doubleAI launches Maxwell, a runtime code optimization model
What happened
On October 19, Hacker News' front page covered doubleAI's launch of Maxwell, a code execution model. Maxwell optimizes code based on the hardware and data present at runtime, and it formally proves equivalence for each change it makes. It runs on CPUs and GPUs; once it has collected enough runtime data, it swaps in code tuned to the current machine while the program keeps running. The page claims a 38x speedup for data processing pipelines and 4.25x for AI inference kernels, and says LTX-2.5 cuts 4K video generation from 250 seconds to 134 seconds per clip on two B200s.
Written by AI from the coverage · updated 2 hours ago
Coverage
Follow the reports to see the story from different sides.
- Hacker News front pageMaxwell: A new kind of model that sits between code and hardware
doubleAI 推出代码执行模型 Maxwell，根据程序运行时的硬件与数据优化代码，并对每项改进进行形式化等价证明。它在 CPU 和 GPU 上运行，收集足够运行时数据后，在程序持续运行期间替换为针对当前机器的优化代码。页面称数据处理流水线加速 38 ×、AI 推理内核加速 4.25 ×，LTX-2.5 在两块 B200 上生成每段 4K 视频的耗时从 250s 降至 134 s。
Heat over time
Not enough continuous observations to draw a trend yet.