Despite faster CPUs, RAM and storage, today’s Windows experience doesn’t feel noticeably different from back in the 2000s when XP and later Windows 7 ruled the roost. To quantify this feeling, [TrigrZolt] decided to run a series of benchmarks on a range of Windows versions.

Covering Windows XP, Vista, 7, 8.1, 10 and 11, the Pro version of each with the latest service packs and updates was installed on the same laptop: a Lenovo ThinkPad X220. It features an Intel i5 2520M CPU, 8 GB of RAM, built-in Intel HD Graphics 3000 and a 256 GB HDD.

For start-up, Windows 8.1 won the race, probably due to having the Fast Boot feature, while Windows 11 came in dead last as it showed the desktop, but struggled to show the task bar. Windows XP’s install size was the smallest and also had the lowest RAM usage with nothing loaded at 800 MB versus 3.3 GB for Windows 11 in last place.

Using the Chrome-based Supermium browser, memory management was tested, with XP performing as poorly as Windows 11, while Windows 7 and 8.1 took home the gold at over two-hundred tabs open before hitting the total RAM usage limit of 5 GB. That XP performed so poorly was however due to an issue with virtual memory and not hitting the RAM limit, which means that Windows 11 is the real dunce here.

This is a pattern that keeps repeating: Windows 11 was last in the battery test, took longer to render a video project in OpenShot, took its sweet time opening the File Explorer window, and opening built-in applications like MS Paint left enough time to fetch a fresh cup of coffee. Not to mention Windows 11 taking the longest to open websites and scoring worst of all in single-threaded CPU-Z.

Much seems to be due to the new code in Windows 11, as Microsoft has opted to start doing major rewrites since Windows 7, hitting a crescendo with Windows 11. Although there’s the unhelpful fact that Windows 11 by default encrypts the storage with the very slow software-based BitLocker, its massive RAM usage and general sluggishness are such a big deal that even Microsoft has acknowledged this and added workarounds for the slow File Explorer in Windows 11 by preloading components into RAM.

All of this appears to be part of the same trend in software development, where more resources are pointlessly used due to developing for the hardware, and performance increasingly takes a backseat to abstractions and indirections that effectively add bloat and latency.

youtube.com/embed/7VZJO-hOT4c?…

hackaday.com/2026/01/02/benchm…

For various reasons, crypto mining has fallen to the wayside in recent years. Partially because it was never useful other than as a speculative investment and partially because other speculative investments have been more popular lately, there are all kinds of old mining hardware available at bargain prices. One of those is the Asrock AMD BC250, which is essentially a cut down Playstation 5 but which has almost everything built into it that a gaming PC would need to run Steam, and [ETA PRIME] shows us how to get this system set up.

The first steps are to provide the computer with power, an SSD, and a fan for cooling. It’s meant to be in a server rack so this part at least is pretty straightforward. After getting it powered up there are a few changes to make in the BIOS, mostly related to memory management. [ETA PRIME] is uzing Bazzite as an operating system which helps to get games up and running easily. It plays modern games and even AAA titles at respectable resolutions and framerates almost out-of-the-box, which perhaps shouldn’t be surprising since this APU has a six-core Zen 2 processor with a fairly powerful RDNA2 graphics card, all on one board.

It’s worth noting that this build is a few weeks old now, and the video has gotten popular enough that the BC250 cards that [ETA PRIME] was able to find for $100 are reported to be much more expensive now. Still, though, even at double or triple the price this might still be an attractive price point for a self-contained, fun, small computer that lets you game relatively easily and resembles the Steam Machine in concept. There are plenty of other builds based on old mining hardware as well, so don’t limit yourself to this one popular piece of hardware. This old mining rig, for example, made an excellent media server.

youtube.com/embed/q_CxcbS5HI8?…

hackaday.com/2026/01/02/a-stea…

Over the decades we have seen many DIY clocks and wrist watches presented, but few are as likely to get you either drawing in the crowds, or quietly snickered at behind your back, as a binary watch of some description does. A wrist watch like [qewer]’s qron0b project which also uses BCD encoding to display the current time is among our more rare project types here, with us having to go all the way back to 2018 for a similar project as well as a BCD desk clock.

As is typical, a single CR2032 coin cell powers the entire PCB, with an ATtiny24A or compatible as the MCU, a DS1302 RTC and the requisite 4×4 LED matrix to display the hours and minutes. Technically three LEDs are unneeded here, but it looks nicely symmetrical this way, and the extra LEDs can be used for other tasks as the firmware is expanded from the current setting and reading of the time.

The AVR C firmware can be found in the above linked GitHub repository, along with the KiCad PCB project and FreeCAD design files for the watch body. The body accepts a 22 mm GT2/GT3-style watch strap to complete the assembly. With a single CR2032 you’re assured of at least a few months of runtime.

hackaday.com/2026/01/02/qron0b…

In my country, we have a saying: the sun is a deadly lazer. Well, it’s not so much a folk saying as a meme, and not so much in one country as “the internet”. In any case, [LiamTronix] was feeling those cancer rays this harvest season when running his electric tractor, and realized that– since he’s already charging it with ground-mounted solar panels anyway–if he’s going to build a roof for his ride, he might as well make charge the batteries.

Another bonus is safety: the old Massey-Ferguson at the heart of the electric tractor build didn’t come with any rollover protection from the factory back in the 1960s. Since having however many tons of tractor roll onto you was bad enough before it got a big hefty battery pack, we heartily approve of including a roll cage in this build. Speaking of battery packs, he’s taking this chance to upgrade to a larger LiFePo pack from the LiIon pack he installed when we first featured this conversion in 2024.

Atop the new roll cage, and above the new battery, [Liam] installed four second-hand 225 W solar panels. Since that’s under 1kW even if the panels have not degraded, the tractor isn’t going to be getting much charge as it runs. In the northern winter, [Liam] is only able to pull 80 W from the set. That’s not getting much work done, but who wants a tractor without a cab or heater when it’s below freezing? In the summer it’s a much better story, and [Liam] estimates that the roof-mounted panels should provide all of the energy needed to run the tractor for the couple hours a day he expects to use it.

If you’re wondering how practical all this is, yes, it can farm — we covered [Liam] putting the project through its paces in early 2025.

hackaday.com/2026/01/02/adding…

As locked-down as the Amazon Echo Show line of devices are, they’re still just ARM-based Android devices, which makes repurposing it somewhat straightforward as long as what you want is another Android device.
Running Home Assistant on an Echo Show 8 with LineageOS.
In this case, we’re talking about the first-generation Amazon Echo Show 8, which is a 2019-era device that got jailbroken back in November by [Rortiz2]. The process was then demonstrated in a video by [Dammit Jeff].

Currently only two devices are supported by this jailbreak, with the Echo Show 5 being the other one. If there’s enough interest, there doesn’t appear to be any technical reason at least for why this support couldn’t be extended to other devices. One major reason for jailbreaking is to put LineageOS on your Echo device courtesy of these Echo Show devices recently beginning to show advertisements, with no way to disable this.

The process of jailbreaking and installing the LineageOS ROM is somewhat long as usual, with plenty of points where you can make a tragic mistake. Fortunately it’s pretty simple as long as you follow the steps and afterwards you can even install the Google apps package if that’s your thing. Just mind the 1 GB RAM and 8 GB of storage on the Echo Show 8. In the case of [Jeff] he mostly replicated the home automation and entertainment features of Amazon’s FireOS with far less locked-down alternatives like Home Assistant.

youtube.com/embed/h0-MlJ38BXw?…

hackaday.com/2026/01/02/jailbr…

Elliot was of at Europe’s largest hacker convention: Chaos Communication Congress. He had an awesome time, saw more projects than you might think humanely possible, and got the flu. But he pulled through and put this audio tourbook for you.

So if you’ve never been to CCC, give it a listen!

html5-player.libsyn.com/embed/…

In the far future, all the cool kids will be downloading MP3s of their favorite podcasts.

Where to Follow Hackaday Podcast

Places to follow Hackaday podcasts:

• iTunes
• Spotify
• Stitcher
• RSS
• YouTube
• Check out our Libsyn landing page

What’s That Sound

• If you think you know what the mystery sound was this week, give us your best guess for a chance at a Hackaday Podcast t-shirt.

hackaday.com/2026/01/02/hackad…

Thanks to the Raspberry Pi, we have easy access to extremely inexpensive machines running Linux that have all kinds of GPIO as well as various networking protocols. And as the platform has improved over the years, we’ve seen more demanding applications on them as well as applications that use an incredibly small amount of power. This project combines all of these improvements and implements a media streaming server on a Raspberry Pi that uses a tiny amount of energy, something that wouldn’t have been possible on the first generations of Pi.

Part of the reason this server uses such low power, coming in just around two watts, is that it’s based on the Pi Zero 2W. It’s running a piece of software called Mini-Pi Media Server which turns the Pi into a DLNA server capable of streaming media over the network, in this case WiFi. Samba is used to share files and Cockpit is onboard for easy web administration. In testing, the server was capable of streaming video to four different wireless devices simultaneously, all while plugged in to a small USB power supply.

For anyone who wants to try this out, the files for it as well as instructions are also available on a GitHub page. We could think of a number of ways that this would be useful over a more traditional streaming setup, specifically in situations where power demand must remain low such as on a long car trip or while off grid. We also don’t imagine the Pi will be doing much transcoding or streaming of 4K videos with its power and processing limitations, but it would be unreasonable to expect it to do so. For that you’d need something more powerful.

youtube.com/embed/rvEQalALV6Y?…

Thanks to [Richard] for the tip!

hackaday.com/2026/01/02/low-co…

There’s folk wisdom in just about every culture that teaches about renewable energy — things like “make hay while the sun shines”. But as an industrial culture, we want to make hay 24/7 and not be at the whims of some capricious weather god! Alas, renewable energy puts a crimp in that. Once again, energy supplies are slowly becoming tied to the sun and the wind.

Since “Make compute while the wind blows” doesn’t have a great ring to it, clearly our civilization needs to come up with some grid-scale storage. Over in Sardinia they’re testing an idea that sounds like hot air, but isn’t — because the working gas is CO2.

The principle is simple: when power is available, carbon dioxide is compressed, cooled, and liquefied into pressure vessels as happens at millions of industrial facilities worldwide every day. When power is required, the compressed CO2 can be run through a turbine to generate sweet, sweet electricity. Since venting tonnes of CO2 into the atmosphere is kind of the thing we’re trying to avoid with this whole rigmarole, the greenhouse gas slash working fluid is stored in a giant bag. It sits, waiting for the next charge cycle, like the world’s heaviest and saddest dirigible. In the test project in Sardinia — backed by Google, amongst others — the gas bag holds 2000 tonnes and can produce 20 megawatts of power for up-to 10 hours.

The scheme does require pressure vessels the size of buildings, which may make some nervous.
That’s not exactly astounding. It gets you through the night, but leaves you hanging if the next day is cloudy. But it’s scalable. The turbine is 20 megawatts, sure, but all you need is land to add extra energy capacity. The 200 MWh pilot plant is a five hectare facility, which is only about 12.3 acres, or roughly 1/10th the size of the Mall of America. It seems like increasing capacity would be fairly trivial; unlike, say, pumped hydro storage, no special topography is required. Ten hours of storage is also notably longer than the six to eight hours grid-scale battery farms usually aim for.

As of this writing, there’s only one of these plants in operation, but expect that to change rapidly. In 2026 the company behind the Sardinia project, Energy Dome, plans on putting in grid-scale storage based on its technology in India and Wisconsin, and that’s before Google gets into it. They’re hoping to roll this technology out at a number of data centers worldwide, though the exact details of the deal aren’t public.

We’ve talked about grid-scale energy storage before, using everything from liquid tin to electric car batteries and big piles of gravel. This methodology has a lot to recommend it over those others in comparison, and should worst come to worst, at least it won’t burn for days like certain batteries we could name. Releasing 2000 tonnes of CO2 might not be as benign as a failure from a liquid air battery, but storing liquid CO2 under pressure is a lot easier holding onto cryogenic air.

All images credited to Luigi Avantaggiato.

hackaday.com/2026/01/02/liquid…

Mentre il settore dibatte su quale rete neurale sia la “più intelligente” e quale framework sia il “più moderno”, tecnologie vecchie di decenni continuano a turbinare silenziosamente sotto la superficie del settore bancario.

Quest’anno, COBOL ha compiuto 66 anni e continua a supportare attività mission-critical: elaborazione delle fatture, gestione delle reti bancomat, compensazione delle carte di credito e liquidazione batch notturna. Dove COBOL risiede, si trovano tipicamente anche i mainframe: in banche, compagnie assicurative, sistemi governativi e grandi sistemi di contabilità e magazzino.

Ciò ha portato a una semplice domanda: quali tecnologie funzioneranno probabilmente tra 50 anni e oltre, anche se tutto intorno a noi cambiasse più volte? È importante considerare che il Cobol è cambiato nel tempo: l’attuale standard COBOL 2023 assomiglia poco alle prime versioni del linguaggio che Grace Hopper ha contribuito a creare, e i mainframe moderni hanno fatto molta strada dai primi sistemi IBM 701 e IBM/360. Ma la continuità rimane; sono le implementazioni a cambiare, non il percorso di sviluppo in sé.

Tra i linguaggi di programmazione, il C, che ha più di mezzo secolo, è un candidato promettente per la longevità. Pur avendo i suoi problemi di sicurezza , rimane estremamente valido in termini di velocità e versatilità: il C può essere utilizzato praticamente su qualsiasi architettura e in qualsiasi attività di basso livello.

Mentre si parla di Rust che sta gradualmente sostituendo il C nella programmazione di sistema, vale la pena ricordare che Rust è già arrivato su Linux, ma i principali punti di forza del C – velocità e portabilità – non sono scomparsi.

Un altro pilastro “indistruttibile” è SQL. È integrato in tutti i principali DBMS relazionali, accumulando enormi quantità di logica di business, stored procedure e query, rendendo praticamente impossibile sostituirli tutti in una volta. JavaScript e TypeScript sono una storia simile: sono spesso criticati e derisi, ma la piattaforma web si basa sulla compatibilità, il che significa che il linguaggio del browser rimarrà rilevante finché il web stesso sopravviverà.

Un livello separato di tecnologie longeve è associato non ai linguaggi, ma agli ecosistemi. Linux esisterà ancora nel 2100 e Git rimarrà lo strumento di sviluppo di base per molto tempo a venire. Sorprendentemente, accanto a loro ci sono gli editor “eterni” vi ed Emacs, così come Bash, che ha superato molti concorrenti e, per inerzia, continua a essere lo strumento standard in molti scenari.

Tra le soluzioni infrastrutturali più recenti, ma già consolidate, vale la pena sottolineare Kubernetes . Ha ricevuto la sua giusta dose di critiche, ma oggi è lo standard de facto per l’orchestrazione dei container cloud e il fondamento dell’approccio “cloud-native”, quindi l’inerzia del mercato è dalla sua parte. Nel mondo applicativo, Photoshop ha alte probabilità di rimanere valido: anche con alternative, l’ecosistema e le abitudini dei professionisti spesso prevalgono sulle argomentazioni ideologiche.

La parte più dolorosa del dibattito riguarda i formati di file. Una volta che un formato diventa uno standard industriale, persiste per anni e decenni, anche se esiste un sostituto più aperto e pratico. Esempi includono DOC e DOCX, che hanno soppiantato approcci più aperti, e PDF , apprezzato per il suo aspetto coerente dei documenti “ovunque e sempre”, ma che in pratica si scontra con una moltitudine di varianti e incompatibilità. Un esempio significativo proviene dall’industria musicale: dopo che un popolare standard di notazione proprietario è stato di fatto abbandonato, gli utenti hanno scoperto che migrare il lavoro accumulato verso altre soluzioni era estremamente difficile, e a volte impossibile, senza perdite.

La conclusione finale è semplice: gli standard aperti e il software open source hanno maggiori probabilità di sopravvivere per decenni, mentre le tecnologie legate a una singola azienda sono intrinsecamente fragili, anche se oggi sembrano “troppo grandi per fallire”.

L'articolo 66 anni e ancora imbattibile: come il COBOL domina silenziosamente il mondo bancario proviene da Red Hot Cyber.

Quando si parla della funzione del DPO è possibile escludere la possibilità di fare gioco di squadra, celebrando una sorte di one-person-band. Quando invece ci si trova a dover esercitare in modo sensato la funzione, allora è necessario ragionare su come fare gioco di squadra. Al di là dell’ovvia ricerca di sinergie con le ulteriori funzioni esterne o interne dell’organizzazione, è fondamentale infatti che venga garantita la capacità di svolgere i compiti attribuiti dall’art. 39 GDPR e, soprattutto, garantire la continuità della sorveglianza. Questo può comportare o lo stanziamento di risorse adeguato per coinvolgere altri professionisti o garantire alcuni collaboratori, o altrimenti la formazione di un team DPO.

Mentre il primo caso rientra nell’ambito dell’adeguatezza dello stanziamento di risorse che l’organizzazione deve fornire a supporto del DPO ai sensi dell’art. 38 par. 2 GDPR, nel secondo caso è invece chi si propone come DPO a dover fornire le evidenze necessarie a dimostrare la capacità di assolvere i propri compiti, come previsto dall’art. 37 par. 5 GDPR. Ovviamente, in entrambi i casi la responsabilità di adempiere agli obblighi previsti dal GDPR è e resta in capo all’organizzazione che deve comunque essere in grado di garantire l’effettività della funzione a partire dalla selezione e per tutto lo svolgimento dell’incarico.

Mentre la dotazione di risorse in termini di collaboratori a supporto del DPO è rimesso alla capacità dell’organizzazione di rispettare il GDPR e alla libertà di cui al principio di accountability, la formazione e conduzione del team DPO incontra necessariamente le regole proprie della funzione che vanno riferite più ad un “Office” che ad un singolo “Officer”.

Best practices per il team DPO.

La formazione del team DPO è presa in considerazione all’interno delle linee guida WP243 sui responsabili della protezione dei dati, indicando delle buone pratiche sia per favorire un’organizzazione trasparente e corretta sia per prevenire conflitti d’interesse. Ad esempio, è necessario non solo che ciascun componente del team possieda i requisiti per ricoprire l’incarico, ma che goda anche delle tutele previste dal GDPR fra cui principalmente quella dell’indipendenza funzionale prevista dall’art. 38 par. 3 GDPR.

Sul piano organizzativo, la ripartizione interna dei compiti è funzionale non solo per garantire l’efficienza del servizio ma anche per individuare un unico soggetto che faccia da referente per i vertici dell’organizzazione. Motivo per cui diventa opportuno dettagliare questo ambito all’interno del contratto di servizi (se esterno) o altrimenti nella descrizione del funzionamento dell’ufficio (se interno).

Quando il gioco di squadra è trasversale.

Ultimo ma non meno importante è l’aspetto invece del gioco di squadra fra DPO di diverse organizzazioni, il più delle volte nel rapporto titolare-responsabile (ma anche: fra contitolari, o titolari differenti). Poiché ciascun DPO svolge i propri compiti secondo un approccio risk-based nei confronti degli interessati, e gli interessati sono il comune denominatore dei trattamenti svolti fra questi soggetti differenti, è quanto mai opportuno – e invero, il più delle volte dimenticato – che si cooperi professionalmente in quanto si perseguono obiettivi se non proprio comuni quanto meno non divergenti: la garanzia della protezione dei dati personali degli interessati.

Questo comporta che nelle qualità personali espresse dallo svolgimento dei compiti sia riscontrabile anche quello spirito di leale collaborazione che invoca integrità e standard deontologici nel contribuire a dare efficace attuazione ai principi del GDPR e ai diritti degli interessati.

L'articolo Perché il DPO non può più essere un lupo solitario (e rischi di far fallire il GDPR) proviene da Red Hot Cyber.