.14 against earlier Python versions to measure performance gains.
Behind the scenes of Bun Install(bun.com)
410 points by Bogdanp | 138 comments
Bun blog post by Lydia Hallie explaining how bun install works internally and why it runs much faster than npm, pnpm, and yarn.
How we made JSON.stringify more than twice as fast(v8.dev)
424 points by emschwartz | 146 comments
V8 blog post explaining the engineering optimizations, including a side-effect-free fast path, that made JSON.stringify more than twice as fast.
Push Ifs Up and Fors Down(matklad.github.io)
495 points by goranmoomin | 177 comments
A blog post by matklad giving two rules of thumb: move if conditions up to callers and push for loops down into batch-processing functions, with Rust examples.
The world could run on older hardware if software optimization was a priority(twitter.com)
757 points by turrini | 682 comments
A post on X by John Carmack arguing that the world could run on older hardware if software optimization were prioritized.
Make Ubuntu packages 90% faster by rebuilding them(gist.github.com)
513 points by jeffbee | 323 comments
A GitHub Gist explaining how rebuilding Ubuntu's jq package from the same source with different build options yields about 90% better performance.
Zlib-rs is faster than C(trifectatech.org)
329 points by dochtman | 437 comments
Trifecta Tech Foundation blog post announcing zlib-rs 0.4.2 and benchmarking its decompression and compression speed against C implementations zlib-ng and zlib-chromium.
A 10x Faster TypeScript(devblogs.microsoft.com)
1699 points by DanRosenwasser | 822 comments
Microsoft TypeScript blog post by Anders Hejlsberg announcing a native port of the TypeScript compiler for roughly 10x faster performance.
Performance of the Python 3.14 tail-call interpreter(blog.nelhage.com)
557 points by signa11 | 160 comments
Blog post by Nelson Elhage examining the performance gains of CPython 3.14's new tail-call bytecode interpreter and reassessing its reported 10-15% speedup.
That's not an abstraction, that's a layer of indirection(fhur.me)
546 points by fagnerbrack | 213 comments
A blog post arguing that many so-called abstractions in codebases are merely layers of indirection that add complexity and hurt performance without real value.
FireDucks: Pandas but Faster(hwisnu.bearblog.dev)
378 points by sebg | 162 comments
A blog post introducing FireDucks, a drop-in Pandas replacement claimed to be up to 100x faster, compared with Pandas and Polars from a finance data perspective.
Using Rust in non-Rust servers to improve performance(github.com)
357 points by amatheus | 199 comments
A blog post in the pretzelhammer rust-blog GitHub repo on using Rust via WebAssembly, FFI, or native modules to speed up servers written in other languages.
The Fastest Mutexes(justine.lol)
852 points by jart | 311 comments
Justine Tunney's blog post benchmarking Cosmopolitan Libc's mutex implementation against other platforms' mutexes under heavy thread contention.
Parsing protobuf at 2+GB/s: how I learned to love tail calls in C (2021)(blog.reverberate.org)
329 points by fanf2 | 140 comments
Josh Haberman's blog post on using Clang's musttail attribute for guaranteed tail calls in C to build a protobuf parser exceeding 2GB/s.
SQLite: 35% Faster Than the Filesystem(sqlite.org)
374 points by yla92 | 127 comments
SQLite documentation page explaining that reading and writing small blobs in SQLite is about 35% faster and uses less disk space than individual files.
Show HN: Simulating 20M Particles in JavaScript(dgerrells.com)
489 points by rustystump | 61 comments
Blog post describing how the author simulated 20 million particles in JavaScript using shared array buffers and multithreading, with an interactive demo.
Google Sheets ported its calculation worker from JavaScript to WasmGC(web.dev)
400 points by microflash | 179 comments
A web.dev case study explaining why Google Sheets moved its calculation worker from JavaScript to WasmGC and the performance results.
After 6 years, I'm over GraphQL(bessey.dev)
1005 points by mattbessey | 602 comments
Blog post by Matt Bessey explaining why, after six years of production use, he no longer recommends GraphQL to most people and what alternatives he prefers.
The return of the frame pointers(brendangregg.com)
630 points by mfiguiere | 230 comments
Brendan Gregg blog post on frame pointers returning as a compiler default in Linux distributions, and how this fixes broken stack traces in profilers and flame graphs.
More powerful Go execution traces(go.dev)
340 points by nalgeon | 114 comments
Go Blog post by Michael Knyszek (March 2024) describing improvements to Go's runtime/trace execution tracer and how traces help diagnose program behavior.
I accidentally Blender VSE(aras-p.info)
483 points by bangonkeyboard | 91 comments
Aras Pranckevičius blogs about how he started contributing code and performance improvements to Blender's Video Sequence Editor.
Don't pass structs bigger than 16 bytes on AMD64(gist.github.com)
340 points by todsacerdoti | 114 comments
A GitHub Gist write-up explaining how avoiding passing structs larger than 16 bytes on AMD64 sped up the Neat compiler by about 2x.
The One Billion Row Challenge(morling.dev)
424 points by madmax108 | 271 comments
Gunnar Morling's blog post announcing the One Billion Row Challenge, a contest to aggregate one billion rows of weather station data as fast as possible in Java.
Rust std fs slower than Python? No, it's hardware(xuanwo.io)
640 points by Pop_- | 209 comments
Blog post investigating why Rust's std::fs file reads appeared slower than Python's, tracing the cause to a hardware/CPU-level issue rather than the language.
Push ifs up and fors down(matklad.github.io)
606 points by celeritascelery | 274 comments
A blog post by matklad giving two rules of thumb: move if conditions up to callers and push for loops down into batch-processing functions, with Rust examples.
Building a high performance JSON parser(dave.cheney.net)
469 points by davecheney | 158 comments
Dave Cheney's GopherCon Singapore 2023 talk notes on designing an efficient Go package by building a high performance JSON parser.
Software disenchantment(tonsky.me)
422 points by InsiderTesla | 286 comments
Nikita Prokopov's blog essay lamenting slow, bloated, inefficient modern software and the industry's tolerance of poor quality.
{n} times faster than C(owen.cafe)
381 points by 414owen | 197 comments
Blog post that takes a clang-compiled C loop and hand-tweaks the x86 assembly in various ways, measuring the speedup of each change.
Data-Oriented Design (2018)(dataorienteddesign.com)
351 points by DeathArrow | 123 comments
Free online reduced edition of Richard Fabian's book Data-Oriented Design, showing the table of contents and an introductory note.
Windows NT on 600MHz machine opens apps instantly. What happened?(twitter.com)
559 points by dceddia | 509 comments
Tweet by Julio Merino showing a video of a ~2000-era 600MHz PC running Windows NT 3.51 and opening apps very quickly.
A single line of code made a 24-core server slower than a laptop (2021)(pkolaczk.github.io)
456 points by xk3 | 138 comments
Blog post by Piotr Kołaczkowski explaining how one line of code caused a parallel program to run slower on a 24-core server than on a 4-core laptop.
Our Plan for Python 3.13(github.com)
477 points by bratao | 333 comments
README in the faster-cpython ideas repo on GitHub outlining the Faster CPython team's performance plan for Python 3.13, by Mark Shannon and Michael Droettboom.
FreeBSD spends 7% of its boot time running a bubblesort on its SYSINITs(twitter.com)
331 points by gslin | 314 comments
Tweet by Colin Percival noting that FreeBSD booting in Firecracker spends 7% of its time bubblesorting SYSINITs, prompting a switch to a faster sort.
Beautiful branchless binary search(probablydance.com)
342 points by ingve | 128 comments
Blog post by Malte Skarupke presenting a branchless C++ binary search based on Shar's algorithm that runs over twice as fast as std::lower_bound in GCC.
We achieved a 6-fold increase in Podman startup speed(redhat.com)
349 points by DrinkWater | 165 comments
Red Hat article explaining how cutting unnecessary processes made Podman container startup six times faster for time-sensitive uses like cars.
Microsoft fixes 5-year-old Defender bug, reducing Firefox-related CPU use by 75%(bugzilla.mozilla.org)
979 points by ylere | 373 comments
Mozilla Bugzilla bug 1441918 tracking high CPU usage by Windows Defender's Antimalware Service Executable when using Firefox, now resolved fixed.