61
Once Claude can measure something, it can make it faster
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.
New
The latest blog posts to take off on Hacker News. 6,942 posts.
62
Making Tailscale Faster
Tailscale performance updates reduce memory use, increase throughput, and speed startup with multi-queue, writev, and netmap caching.
63
Data-only attacks are easier than you think (2024)
Data-only attacks, those that do not affect a program’s control flow, have long been considered too sophisticated and niche to pose a practical threat. With our research, however, we have built a tool that automatically generates them with surprising ease. We explain how such attacks work, and why our tool, Einstein…
64
Fearless SIMD v1.0
Fearless SIMD v1.0 is here
65
Software sandboxing: The basics (2025)
Diving into the territory of software sandboxing is diving into mostly uncharted territory. The necessary pieces to implement good sandboxing in your software are scattered all-around and the pioneers haven’t yet gathered enough knowledge into an unified mappa mundi that can guide new sailors through some well…
66
67
How did AMD Ryzen get 50% faster in two years?
People still tell me that CPUs are boring. That nothing much happens anymore. Let us look at AMD Ryzen 7 processors from 2022 to 2024: the 5800X3D (Zen 3), the 7800X3D (Zen 4) and the 9800X3D (Zen 5). They are comparable 8-core chips with 3D V-Cache. I have written about them before, in How stagnant … Continue reading…
68
Saving another 100TB of RAM
Cloudflare's global network is immense but not limitless. As we look for small ways to trim our resource usage, we sometimes get lucky and we can cut significantly more. Here’s how we reduced one of our Pingora-based service's RAM usage with statistics.
69
The scourge of x86 emulation
Welcome to the first feature article on our site. We’re going to cover an ongoing problem with x86 emulation that affects every application that we emulate. This comes down to a single over-arching term that has wide-reaching ramifications; Emulating the x86 Total Store Ordering memory model (x86-TSO) .
70
How Uber Protects Against Retry Storms
What if retries could know when they’re helping, and when they’re making things worse? Uber Engineering built error ownership to prevent and contain retry storms.
71
You can run Git on object storage if you re-make packfiles
Git packfiles were designed for mmap and local disk, so pulling one object out of a bucket means guessing at a byte range. I wrote a new format instead.
72
A 32-year-old bug walks into a Telnet server
A long, long time ago, in a land free of binary exploit mitigations, when Unix still roamed the Earth, there lived a pre-authentication Telnetd vulnerability. In fact, this vulnerability was born so long ago (way back in 1994) that it may even be older than you. To put the
73
The engineering behind the US Strategic Petroleum Reserve
One of the most fascinating things I’ve learned about recently is the engineering behind the US Strategic Petroleum Reserve. Here are the key requirements it’s designed to meet: Store …
74
Building a Linux GPU Driver for the M4 Mac Mini in One Month
Personal website of Cody Ho
75
76
Performance Improvements in .NET 11
Take a tour through hundreds of performance improvements in .NET 11.
77
Running Ubuntu on the Lenovo IdeaPad Duet
I own a Lenovo IdeaPad Duet Chromebook, a ChromeOS 2-in-1 ARM64 tablet. I had been running ChromeOS, then postmarketOS, which suddenly broke. Time for some firmware hacking to try and get Ubuntu to run there!
78
Distributed Systems Classics (2017)
A selected list of timeless and influential papers in distributed systems that shaped the research in the field. Intended to serve as a good…
79
Principles for Fast Tokio Applications
A microscope for Tokio and Rust applications
80
Dropping eBPF CPU Cost by About 90% with Memoization (Not AI Gen)
My brother and I spent a lot of time designing our eBPF security agent to be really fast from the ground up, but recently we discovered we could make it much faster using memoization! A couple of weeks ago, I profiled the eBPF code and found that the most expensive part of the protection isn’t actually enforcing a…
81
Flawed routers flood University of Wisconsin internet time server (2003)
ABSTRACT In May 2003, the University of Wisconsin - Madison found that it was the recipient of a continuous large scale flood of inbound Internet traffic destined for one of the campus' public Network Time Protocol (NTP) servers. The flood traffic rate was hundreds-of-thousands of packets-per-second, and hundreds of…
82
Why is the x86 undefined instruction called ud2? Why 2?
It came after ud0 and ud1.
83
Microcode in Intel's 8087 floating-point chip: the scale instruction
In the 1970s, floating-point arithmetic was a mess. Computer manufacturers had a dozen incompatible arithmetic standards. Moreover, floating...
84
Retrospectively Reverse-Engineering Apple's Neural Engine
I stopped working on the reverse-engineered Apple Neural Engine (ANE) driver three years ago, upon a sad mini realization that the ANE block is just not that useful, and I could be doing more useful things, and moved onto upstreaming other, more useful, blocks. The ANE's architecture was too opinionated to build a…
85
118M Queries per Second on Neki
We ran a massive, sharded Postgres database at 118.5 million queries per second, with 200k queries per second on each shard across 512 shards.
86
Measuring the sloppiness of code
Exploring how to measure code sloppiness, why correct code can still erode a codebase, and why human intuition and taste still matter.
87
Neki – Sharded Postgres
Neki, sharded Postgres by PlanetScale, is now available in platform preview.
88
Getting 50 GB/S Back from the Apple Neural Engine
An RTL performance erratum in the Apple M3 Neural Engine throttles DRAM weight streaming throughput down to 17–19 GB/s from the nominal 45–60 GB/s, whenever the total weight size is an integer multiple of 1 MiB, which currently affects 7 of ANEMLL ’s 15 models. Avoiding the problematic path in the kernel DMA engine's…
89
Performance of WebAssembly Runtimes in 2026
I wanted to know if WebAssembly runtimes are getting faster.
90
Compute-efficient pretraining and scaling to trillion-parameter models
Research update on compute-efficient pretraining and scaling to trillion-parameter models.