Developer Details How He Built a Custom Bootloader Taking an OS from BIOS to 64-bit Mode

·3 views

A developer building a hobby operating system called Nyvela has documented the process of writing a custom bootloader from scratch using legacy BIOS firmware. The bootloader is divided into two stages: Stage 1 handles disk loading, memory mapping, and A20 gate enabling before transitioning the CPU from 16-bit Real Mode to 32-bit Protected Mode. Stage 2 then configures paging, sets up descriptor tables, and enables Long Mode, allowing the system to run a 64-bit kernel. The developer chose legacy BIOS over the more modern UEFI standard because it is the boot path Nyvela currently supports. The project offers a detailed look at low-level x86 architecture concepts such as memory protection, privilege levels, and the step-by-step CPU mode transitions required before an operating system kernel can execute.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Poetry: A Smarter Way to Manage Python Project Dependencies

Poetry is an open-source package management tool for Python that simplifies project setup, dependency handling, and environment isolation. It automatically generates configuration files like pyproject.toml and poetry.lock, making projects easier to share and reproduce across systems. Developers can create a new project using a single command, then add, update, or remove dependencies with straightforward CLI instructions. Poetry also creates isolated virtual environments per project, preventing version conflicts between different codebases. Its built-in dependency resolver reduces compatibility issues, making it a practical choice for both beginner and experienced Python developers.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Prompt Engineering or Fine-Tuning? How to Know When to Switch

Teams building LLM applications typically begin with prompt engineering, which requires no infrastructure, is instantly reversible, and costs little compared to model training. Prompt engineering works by modifying the input to the model, while fine-tuning directly updates the model's weights to embed desired behaviour by default. Experts recommend staying in prompt-engineering mode until evaluation scores stop improving across three to four consecutive changes, signalling a performance plateau. Fine-tuning becomes viable only when the task is narrow, thousands of real labelled examples are available, and cost or latency constraints make a leaner model preferable. Common warning signs that prompting has hit its limit include diminishing eval gains, conflicting instructions, bloated system prompts, and the same corrections being resent on every API call.

0 comments Read more at DEV Community

ProgrammingHacker News ·

French High Schools Test Whether Reading Newspapers Shapes Teen Habits

A nationwide experiment was conducted in French high schools to study the effects of making teenagers read newspapers regularly. The initiative, documented in a CEPR discussion paper, aimed to examine whether structured news consumption influences students' reading habits, civic awareness, or academic outcomes. The study represents a large-scale policy intervention targeting adolescents across France. Researchers sought to understand the broader educational and social implications of integrating print media into school curricula.

0 comments Read more at Hacker News

ProgrammingHacker News ·

Developer Builds Interactive Tool to Visualize LLM Attention Mechanisms

A developer has released an open-access tool that visually demonstrates how large language models (LLMs) distribute attention across input tokens. The project, shared on Hacker News under 'Show HN', allows users to explore the internal attention patterns that influence how LLMs process and respond to text. The tool is hosted at ishamf.dev and appears aimed at making model interpretability more accessible to researchers and curious users. Such visualizations can help demystify how transformer-based models weigh relationships between words or tokens during inference.

0 comments Read more at Hacker News