Author: Bryan Cockfield

1576 Articles

Genetic Algorithm Runs On Atari 800 XL

February 21, 2025 by 7 Comments

For the last few years or so, the story in the artificial intelligence that was accepted without question was that all of the big names in the field needed more compute, more resources, more energy, and more money to build better models. But simply throwing money and GPUs at these companies without question led to them getting complacent, and ripe to be upset by an underdog with fractions of the computing resources and funding. Perhaps that should have been more obvious from the start, since people have been building various machine learning algorithms on extremely limited computing platforms like this one built on the Atari 800 XL.

Unlike other models that use memory-intensive applications like gradient descent to train their neural networks, [Jean Michel Sellier] is using a genetic algorithm to work within the confines of the platform. Genetic algorithms evaluate potential solutions by evolving them over many generations and keeping the ones which work best each time. The changes made to the surviving generations before they are put through the next evolution can be made in many ways, but for a limited system like this a quick approach is to make small random changes. [Jean]’s program, written in BASIC, performs 32 generations of evolution to predict the points that will lie on a simple mathematical function.

While it is true that the BASIC program relies on stochastic methods to train, it does work and proves that it’s effective to create certain machine learning models using limited hardware, in this case an 8-bit Atari running BASIC. In previous projects he’s also been able to show how similar computers can be used for other complex mathematical tasks as well. Of course it’s true that an 8-bit machine like this won’t challenge OpenAI or Anthropic anytime soon, but looking for more efficient ways of running complex computation operations is always a more challenging and rewarding problem to solve than buying more computing resources.

Continue reading “Genetic Algorithm Runs On Atari 800 XL” →

Open-Source Random Numbers

February 20, 2025 by 25 Comments

Whether it’s a game of D&D or encrypting top-secret information, a wide array of methods are available for generating the needed random numbers with high enough entropy for their use case. For a tabletop game this might be a single die but for more sensitive applications a more robust method of generating random numbers is needed. Programmers might reach for a rand() function of some sort, but these pseudorandom numbers don’t cut the mustard for encryption. For that you’ll need a true random number generator (RNG), and this open-source hardware RNG uses one of the better methods we’ve seen.

The device, called RAVA, is based on a property found in many electronic devices called avalanche breakdown. Avalanche breakdown occurs when a high voltage (in this case approximately 25V) is applied in the reverse bias direction, with this device using a pair of Zener diodes. When this high voltage is applied, an “avalanche” of electrons occurs which allows the diodes conduct in the opposite direction that they would when they are forward biased. This isn’t a constant current flow, though; there are slight variations over time which can be amplified and used as the random number generator. The noise is amplified over a series of op amps and then fed to an ATmega32U4 microcontroller which can provide the user with 136.0 Kbit/s of random data.

Unlike other random number generators, this device is based on a method generally accepted to be truly random. Not only that, but since it’s based on discrete hardware it can be accessed directly for monitoring and replacement in case of faults, unlike other methods which are more “black boxes” and are more opaque in their processes which are thus harder to audit. We also appreciate it’s open-source nature as well, and for some more information on it be sure to check out the paper on it in IEEE. If you’re looking for something to generate random numbers but will also bring some extra flair to the next game night, take a look at this radioactive dice replacement.

Measuring Local Variances In Earth’s Magnetic Field

February 17, 2025 by 11 Comments

Although the Earth’s magnetic field is reliable enough for navigation and is also essential for blocking harmful solar emissions and for improving radio communications, it’s not a uniform strength everywhere on the planet. Much like how inconsistencies in the density of the materials of the planet can impact the local gravitational force ever so slightly, so to can slight changes impact the strength of the magnetic field from place to place. And it doesn’t take too much to measure this impact on your own, as [efeyenice983] demonstrates here.

To measure this local field strength, the first item needed is a working compass. With the compass aligned to north, a magnet is placed with its poles aligned at a right angle to the compass. The deflection angle of the needle is noted for varying distances of the magnet, and with some quick math the local field strength of the Earth’s magnetic field can be calculated based on the strength of the magnet and the amount of change of the compass needle when under its influence.

Using this method, [efeyenice983] found that the Earth’s magnetic field strength at their location was about 0.49 Gauss, which is well within 0.25 to 0.65 Gauss that is typically found on the planet’s surface. Not only does the magnetic field strength vary with location, it’s been generally decreasing in strength on average over the past century or so as well, and the poles themselves aren’t stationary either. Check out this article which shows just how much the poles have shifted over the last few decades.

Decoy Killswitch Triggers Alarm Instead

February 17, 2025 by 43 Comments

There are a few vehicles on the road that are targeted often by car thieves, whether that’s because they have valuable parts, the OEM security is easily bypassed, or even because it’s an antique vehicle that needs little more than a screwdriver to get started. For those driving one of these vehicles an additional immobilization feature is often added, like a hidden switch to deactivate the fuel pump. But, in the continual arms race between thieves and car owners, this strategy is easily bypassed. [Drive Science] hopefully took one step ahead though and added a decoy killswitch instead which triggers the alarm.

The decoy switch is placed near the steering column, where it would easily be noticed by a thief. Presumably, they would think that this was the reason the car wouldn’t start and attempt to flip the switch and then start the ignition. But secretly, the switch activates a hidden relay connected to the alarm system, so after a few seconds of the decoy switch activating, the alarm will go off regardless of the position of this switch. This build requires a lot of hiding spots to be effective, so a hidden method to deactivate the alarm is also included which resets the relay, and another killswitch which actually disables the fuel pump is also added to another secret location in the car.

As far as “security through obscurity” goes, a build like this goes a long way to demonstrate how this is an effective method in certain situations. All that’s generally needed for effective car theft prevention is to make your car slightly more annoying to steal than any other car on the road, and we think that [Drive Science] has accomplished that goal quite well. Security through obscurity is generally easily broken on things deployed on a much larger scale. A major European radio system was found to have several vulnerabilities recently thanks in part to the designers hoping no one would look to closely at them.

Continue reading “Decoy Killswitch Triggers Alarm Instead” →

Parametric Design Process Produces Unique Speakers

February 17, 2025 by 28 Comments

When building one-off projects, it’s common to draw up a plan on a sheet of paper or in CAD, or even wing it and hope for the best outcome without any formal plans. Each of these design philosophies has its ups and downs but both tend to be rigid, offering little flexibility as the project progresses. To solve this, designers often turn to parametric design where changes to any part of the design are automatically reflected throughout the rest, offering far greater flexibility while still maintaining an overall plan. [Cal Bryant] used this parametric method to devise a new set of speakers for an office, with excellent results.

The bulk of the speakers were designed with OpenSCAD, with the parametric design allowing for easy adjustments to accommodate different drivers and enclosure volumes. A number of the panels of the speakers are curved as well, which is more difficult with traditional speaker materials like MDF but much easier with this 3D printed design. There were a few hiccups along the way though; while the plastic used here is much denser than MDF, the amount of infill needed to be experimented with to achieve a good finish. The parametric design paid off here as well as the original didn’t fit exactly within the print bed, so without having to split up the print the speakers’ shape was slightly tweaked instead. In the end he has a finished set of speakers that look and sound like a high-end product.

There are a few other perks to a parametric design like this as well. [Cal] can take his design for smaller desk-based speakers and tweak a few dimensions and get a model designed to stand up on the floor instead. It’s a design process that adds a lot of options and although it takes a bit more up-front effort it can be worth it while prototyping or even for producing different products quickly. If you want to make something much larger than the print bed and slightly changing the design won’t cut it, [Cal] recently showed us how to easily print huge objects like arcade cabinets with fairly standard sized 3D printers.

USB Stick Hides Large Language Model

February 17, 2025 by 25 Comments

Large language models (LLMs) are all the rage in the generative AI world these days, with the truly large ones like GPT, LLaMA, and others using tens or even hundreds of billions of parameters to churn out their text-based responses. These typically require glacier-melting amounts of computing hardware, but the “large” in “large language models” doesn’t really need to be that big for there to be a functional, useful model. LLMs designed for limited hardware or consumer-grade PCs are available now as well, but [Binh] wanted something even smaller and more portable, so he put an LLM on a USB stick.

This USB stick isn’t just a jump drive with a bit of memory on it, though. Inside the custom 3D printed case is a Raspberry Pi Zero W running llama.cpp, a lightweight, high-performance version of LLaMA. Getting it on this Pi wasn’t straightforward at all, though, as the latest version of llama.cpp is meant for ARMv8 and this particular Pi was running the ARMv6 instruction set. That meant that [Binh] needed to change the source code to remove the optimizations for the more modern ARM machines, but with a week’s worth of effort spent on it he finally got the model on the older Raspberry Pi.

Getting the model to run was just one part of this project. The rest of the build was ensuring that the LLM could run on any computer without drivers and be relatively simple to use. By setting up the USB device as a composite device which presents a filesystem to the host computer, all a user has to do to interact with the LLM is to create an empty text file with a filename, and the LLM will automatically fill the file with generated text. While it’s not blindingly fast, [Binh] believes this is the first plug-and-play USB-based LLM, and we’d have to agree. It’s not the least powerful computer to ever run an LLM, though. That honor goes to this project which is able to cram one on an ESP32.

Continue reading “USB Stick Hides Large Language Model” →

Game Bub Plays ROMs And Cartridges

February 15, 2025 by 10 Comments

With today’s technology, emulating video game consoles from the 90s or before is trivial. A Raspberry Pi and a controller of some sort is perhaps the easiest and simplest way to go to get this job done, but to really impress the masses some extra effort is required. This handheld from [Eli] called the Game Bub not only nails the appearance and feel of the first three generations of Nintendo handhelds but, thanks to its FPGA, can play not only ROMs but the original game cartridges as well.

As [Eli] notes, the FPGA is not strictly necessary for emulation, but does seem to be better at interfacing with physical hardware like controllers and game cartridges. For this task an Xilinx XC7A100T with integrated memory was chosen, with a custom PCB supporting the built-in controller, speaker, a rechargeable lithium battery, and a 480×320 display (that had to be rotated out of portrait mode). An SD Card reader is included for any ROM files, and there’s also a ESP32-S3 included to give the handheld WiFi and Bluetooth capabilities, with future plans to support the communications protocol used by the Game Boy Advance Wireless Adapter.

There are a few other features with the Game Bub as well, including the ability to use an authentic link cable to communicate with the original Game Boy and Game Boy Color, and a Switch-like dock that allows the Game Bub to be connected to an external monitor. It’s also open source, which makes it an even more impressive build. Presumably it doesn’t include the native ability to dump cartridges to ROM files but you don’t need much more than a link cable to do that if you need to build your ROM library.

Continue reading “Game Bub Plays ROMs And Cartridges” →