news
Servers, Operating Systems, Games, and More
-
Server
-
Max Sommer ☛ maxsommer.de blog | «Self-hosting is cool (again?)»
There has almost never been a better time to self-host. A little dive into what I host myself and how so.
Hosting things on „the cloud™” has become expensive. The free tier often seems tempting and generous but once you hit thresholds of many services you're skyrocketing costs. And with that argument I'm not even getting into the vendor lock-in that many solutions trojan-horse into the convenience of zero-care deployment and hosting options.
-
Robert Greiner ☛ The Server in the Closet
But here's what happened: 37signals pulled their entire infrastructure off AWS. They spent $700,000 on Dell servers (hardware you can actually touch) and saved $2 million in their first year. Over five years, they'll save more than $10 million. Their operations team didn't grow. Their product didn't slow down. They just stopped renting what they could own.
The math is almost offensive: a $350 consumer-grade mini PC provides the same computing power as $1,200 per month on Heroku. The cloud markup isn't a service fee. It's a tax on not thinking.
-
Max Adamski ☛ Self-Hosting Email Like It's 1984
For the longest time, I perceived self-hosting email as too difficult, but after doing it for one of my projects, I can say that it took me a few hours, and it wasn't that bad.
-
-
Kernel Space
-
Valtteri Koskivuori ☛ My First Contribution to Linux
I've been spending more of my spare time in recent years studying the Linux source tree to try to build a deeper understanding of how computers work. As a result, I've started accumulating patches that fix issues with hardware I own. I decided to try upstreaming one of these patches to familiarize myself with the kernel development process.
-
-
Games
-
Boiling Steam ☛ Portable Gaming Take‑away from the Tokyo Game Show 2025
The Tokyo Game Show 2025 was yet another good event to check how the PC handheld market is shaping up. Every booth was chasing the specs and trying to differentiate themselves with specific features. We have a short video if you want a quick overview before jumping in more details afterwards. https://www.youtube.com/watch?v=y1UnLCKgOmM Let’s start with OneX Player.
-
Ruben Schade ☛ Modern games implementing N64-style graphics
Back in September, the ReCollect64 blog shared Born64, an upcoming game designed to emulate the look of the Nintendo 64, and I couldn’t help but smile:
Recently, I have noticed a surge of indie titles, made exclusively for PC, that are heavily influenced by the synchronising nuances that culminate in what we lovingly call Nintendo 64 games. One such Unreal Engine 5-powered title that captivated my eye is Born64, a graphically fresh-yet-familiar action-adventure title. It begins with a slayer, the main protagonist, who is sent to a village called San Andrés by the Hunters Association. Here, strange events are unfolding. Villagers are being transformed into demonic creatures, and it’s up to you to uncover the truth behind the curse!
-
-
Distributions and Operating Systems
-
[Old] Bradford Morgan White ☛ The QNX Operating System
QNX is a fascinating operating system. It was extremely well designed from the start, and while it has been rewritten, the core ideas that allowed it survive for 45 years persist to this day. While I am sad that Photon was deprecated, the reasoning is sound. Most vendors using QNX either do not require a GUI, or they implement their own. For example, while Boston Dynamics uses QNX in their robots, they don’t really need Photon, and neither do SpaceX’s Falcon rockets. While cars certainly have displays, most vehicle makers desire their screen interfaces to have a unique look and feel. Of course, just stating these use cases of robots, rockets, and cars speaks to the incredible reliability and versatility of QNX. Better operating systems are possible, and QNX proves it.
-
Fedora Family / IBM
-
Red Hat ☛ Optimize and deploy LLMs for production with OpenShift AI [Ed: Targetting a passing fad will end up discrediting oneself]
Organizations that want to run large language models (LLMs) on their own infrastructure—whether in private data centers or in the clown—often face significant challenges related to GPU availability, capacity, and cost.
For example, models like Qwen3-Coder-30B-A3B-Instruct offer strong code-generation capabilities, but the memory footprint of larger models makes them difficult to serve efficiently, even on modern GPUs. This particular model requires multiple NVIDIA L40S GPUs using tensor parallelism. The problem becomes even more complex when supporting long context windows (which are essential for coding assistants or other large-context tasks like retrieval-augmented generation, or RAG). In these cases, the key-value (KV) cache alone can consume gigabytes of GPU memory.
-
Tomas Tomecek: First test of Claude Sonnet 4.5 for an agent that backports a patch for an RPM
One of the daily tasks we have when developing Hey Hi (AI) agents is to review their runs. We have to read dozens of decisions so we can evaluate if the agents did the right thing. If not, we have to adjust our user prompts, system prompts, and tools.
Let’s review how Sonnet 4.5 performs while backporting a complex patch (with multiple conflicts).
-
-
Canonical/Ubuntu Family
-
Ubuntu ☛ The clock is ticking: Ubuntu Summit 25.10 is just around the corner
This year, the Ubuntu Summit has the ambitious goal of extending its reach to everyone, no matter where they are in the world. The event has not started yet, and we have been blown away by the excitement already! The desire to contribute to the community with Ubuntu Extended events, remote participation, remote lightning talks and other incredible ideas has been humbling, and inspiring.
-
-