RoofLang: Enabling AI-Driven Architecting of LLM Inference Systems
This matters because it introduces a new tool, RoofLang, that could significantly improve the performance of LLMs like DeepSeek V4-series by optimizing their inference architecture, potentially leading to faster processing and better user experience.
RoofLang is a new domain-specific language (DSL) designed to help AI optimize LLM inference systems. It provides a general workload representation, a verifiable mutation space, and an implementation-independent evaluator. In an evaluation, RoofLang showed that DeepSeek V4-series models could achieve 3.5-39.5x higher peak decode throughput than other models. A persistent optimizer agent also discovered new architectures that improved DeepSeek V4 Pro's throughput and interactivity on NVIDIA B300 by 6.23-50.1%.
AI is beginning to make substantive contributions to LLM inference optimization. Existing AI optimizations are predominantly profiling-based. Profiling-bound feedback confines the search to the capabilities and performance of an existing software stack, preventing a fundamentally better architecture of LLM inference systems from being identified. To enable the AI-driven LLM inference system architecting loop, we argue that a general workload representation, a verifiable mutation space, and an implementation-independent evaluator are required. We present the RoofLang domain-specific language (DSL) that provides these features. In our evaluation, RoofLang reveals that DeepSeek V4-series models could achieve 3.5-39.5$\times$ higher peak decode throughput than other representative models. This gap is disproportionate to their total parameter counts and arises largely from compact KV-cache designs that support larger batches and reduce memory traffic. A persistent optimizer agent further discovered several new architectures that improved both throughput and interactivity of DeepSeek V4 Pro on NVIDIA B300 by 6.23-50.1%.
- 宝玉09-12 18:10原文