61
What happens when a GPU reads memory
Following an LDG.E SASS instruction through the hardware units in an RTX 4090.
New
The latest blog posts to take off on Hacker News. 2,560 posts.
62
TigerBeetle Core System Architecture: Deconstructing Performance Engineering
An in-depth technical analysis of TigerBeetle's core architecture, exploring how static memory allocation, zero-copy io uring interfaces, and Zig-based…
63
We rebuilt the Linux MicroVM stack on Apple Silicon
Firecracker needs KVM and therefore a Linux host, and the image tooling around it needs loop mounts and Linux userland. We rebuilt the stack on Apple's Virtualization.framework to boot the same OCI images as microVMs on macOS and on Linux.
64
The August 17 outage
An update on the August 17 outage and the steps we're taking to improve reliability.
65
Linux 7.2
Igalia is an open source consulting firm specialised in the development of innovative projects and solutions. Our engineers have expertise in a wide range of technological areas, including browsers and client-side web technologies, graphics pipeline, compilers and virtual machines. We have the most WPE, WebKit…
66
A look under our trunk: what's in our compute
Compute is the brain of the Waymo Driver, translating raw sensor data into real-time driving commands. Operating demonstrably safe, physical AI on the road demands a fundamental shift towards a system engineered for deterministic, low-latency performance. Over the past decade, we have co-designed our hardware…
67
Kubernetes on Oxide: How customer needs shaped our integrations
How customer needs shaped Oxide's Kubernetes integrations.
68
SIMD in the 90s: Programming Intel's Pentium MMX
If you used PCs in the 90s, you probably remember the 'Intel Inside MMX' stickers. This article explores Intel's major attempt at bringing SIMD instructions to mainstream desktop CPUs.
69
Breaking the WAL
I did this from a lawn chair on the Sunshine Coast.
70
Rust SIMD on the GPU
At VectorWare , we are building the first GPU-native software company . Today, we are excited to announce that we can successfully use Rust's portable SIMD ( core::simd ) on the GPU. This milestone marks a significant step towards our vision of enabling developers to write complex, high-performance applications that…
71
Asynchronous I/O in DuckDB: Work, Thread, Work
Starting with v2.0, scheduled for fall 2026, DuckDB will support asynchronous reads of Parquet and CSV files. This can significantly speed up queries when synchronous I/O does not saturate the available bandwidth, as is typical in EC2/S3 compute-storage setups.
72
We replaced Redis with MySQL for inventory reservations and it scaled
How we used SKIP LOCKED, composite primary keys, and connection visibility to hit our scale targets.
73
Triton: DirectX 11 Driver for QEMU
In the prequel, we introduced Neptune, a Direct3D protocol forwarding layer for VirtIO. Neptune allowed us to serialize Direct3D API calls across the hypervisor boundary and this allowed us to run Wine games on a Linux guest with a Linux host faster than with DXVK directly in the guest. Admittedly, the payoff there…
74
Smaller, faster, safer: running Kimi and GLM at scale
Serving frontier models like Kimi and GLM means fighting for GPU memory. Here's how we quantize KV caches, compress model weights, and add integrity checks to serve them faster, cheaper, and safely.
75
76
NetBSD 11.0
The NetBSD project is pleased to (finally) announce the NetBSD 11.0 release ! See the release announcement for details.
77
A big win for Android interoperability
Google gatekept key Android features, and the European Commission asked what we thought about it. Now Alphabet must open 11 Android features to all.
78
Making Postgres queues scale
Lessons learned from scaling Postgres-backed durable queues for users running tens of billions of workflows per month.
79
SQLite in Production: Optimizing WAL Mode, Concurrency, and VFS Layers
Transitioning SQLite from a local development tool to a production-grade database requires a deep understanding of its internal mechanics. This article explores how to tune WAL mode, manage busy handlers, and leverage custom Virtual File System (VFS) layers to achieve ultra-low latency.
80
Postgres LISTEN/NOTIFY actually scales
How we optimized Postgres LISTEN/NOTIFY-backed data streams to 60K writes per second on a single Postgres server, with millisecond-scale latency.
81
SIMD for Collision
In a previous post SIMD Matters I discussed the big wins I got from graph coloring to make the contact solver faster. That same approach exists in Box3D. I like to call this approach “wide SIMD”. The idea is to process multiple work units at the same time. In the contact solver this means solving four contact points…
82
GC and Exceptions in Wasmtime
The Wasm GC and exceptions proposals are both enabled by default in today’sWasmtime 47 release! We are excited to help bring more languages to WebAssemblyand everywhere that Wasmtime runs. Getting to this point involved large Wasmtimecha...
83
Making 768 servers look like 1
How to make 768 distinct Postgres servers look like 1 to your applications.
84
Running Gemma 4 26B at 5 tokens/sec on a 13-year-old Xeon with no GPU
A storage appliance that shipped before AVX2 existed now runs Google's Gemma 4 26B in my basement, no GPU. Here's how we got a modern MoE model onto a CPU older than the instructions its inference engine assumed were there.
85
GhostLock, a stack-UAF that has existed in all Linux distributions for 15 years
GhostLock (CVE-2026-43499) is a Linux kernel vulnerability found by Nebu that exists in every major distribution since 2011. Triggering the bug does not require any special kernel config or privilege. By turning it into a 97% stable privilege escalation and container escape, Google has rewarded us $92,337 in…
86
Cloudflare Meerkat - Globally distributed consensus
Cloudflare Research is building a global consensus service called Meerkat that uses a new consensus algorithm called QuePaxa. We plan to use Meerkat to build a strongly consistent, fault-tolerant key-value store, and other applications.
87
Why we built yet another Postgres connection pooler
PgDog is a connection pooler, load balancer, and sharding proxy for PostgreSQL. Scale Postgres horizontally without rewriting your application.
88
Inference Optimization for MiMo v2.5: Pushing Hybrid SWA Efficiency to the Limit
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.
89
90
PostgreSQL and the OOM killer: Why we use strict memory overcommit
In this blog post, we will explain how strict memory overcommit protects your database from catastrophic OOM (out of memory) kills. We will also share how a one character kernel bug forced us to temporarily disable this setting. Finally, we will explain our heuristic for determining the right memory overcommit limit…