Pitting A Modern Chess Engine On A Pentium PC Against 1996’s Deep Blue

November 5, 2026 by Maya Posch 7 Comments

Although premature optimization can be a really bad thing, late optimization is sometimes unavoidable because we simply didn’t know about certain algorithms yet to massively improve the solving of a problem. When the 1990s was all about beating chess grandmasters at their game, IBM went so far as to build a the Deep Blue custom super computer for it. This is what [Nicholas Carlini] managed to effectively beat using a version of the Stockfish chess engine running on an era-appropriate Pentium PC.

A common way to measure one’s skill in chess involves the Elo rating system, named after chess master and physics professor [Arpad Elo]. As a relative system, it can be used to compare different players including chess engines against each other, with IBM’s Deep Blue scoring 2700, with the modern-day Stockfish engine hitting around 3700. This makes Stockfish statistically much more likely to win against a human chess grandmaster than Deep Blue, but also raises the question of how much of this advantage is due to Stockfish running on a much faster system.

To investigate the impact here, [Nicholas] used a Pentium Pro 200 MHz system with 64 MB of RAM (Fast Page or EDO SIMMs), which would have been a rather nice system for about 1995. Two chess engines were ported to this platform, being Crafty and Stockfish. Next both engines were tweaked with the help of an LLM coding bot to see what difference various compilation flags and algorithms would make.

Based on the results it would seem that the answer is ‘yes’: this 200 MHz Pentium Pro could have likely beaten even a chess grandmaster like [Kasparov] and Deep Blue in at least a few games, though without going back in time by about thirty years we’ll never know for certain.

Header: Wilfredor, CC0

Enable Resizable BAR On Unsupported Motherboards

October 26, 2026 by Maya Posch 23 Comments

Recently [Rod Burlamaqui] released the arc-rebar project on GitHub, enabling Intel GPUs to work significantly more efficiently if the system’s mainboard does not support resizable BAR. It does require a Debian or Ubuntu installation and of course a supported Intel GPU.

Although the resizable base address register (ReBAR) feature has been part of the PCIe specification for a while, its use was quite limited until Intel released its Arc series of graphic cards. These use this feature to map much more of the GPU’s video RAM (VRAM) into the PC’s address space than the standard 256 MB, assuming that the system is capable of decoding addresses above 4 GB, i.e. allow for 64-bit BARs.

For GPU manufacturers like AMD and Nvidia this wasn’t so much of an issue as their GPU designs developed in a world where ReBAR didn’t exist, while Intel’s new GPU generations were designed with ReBAR as a core feature, which naturally cripples performance if the mainboard does not have it enabled during PCIe bus enumeration.

This initramfs-based fix runs before the GPU driver loads and performs the requisite BAR configuration. In testing with an Intel Arc B580 GPU on a 2015-era Intel Xeon mainboard the difference between the default and the fixed bandwidth was a bump from a pitiful 15 GB/s to a respectable 392 GB/s, showing how this one fix can breathe new life into older systems.

Backup Tapes Come Back To The Homelab

October 12, 2026 by Tyler August 41 Comments

It’s one of the great mysteries of computing, how somehow in the Year Of OLlama 2026, tape remains the god of cheap storage. A while back [Ben Pettis] recounted his experience setting up an LTO-5 tape drive in his homelab, an experience which only becomes more relevant as hard drive prices stay high.

LTO-5 is already well out of date in the professional world, after all– a new server install would be debating the merits of LTO-9 versus LTO-10, but for the homelab it’s a matter of what’s affordable on the used market with decently-sized cartridges. If you can live with 1.5 TB of uncompressed data on a roughly palm-sized Linear Tape, Open cartridge then LTO-5 could be good enough. Yes, the tapes are advertized as 3TB, but that’s assuming a 2:1 compression ratio on the data, which you’re not likely to get on the photos and video that make up much of a data horder’s storage. That’s one thing [Ben] was quick to clear up for us. He also detailed his adventures with the cheaper-to-buy fiber channel model of drive.

The label would have you believe this box stores 30 TB.

Tape drives usually come with an SAS or fiber channel interface– fiber channel being, you guessed it, an optical solution unlike the SCSI-like SAS connection you may be familiar with for block storage devices. With a PCIe adapter it’s not hard to set up, provided you get the cables the right way around! That’s another detail we’re glad [Ben] shared so we can find it again if we have the same problem: with no switch in the circuit, you need a crossover cable to bring TX to RX. Which is obvious in retrospect, but obscure to those of us just learning about the standard.

The advantage, as [Ben] makes clear and everyone has always known is simply price. As of this writing, diskprices.com has 16$/TB as the lowest cost, while [Ben] was able to pick up lightly-used tapes on eBay for only 2$/TB. If you have a lot of data, those savings amortize the drive costs relatively quickly, since a used drive like [Ben]’s can be had for a few hundred bucks.

Is it the ultimate solution for our data storage woes? That depends on your needs, but perhaps not considering we get as many tips on storing data on reel-to-reel as we do on LTO. We even saw a tape-loading robot repurposed to serve bar one time.

Failing To Make A Fanless PC

October 5, 2026 by Fenix Guthrie 31 Comments

The discerning modern gamer’s quest is to find a quieter computer without sacrificing performance. That proves challenging when modern components are highly efficient at producing monumental quantities of thermal energy. So the typical solution ends up moving enough air through the computer to create a rather tumultuous racket. So why have fans at all? Well, as [Billet Labs] demonstrates, that’s not necessarily a good idea.

The build begins with a block of aluminum which, with a bit of hand-metal work, may become the base of the PC “case.” A dry-fit of the aesthetically modified parts taken from a previous attempt at a fanless PC reveals the steampunk aesthetic that defines the build. On final assembly, a large amount of thermal paste under the flex ATX power supply and 6mm of thermal pads under the motherboard provide passive cooling to the block of aluminum they are attached to

The cooling loop begins with three radiators: a big one, a small one, and a slightly smaller one. But raw metal PC radiators are difficult and expensive to come by these days. Some paint stripping and polishing will fix that problem. A few hand-made mounting brackets complete the aesthetic. All three radiators are stacked above the motherboard, and on top of each other, positioned horizontally to encourage passive airflow through them.

Continue reading “Failing To Make A Fanless PC” →

Coreboot Hikes The Bay Trail To DRAM Initialization

September 19, 2026 by Tyler August 18 Comments

This article is written on an open-source operating system, but not an open-source machine — the BIOS isn’t open-source, and even if it were supported by the coreboot project (formerly LinuxBIOS), there would still be a whole host of binary blobs required to get it to boot. On one vintage architecture, there’s one less blob, as coreboot can now initialize DRAM on Intel Bay Trail SOCs, as [Mate Kukri] presented in a talk at the recent Open Source Firmware Conference.

Bay Trail isn’t exactly cutting-edge hardware, to be sure — the SoCs are over a decade old at this point, and were only used in low-performance mobile applications like Chromebooks. On the other hand, coreboot has been on Chromebooks for at least as long. Getting DRAM set up is difficult because, well, you don’t have any memory to work with until you do. Traditionally, the way you did that was to call on one of the many proprietary ‘binary blobs’ provided with next to no documentation by the manufacturer. Reverse engineering that requires some serious bus-sluthing, which was done in software with the SerialICE debugger and the Unicorn Engine CPU emulator. The talk focused on that technique and how it might be applied more widely, rather than getting into the weeds of how to do DRAM init on one obsolete SOC. At some point the whole thing should be archived on the OFSC website so those of us not lucky enough to attend in person can hear what [Mate] — and all the other speakers — had to say.

Continue reading “Coreboot Hikes The Bay Trail To DRAM Initialization” →

Hackaday Europe 2026: Bare Metal Made Easy

September 15, 2026 by Zoe Skyforest 27 Comments

When we talk about programming in “bare metal,” it basically means writing software that runs right on the hardware with no operating system or abstraction layers in between. This gives the program the most direct possible access to peripherals and memory, with the tradeoff being that you don’t get the protection and ancillary features that come with an OS.

Sylvain Huet came down to Hackaday Europe 2026 to talk about making bare metal easy. Not just by ignoring operating systems and ever-bloating dependencies, but by rethinking the way we approach software development and by building a transparent platform from the ground up.

Continue reading “Hackaday Europe 2026: Bare Metal Made Easy” →

If It’s Electronic Paper, Why Not Print To It?

September 10, 2026 by Tyler August 42 Comments

E-ink is supposed to be the electronic version of dead-tree paper, but there’s one enormous difference: how you get your images and text onto it. For paper, you print on it. E-ink? Well, it depends on what it’s hooked up to. If it’s an ESP32C3-based Xteink X4 and X3, then you can just print to that, too, thanks to [Nishant Joshi]’s fork of the popular CrossPoint firmware.

The fork implements the Internet Printer Protocol, IPP, on the ESP32-C3. It makes loading documents a breeze: select the printer in your operating system, and press print. To the computer, the reader advertises itself as a printer that provides monochrome, single-sided output at 300 DPI on A5 paper, and accepts Apple raster and PWG raster formats. Of course these raster images are enormous compared to the paltry RAM available on the microcontroller, so pixels are received and written row-by-row — both to the screen directly, and to the SD card for later perusal. That makes a folder on the SD card the equivalent of a printer’s output tray.

This interface is amazing in its simplicity, and we’d love to see it in other E-ink devices. It reminds us somehow of AlphaSmart word processors — which are still relevant today in large part because their ‘file transfer’ mechanism is to pretend to be a USB keyboard and type the document onto your computer. As long as IPP is available, so too will you be able to get documents onto this e-reader.