Metraipolis › archive › dc_972299cd9ee2
FlowTT: Exploiting Computation Flow Reuse in Irregular Tensor-Train Embedding — arXiv:2609.03459
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.03459
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-17T01:47:28.287Z
- SHA-256 of the archived text
53f408eb2f2a5146ee8cb51ae82b9c6ff45d95bb5515e88cbb226056def16f41- Stable URL of this page
https://metraipolis.com/documento/dc_972299cd9ee2
How to cite
FlowTT: Exploiting Computation Flow Reuse in Irregular Tensor-Train Embedding — arXiv:2609.03459 — arXiv (en), https://arxiv.org/abs/2609.03459 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_972299cd9ee2 (Metraipolis archive dc_972299cd9ee2, sha256 53f408eb2f2a5146ee8cb51ae82b9c6ff45d95bb5515e88cbb226056def16f41)
BibTeX
@misc{dc972299cd9ee2,
title = {FlowTT: Exploiting Computation Flow Reuse in Irregular Tensor-Train Embedding — arXiv:2609.03459},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.03459},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_972299cd9ee2; sha256 53f408eb2f2a5146ee8cb51ae82b9c6ff45d95bb5515e88cbb226056def16f41},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.03459. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Tensor-Train (TT) decomposition effectively compresses large embedding tables in recommendation models, but TT-based embedding lookup remains inefficient because partially shared computation flows across input indices are not fully reused and intermediate results are repeatedly materialized off-chip between sequential TT-core contractions. We present FlowTT, a flow-aware GPU execution framework that reformulates TT gather as a set of prefix-shared irregular computation flows. FlowTT combines flow-aligned prefix-based index grouping, a fused TT-embedding execution path with on-chip intermediate retention, and persistent-thread scheduling with chunk-based work stealing and L2 checkpointing to preserve reuse under skewed workloads. By co-designing task formation, data buffering, and scheduling with the structure of TT gather, FlowTT reduces redundant TT-core operations, global-memory traffic, and load imbalance. On Meta's synthetic recommendation benchmarks (Meta-240, Meta-480, and Meta-788), FlowTT consistently achieves the lowest latency compared to existing methods. At batch size 32,768, it reduces latency by up to 42.2% in inference and 49.2% in training relative to EcoRec, while also achieving the lowest inference peak memory usage. These results show that exposing prefix-shared computation is key to efficient TT-based embedding execution. Tema degli esploratori: calcolo e infrastrutture (retro-catalogo arXiv: annunci del 2026-09-04 in cs.DC). Categorie arXiv: Distributed, Parallel, and Cluster Computing (cs.DC) ; Hardware Architecture (cs.AR). Pagina del lavoro: https://arxiv.org/abs/2609.03459 — 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 2026-10-17T01:12:01.336Z dalla pagina https://arxiv.org/catchup/cs.DC/2026-09-04?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).