Topics /
The most popular blog posts about systems & performance on Hacker News. 555 posts.
22 May 2017 by Phillip Johnston • Last updated 27 October 2023I want to make a few points up front to short-circuit common misunderstandings with this article: I have a zero warning policy on my projects. I turn on a lot of warnings. -Wall, -Wextra, and a whole host of specific warning flags. I supplement … Continue…
I've been doing a bit of work cleaning up the CSS on this blog in the past couple days — the CSS is from a template that I copied more than five years ago at this point, and it was pretty crufty even when I first got it.
Suppose you’re not using a C runtime on Linux, and instead you’re programming against its system call API. It’s long-term and stable after all. Memory management and buffered I/O are easily solved, but a lot of software benefits from concurrency. It would be nice to also have thread spawning capability. This article…
Back in spring 2020, I was wrapping up the distributed systems course I was teaching, and for the last lecture, decided to spend a little time poking at the question of who actually invented vector clocks. Most people who need something to cite for vector clocks cite Friedemann Mattern’s “Virtual Time and Global…
Vigneshwaran C explained the internals of PostgreSQL Logical Replication during PGConf India - if you were not able to attend, you can read all here.
What’s the “right” level of CPU utilization for a server? If you look at a monitoring dashboard from a well-designed and well-run service, what CPU utilization should we hope to see, averaged over a day or two? It’s a very general question, and it’s not clear it should have a single answer. That said, for a long time…
What kind of durability guarantees fsync system call provides in Linux? Let's find out.
Our systems and ourselves
Box64 and RISC-V, the challenges of wrtting the Dynarec, and benchmarks
More or less the title this time, but first, a story about SIMD. There are three levels of understanding how SIMD works (well, at least I am level 3 at the moment):
[Update: 13-Apr-2023 with some of the background context and some things I learned from hackernews ] [Update: 18-Dec-2023] This hiduti...
There’s a lot of excitement right now about human-in-the-loop systems supercharged by foundation models including chat assistants ( ChatGPT ), word processing ( Microsoft Office ), graphic design ( Stable Diffusion ), and code editing ( Copilot ). However, these systems only begin to scratch the surface of the role…
You're probably reading this page because you've attempted to access some part of my blog (Wandering Thoughts) or CSpace , the wiki thing it's part of. Unfortunately you're using a browser version that my anti-crawler precautions consider suspicious, most often because it's too old (most often this applies to versions…
A summary of highlights from the September 2023 release of Chapel 1.32.0
A streaming SQL engine keeps queries’ results up to date without ever having to recalculate them, even as the underlying data changes. To explain this, imagine a simple query, such as SELECT count(*) FROM humans
Yesterday Lauri Vasama showed me this awesome Godbolt: struct S { static std::unique_ptr<S> make() noexcept; ~S(); static std::unique_ptr<S> factory() { std::unique_ptr<S> s = S::make(); return M(s); } }; void take_ownership(S*) noexcept; void test() { std::unique_ptr<S> p = S::factory()…
AI is powered by a virtuous circle of data, algorithms (“models”), and compute. Growth in one pushes needs in the others and can grossly affect the developer experience on aspects like usability and performance. Today, we have more data and more AI model research than ever before, but compute isn’t scaling at the same…
A decision that arises over and over when designing concurrent programs is whether to represent program state in control flow or as data. This post is about what that decision means and how to approach it. Done well, taking program state stored in data and storing it instead in control flow can make programs much…
Graphing TLA+ state spaces for fun and profit.
A short tale of how horrible code yields clean performance.
The database zeitgeist has decided that ANSI SQL '92 isn't enough. We need something better.
Spoiler: Yes. (But the "why" matters. A lot.)
A few months after my contract with Haiku, Inc. began, I rewrote the implementation of the Haiku kernel’s condition variables (as opposed to our userspace condition variables, which are from POSIX.) As this new implementation has run in Haiku …
This is the second post in the series about effortless performance improvements in C++. Check the first post to get the full story!
This is the story of a clever trick that's been around for at least 35 years, in which array values can be left uninitialized and then read during normal operations, yet the code behaves correctly no matter what garbage is sitting in the array. Like the best programming tricks, this one is the right tool for the job…
Detailed explanation on how RISC-V boot process works and what role SBI has in the RISC-V software stack. Concrete example with OpenSBI.
This post describes the design and implementation of a runtime benchmark suite for measuring the performance of Rust programs, which was recently added into the Rust compiler suite. I have recently blogged about how the whole benchmark suite works, so feel free to read that post first if you want to gather a bit more…
What is a LAN? What is a network segment? How are L1, L2, and L3 segments related? What does it take to send an IP packet from one host to another? What is the difference between LAN, VLAN, and VXLAN?
I’ve been writing a pubsub listener for work recently and had this silly “shower thought”: the ideal queue should be empty. Why? If the queue is not empty, it means that whatever job is consuming from the queue is not keeping up with messages coming into it. Ok, then if queues are supposed to be empty, then what are…
