Metraipolis › archive › dc_16531ebdd72c
KerColle: Unlocking Fine-Grained GPU Concurrency in Vision-Language-Action Models — arXiv:2609.22335
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.22335
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-09T06:55:51.164Z
- SHA-256 of the archived text
d7cb05540c0d595a7ebcad883e2416332c575abe55fddfd10468b7d5eaa7ad97- Stable URL of this page
https://metraipolis.com/documento/dc_16531ebdd72c
How to cite
KerColle: Unlocking Fine-Grained GPU Concurrency in Vision-Language-Action Models — arXiv:2609.22335 — arXiv (en), https://arxiv.org/abs/2609.22335 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_16531ebdd72c (Metraipolis archive dc_16531ebdd72c, sha256 d7cb05540c0d595a7ebcad883e2416332c575abe55fddfd10468b7d5eaa7ad97)
BibTeX
@misc{dc16531ebdd72c,
title = {KerColle: Unlocking Fine-Grained GPU Concurrency in Vision-Language-Action Models — arXiv:2609.22335},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.22335},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_16531ebdd72c; sha256 d7cb05540c0d595a7ebcad883e2416332c575abe55fddfd10468b7d5eaa7ad97},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.22335. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Vision-Language-Action (VLA) models have emerged as foundational models for next-generation robotics. High VLA inference throughput is critical for meeting the control-rate requirements of robots. VLA models comprise two phases, a vision-language model (VLM) and an action head, that can be decoupled and executed asynchronously and concurrently across independent robot requests. Through a detailed characterization of four state-of-the-art VLAs, we observe that GPUs are severely underutilized in VLA inference as the Cooperative Thread Array (CTA) scheduler of GPU is unable to fully overlap the two independent phases of VLA execution. We identify that this inefficiency is caused by head-of-line blocking in the hardware thread-block dispatcher. We demonstrate that prior scheduling frameworks do not address the challenges posed by VLA concurrency. First, the independent phases across different robot requests each comprise numerous kernels, and at any given time, there are different combinations of kernels that are executed in parallel. This makes static or ahead-of-time scheduling policies largely ineffective. Second, many of the action-head operators are short-running kernels and there are numerous such kernels. This leaves no headroom for online profiling or preemption-based mechanisms. To address these challenges, we present KerColle, a lightweight GPU scheduling framework that leverages online Streaming Multiprocessor (SM) utilization and individual kernel resource requirements to intelligently and dynamically co-schedule kernels to efficiently overlap the two phases of execution by (1) mitigating head-of-line blocking, and (2) co-scheduling kernels with complementary resource requirements. We demonstrate in simulation, across two GPU architectures, for 4 state-of-the-art VLA models, that KerColle delivers throughput gains of up to $28\%$. Tema degli esploratori: calcolo e infrastrutture (retro-catalogo arXiv: annunci del 2026-09-22 in cs.AR). Categorie arXiv: Hardware Architecture (cs.AR) ; Robotics (cs.RO). Pagina del lavoro: https://arxiv.org/abs/2609.22335 — 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 2026-10-09T06:50:43.557Z dalla pagina https://arxiv.org/catchup/cs.AR/2026-09-22?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).