2026年8月19日，Snowboard Kids反编译项目团队宣布，经典Nintendo 64游戏《Snowboard Kids》在84天内完成了100%反编译，所有函数均有匹配的C实现 1。该项目的反编译仓库公开时间跨度为2026年5月28日至2026年8月19日 1。
《Snowboard Kids》共包含2,145个函数，项目采用了IDO 5.3编译器，这与续作《Snowboard Kids 2》所使用的GCC 2.7.2编译器不同 1。作为对比，包含2,995个函数的续作《Snowboard Kids 2》反编译耗时达596天，而包含1,791个函数的《Pilotwings 64》反编译则耗时74天 1。
在反编译过程中，团队测试了GPT-5.5、GPT-5.6、Claude 4.5、Claude Fable以及GLM 5.2等多种前沿大模型与Agent工具链以加速进程，其中Codex表现最佳 1。尽管引入了AI工具，项目仍高度依赖人类专家对IDO专有编译器的深入理解，约4.8%的匹配提交涉及专家的人工干预 1。
未来，团队计划继续完善相关文档并进行静态重编译 1。此外，他们还考虑将内容移植至续作引擎，或者对PlayStation平台的《Snowboard Kids Plus》进行反编译 1。
A development team has announced the 100% decompilation of the classic Nintendo 64 game Snowboard Kids, achieving matching C implementations for all functions within 84 days 1. The public repository for this decompilation project spans from May 28, 2026, to August 19, 2026 1. In comparison, the sequel Snowboard Kids 2 contains 2,995 functions and took 596 days to decompile, whereas Snowboard Kids consists of 2,145 functions 1. For additional context, the decompilation of another N64 title, Pilotwings 64, which contains 1,791 functions, took 74 days 1.
The project accelerated its progress by leveraging cutting-edge large language models and agent toolchains, including Codex and Claude 1. Among the tested models—GPT-5.5, GPT-5.6, Claude 4.5, Claude Fable, and GLM 5.2—Codex demonstrated the best performance 1. However, approximately 4.8% of the matching commits still required manual intervention from human experts with a deep understanding of the proprietary IDO 5.3 compiler, which differs from the GCC 2.7.2 compiler used in Snowboard Kids 2 1. Looking ahead, the team plans to further refine the documentation and perform static recompilation, while also considering porting content to the sequel's engine or decompiling the PlayStation version of Snowboard Kids Plus 1.
评论
还没有评论，欢迎留下第一条。