I don't normally do follow-ups and never this quick. After posting that last article, it was fun to read the comments on Reddit and Hacker News as they rolled in. I even found other discussions . I couldn't help wonder, "Could I have made it even more performant?" . After heading home I decided to take another look.
An optimization that made Pivot significantly faster on GROUP BY queries with a Top-K operator above them, and a walkthrough of how analytical databases run GROUP BY in the first place.
Let's talk about the "Windows Baseline Security Mode," too.
Discover how Bootc and OSTree are revolutionizing Linux system deployment, offering modern, efficient, and immutable management.
Modern processors have the ability to execute many instructions per cycle, on a single core. To be able to execute many instructions per cycle in practice, processors predict branches. I have made the point over the years that modern CPUs have an incredible ability to predict branches. It makes benchmarking difficult…
So where has the last six months gone ? I was planning on getting images done for Fedora 44 Beta but I was unwell and busy and ran out of time. So what better time to get them out than Pi Day! .
For many years, I sometimes discovered that commands I was sure I had run were no longer present in my Z shell history file (~/.zsh_history). In this article, I will show you how I tracked down the bug. Spoiler: ultimately, patching Zsh to make it crash loudly and analyzing the crash’s core dump was the winning…
Editors’ note: AI has been actively pushing the frontier of applied formal methods for computing systems. In this article, the Specula team wrote about their experience of evaluating LLMs on modeling system code, the basic capability for agentic model checking, using TLA+, a specification language f
In the world of matching decompilation, projects use a tool which consumes an input binary and lifts objects for comparison and linkage.
If we are running any eBPF workload or writing eBPF code, we want to measure its performance impact, and in this post we will demonstrate an example of how to do it. In this example, our goal is to measure the performance of file open operations, one of the most critical functions in the OS. Our code used file open…
How Lakebase's storage architecture pushes down full page write generation to cut WAL traffic by 94% and boost Postgres write throughput by up to 5x.
At VectorWare , we are building the first GPU-native software company . Today, we are excited to announce that we can successfully use Rust's std::thread on the GPU. This milestone marks a significant step towards our vision of enabling developers to write complex, high-performance applications that leverage the full…
Someone opened a PR to implement readahead in Turso. It was a throwaway implementation, but a good excuse to measure io_uring and understand more about it.
A WebAssembly module's linear memory can be shared directly with the Apple Silicon GPU: no copies, no serialization, no intermediate buffers. Here's how the zero-copy chain works, what we measured, and what it enables for stateful AI inference.
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.
Two services running on the same machine. One of them opens a listening TCP socket bound to localhost, the other one connects to that. They exchange data. Every now and then , the service that initiated the connection gets an ECONNRESET while reading data from the socket -- but no other errors show up in the logs, no…
When talking about TLA+, I keep referring to "abstraction" as the most important thing to learn . And it is about the hardest to learn as w...
Recently, I've been thinking a lot about scaffolding and languages for managing systems of LLMs coordinating with each other — new programming languages might be the ideal solution for this area.
After learning Apple killed off FireWire (IEEE 1394) support in macOS 26 Tahoe, I started looking at alternatives for old FireWire equipment like hard drives, DV cameras, and A/V gear. I own an old Canon GL1 camera, with a 'DV' port. I could plug that into an old Mac (like the dual G4 MDD above) with FireWire—or even…
The author of Java 26's MemoryMXBean.getTotalGcCpuTime() analyzes the CPU-memory trade-off and why pause times fail to capture explicit GC cost.
I want to make this clear in the first sentence because its biggest chance that people will read it – this article is entirely based on work done by Christian Hofstede-Kuhn (Larvitz) that wro…
Research update on compute-efficient pretraining and scaling to trillion-parameter models.
A customer's Go application kept crashing on a 32-bit embedded Linux system. We traced the fatal netpoll error to a pointer aliasing bug in the Go runtime.
I’m writing this so you don’t slap your forehead like I almost did when I decided to run swap in production to absorb memory spikes.
In 1994 I got my first computer: an Intel i486 DX2-66 with 4 MB RAM and a 512MB harddisk. The software was IBMs OS/2 and Microsofts Windows 3.11. In the next four years I was upgrading this machine every few months with more RAM (up to 16MB), a CD-ROM-drive and a soundblaster card. So I learned upgrading this machine…
Before we explore FreeBSD jails, it is worth refreshing our understanding of how Linux solved the same problem with LXC (Linux Containers).
Dead tuples from high-churn job queues can silently degrade your Postgres database when vacuum falls behind—especially alongside competing workloads. Traffic Control keeps cleanup on track.
MariaDB is a popular open source SQL database which began as a fork of MySQL . MariaDB Galera Cluster is an active-active replication system for MariaDB which allows reads and writes on every node. In 2015 this author analyzed MariaDB with Galera Cluster and found that although Galera claimed to offer Snapshot…
To kick off 2026, I had clear objectives in mind: decommissioning my trusty VPS and setting up its successor. Embracing a complete paradigm shift, I built myself a container-centric, declarative, and low-maintenance setup for the years to come.
An end-to-end engineering practice for the MiMo-V2.5 series inference system, covering KVCache management, tiered caching, SWA-aware prefix cache trees, scheduling, Prefill/Decode pipelines, and multimodal optimizations.
