Metraipolis › archive › dc_16ff2071e333
Hybrid GPU-CPU Retrieval for Personalized Search at Ultra-Large Scale — arXiv:2609.21281
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.21281
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-10T01:40:10.056Z
- SHA-256 of the archived text
1ee884b4b9a22f1835924b8a59cd0b74120bbb0caf285cf52cfa148f0eadfeec- Stable URL of this page
https://metraipolis.com/documento/dc_16ff2071e333
How to cite
Hybrid GPU-CPU Retrieval for Personalized Search at Ultra-Large Scale — arXiv:2609.21281 — arXiv (en), https://arxiv.org/abs/2609.21281 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_16ff2071e333 (Metraipolis archive dc_16ff2071e333, sha256 1ee884b4b9a22f1835924b8a59cd0b74120bbb0caf285cf52cfa148f0eadfeec)
BibTeX
@misc{dc16ff2071e333,
title = {Hybrid GPU-CPU Retrieval for Personalized Search at Ultra-Large Scale — arXiv:2609.21281},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.21281},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_16ff2071e333; sha256 1ee884b4b9a22f1835924b8a59cd0b74120bbb0caf285cf52cfa148f0eadfeec},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.21281. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Embedding-based retrieval on user-generated content at the trillion-document scale exposes a sharp conflict between two production demands: deep, expressive personalization for queries with rich user intent, and broad coverage of a massive inventory under fixed latency and resource budgets. We characterize this as the personalization-scale paradox: hosting the full serving inventory in GPU memory is too resource intensive, while CPU compute cannot execute the same interaction-heavy model on the latency-critical path. We present a hybrid GPU-CPU co-serving system that resolves the paradox through orchestration rather than a new model class. A high-depth GPU pathway fuses retrieval and interaction pre-ranking over a curated online pool on the order of a billion documents, while a high-breadth CPU pathway searches an independently selected online inventory roughly twenty times larger with lightweight personalized scoring. Either or both pathways can run per request; candidates are deduplicated before shared downstream ranking. The system is deployed in production. A full-system A/B test against the legacy CPU-only configuration improves model-scored relevance and substantive engagement, while separate pathway experiments show positive value at their own deployment scopes. Retrieval logs show that the pathways contribute structurally distinct candidates, production serving measurements characterize their latency, and a matched capacity plan quantifies the economic rationale for assigning modeling depth to GPUs and inventory breadth to CPUs. Together, these results validate a practical, independently evolvable depth-breadth architecture for ultra-large-scale personalized search. Tema degli esploratori: calcolo e infrastrutture (retro-catalogo arXiv: annunci del 2026-09-21 in cs.DC). Categorie arXiv: Information Retrieval (cs.IR) ; Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG); Performance (cs.PF). Pagina del lavoro: https://arxiv.org/abs/2609.21281 — 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 2026-10-10T01:05:30.979Z dalla pagina https://arxiv.org/catchup/cs.DC/2026-09-21?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).