Metraipolis › archive › dc_300cce32edd1

RoofLang: Enabling AI-Driven Architecting of LLM Inference Systems — arXiv:2609.12551

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.12551
License (as archived)
CC0-1.0
Archived
2026-10-16T03:03:28.894Z
SHA-256 of the archived text
49581cbc9d9e2e2902bcc0fa4f85fb1eb9536e954cfcac6700719a5121f5fcc2
Stable URL of this page
https://metraipolis.com/documento/dc_300cce32edd1

How to cite

RoofLang: Enabling AI-Driven Architecting of LLM Inference Systems — arXiv:2609.12551 — arXiv (en), https://arxiv.org/abs/2609.12551 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_300cce32edd1 (Metraipolis archive dc_300cce32edd1, sha256 49581cbc9d9e2e2902bcc0fa4f85fb1eb9536e954cfcac6700719a5121f5fcc2)
BibTeX
@misc{dc300cce32edd1,
  title = {RoofLang: Enabling AI-Driven Architecting of LLM Inference Systems — arXiv:2609.12551},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.12551},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_300cce32edd1; sha256 49581cbc9d9e2e2902bcc0fa4f85fb1eb9536e954cfcac6700719a5121f5fcc2},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.12551. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

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%.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-14 in cs.AI).
Categorie arXiv: Distributed, Parallel, and Cluster Computing (cs.DC) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.12551 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-16T02:51:27.922Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-14?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).