{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,2]],"date-time":"2026-06-02T17:12:34Z","timestamp":1780420354174,"version":"3.54.1"},"publisher-location":"New York, NY, USA","reference-count":28,"publisher":"ACM","license":[{"start":{"date-parts":[[2026,6,20]],"date-time":"2026-06-20T00:00:00Z","timestamp":1781913600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/legalcode"}],"content-domain":{"domain":["dl.acm.org"],"crossmark-restriction":true},"short-container-title":[],"published-print":{"date-parts":[[2026,6,21]]},"DOI":"10.1145\/3812836.3814781","type":"proceedings-article","created":{"date-parts":[[2026,6,2]],"date-time":"2026-06-02T16:54:53Z","timestamp":1780419293000},"page":"322-327","update-policy":"https:\/\/doi.org\/10.1145\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["Phase Matters: Characterizing Heterogeneous Vision-Language Inference on a Mobile SoC"],"prefix":"10.1145","author":[{"ORCID":"https:\/\/orcid.org\/0009-0005-0313-4690","authenticated-orcid":false,"given":"Aryama V","family":"Murthy","sequence":"first","affiliation":[{"name":"International Institute of Information Technology Hyderabad, Hyderabad, Telangana, India"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0008-6658-878X","authenticated-orcid":false,"given":"Yashas N","family":"Kotre","sequence":"additional","affiliation":[{"name":"International Institute of Information Technology Hyderabad, Hyderabad, Telangana, India"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0007-5988-5658","authenticated-orcid":false,"given":"Prathmesh","family":"Sharma","sequence":"additional","affiliation":[{"name":"International Institute of Information Technology Hyderabad, Hyderabad, Telangana, India"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0002-3446-7488","authenticated-orcid":false,"given":"Pragya","family":"Mishra","sequence":"additional","affiliation":[{"name":"International Institute of Information Technology Hyderabad, Hyderabad, Telangana, India"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0005-5397-5578","authenticated-orcid":false,"given":"Sanjith","family":"Ganapathi","sequence":"additional","affiliation":[{"name":"International Institute of Information Technology Hyderabad, Hyderabad, Telangana, India"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-9868-8944","authenticated-orcid":false,"given":"Priyesh","family":"Shukla","sequence":"additional","affiliation":[{"name":"International Institute of Information Technology Hyderabad, Hyderabad, Telangana, India"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"320","published-online":{"date-parts":[[2026,6,20]]},"reference":[{"key":"e_1_3_2_1_1_1","volume-title":"FastVLM: Efficient vision encoding for vision language models,\" CVPR","author":"Bhatt P.","year":"2025","unstructured":"P. Bhatt et al., \"FastVLM: Efficient vision encoding for vision language models,\" CVPR, 2025."},{"key":"e_1_3_2_1_2_1","volume-title":"MobileVLM: A fast, reproducible and strong vision language assistant for mobile devices,\" arXiv:2312.16886","author":"Chu X.","year":"2023","unstructured":"X. Chu et al., \"MobileVLM: A fast, reproducible and strong vision language assistant for mobile devices,\" arXiv:2312.16886, 2023."},{"key":"e_1_3_2_1_3_1","volume-title":"LiteVLM: Lightweight vision-language models for edge deployment,\" arXiv:2406.07496","author":"Yuan H.","year":"2024","unstructured":"H. Yuan et al., \"LiteVLM: Lightweight vision-language models for edge deployment,\" arXiv:2406.07496, 2024."},{"key":"e_1_3_2_1_4_1","volume-title":"A highly capable language model locally on your phone,\" arXiv:2404.14219","year":"2024","unstructured":"Microsoft Research, \"Phi-3 technical report: A highly capable language model locally on your phone,\" arXiv:2404.14219, 2024."},{"key":"e_1_3_2_1_5_1","volume-title":"An image is worth 16x16 words: Transformers for image recognition at scale,\" ICLR","author":"Dosovitskiy A.","year":"2021","unstructured":"A. Dosovitskiy et al., \"An image is worth 16x16 words: Transformers for image recognition at scale,\" ICLR, 2021."},{"key":"e_1_3_2_1_6_1","volume-title":"NanoVLM: Efficient vision-language models you can train and deploy on the edge,\" arXiv:2504.05299","author":"Lauren\u00e7on H.","year":"2025","unstructured":"H. Lauren\u00e7on et al., \"NanoVLM: Efficient vision-language models you can train and deploy on the edge,\" arXiv:2504.05299, 2025."},{"key":"e_1_3_2_1_7_1","unstructured":"Google DeepMind \"Gemma 3n: An efficient multimodal model for edge devices \" Technical Report 2025."},{"key":"e_1_3_2_1_8_1","volume-title":"Qwen technical report,\" arXiv:2309.16609","author":"Bai J.","year":"2023","unstructured":"J. Bai et al., \"Qwen technical report,\" arXiv:2309.16609, 2023."},{"key":"e_1_3_2_1_9_1","volume-title":"Qwen2-VL: Enhancing vision-language model's perception of the world at any resolution,\" arXiv:2409.12191","author":"Wang P.","year":"2024","unstructured":"P. Wang et al., \"Qwen2-VL: Enhancing vision-language model's perception of the world at any resolution,\" arXiv:2409.12191, 2024."},{"key":"e_1_3_2_1_10_1","volume-title":"Learning transferable visual models from natural language supervision,\" ICML","author":"Radford A.","year":"2021","unstructured":"A. Radford et al., \"Learning transferable visual models from natural language supervision,\" ICML, 2021."},{"key":"e_1_3_2_1_11_1","volume-title":"Sigmoid loss for language image pre-training,\" ICCV","author":"Zhai X.","year":"2023","unstructured":"X. Zhai et al., \"Sigmoid loss for language image pre-training,\" ICCV, 2023."},{"key":"e_1_3_2_1_12_1","unstructured":"Qualcomm Technologies Inc. \"Snapdragon 8 Elite mobile platform product brief \" 2024."},{"key":"e_1_3_2_1_13_1","unstructured":"Qualcomm Technologies Inc. \"Unlocking on-device generative AI with an NPU and heterogeneous computing \" 2024."},{"key":"e_1_3_2_1_14_1","unstructured":"Qualcomm Technologies Inc. \"Qualcomm AI Engine direct: QNN SDK documentation \" 2024."},{"key":"e_1_3_2_1_15_1","volume-title":"Lightweight runtime for on-device inference,\" https:\/\/ai.google.dev\/edge\/litert","year":"2024","unstructured":"Google, \"LiteRT: Lightweight runtime for on-device inference,\" https:\/\/ai.google.dev\/edge\/litert, 2024."},{"key":"e_1_3_2_1_16_1","volume-title":"Fast on-device LLM inference with NPUs,\" ASPLOS","author":"Xu D.","year":"2025","unstructured":"D. Xu et al., \"Fast on-device LLM inference with NPUs,\" ASPLOS, 2025."},{"key":"e_1_3_2_1_17_1","volume-title":"Characterizing mobile SoC for accelerating heterogeneous LLM inference,\" SOSP","author":"Chen L.","year":"2025","unstructured":"L. Chen et al., \"Characterizing mobile SoC for accelerating heterogeneous LLM inference,\" SOSP, 2025."},{"key":"e_1_3_2_1_18_1","volume-title":"NNV12: An efficient on-device inference framework for mobile vision tasks,\" MobiSys","author":"Cao Z.","year":"2024","unstructured":"Z. Cao et al., \"NNV12: An efficient on-device inference framework for mobile vision tasks,\" MobiSys, 2024."},{"key":"e_1_3_2_1_19_1","volume-title":"PASK: Preloaded-and-shared kernel for fast ondevice DNN inference,\" MobiCom","author":"Liu H.","year":"2024","unstructured":"H. Liu et al., \"PASK: Preloaded-and-shared kernel for fast ondevice DNN inference,\" MobiCom, 2024."},{"key":"e_1_3_2_1_20_1","volume-title":"LLM.int8(): 8-bit matrix multiplication for transformers at scale,\" NeurIPS","author":"Dettmers T.","year":"2022","unstructured":"T. Dettmers et al., \"LLM.int8(): 8-bit matrix multiplication for transformers at scale,\" NeurIPS, 2022."},{"key":"e_1_3_2_1_21_1","volume-title":"AWQ: Activation-aware weight quantization for LLM compression and acceleration,\" MLSys","author":"Lin J.","year":"2024","unstructured":"J. Lin et al., \"AWQ: Activation-aware weight quantization for LLM compression and acceleration,\" MLSys, 2024."},{"key":"e_1_3_2_1_22_1","volume-title":"FlashAttention: Fast and memory-efficient exact attention with IO-awareness,\" NeurIPS","author":"Dao T.","year":"2022","unstructured":"T. Dao et al., \"FlashAttention: Fast and memory-efficient exact attention with IO-awareness,\" NeurIPS, 2022."},{"key":"e_1_3_2_1_23_1","volume-title":"GQA: Training generalized multi-query transformer models from multi-head checkpoints,\" EMNLP","author":"Ainslie J.","year":"2023","unstructured":"J. Ainslie et al., \"GQA: Training generalized multi-query transformer models from multi-head checkpoints,\" EMNLP, 2023."},{"key":"e_1_3_2_1_24_1","volume-title":"RoFormer: Enhanced transformer with rotary position embedding,\" Neurocomputing","author":"Su J.","year":"2024","unstructured":"J. Su et al., \"RoFormer: Enhanced transformer with rotary position embedding,\" Neurocomputing, 2024."},{"key":"e_1_3_2_1_25_1","volume-title":"On-device AI across mobile, embedded and edge for PyTorch,\" https:\/\/pytorch.org\/executorch\/","year":"2024","unstructured":"Meta AI, \"ExecuTorch: On-device AI across mobile, embedded and edge for PyTorch,\" https:\/\/pytorch.org\/executorch\/, 2024."},{"key":"e_1_3_2_1_26_1","volume-title":"MLC-LLM: Universal LLM deployment engine with ML compilation,\" arXiv:2404.09542","author":"Chen T.","year":"2024","unstructured":"T. Chen et al., \"MLC-LLM: Universal LLM deployment engine with ML compilation,\" arXiv:2404.09542, 2024."},{"key":"e_1_3_2_1_27_1","unstructured":"Google \"TensorFlow Lite delegates \" https:\/\/www.tensorflow.org\/lite\/performance\/delegates 2024."},{"key":"e_1_3_2_1_28_1","doi-asserted-by":"publisher","DOI":"10.1145\/1498765.1498785"}],"event":{"name":"MobiSys Workshop '26: 24th Annual International Conference on Mobile Systems, Applications and Services Workshops","location":"University of Cambridge Cambridge United Kingdom","acronym":"MobiSys Workshop '26","sponsor":["SIGMOBILE ACM Special Interest Group on Mobility of Systems, Users, Data and Computing","SIGOPS ACM Special Interest Group on Operating Systems"]},"container-title":["Proceedings of the 24th Annual International Conference on Mobile Systems, Applications and Services Workshops"],"original-title":[],"deposited":{"date-parts":[[2026,6,2]],"date-time":"2026-06-02T16:57:18Z","timestamp":1780419438000},"score":1,"resource":{"primary":{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3812836.3814781"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,6,20]]},"references-count":28,"alternative-id":["10.1145\/3812836.3814781","10.1145\/3812836"],"URL":"https:\/\/doi.org\/10.1145\/3812836.3814781","relation":{},"subject":[],"published":{"date-parts":[[2026,6,20]]},"assertion":[{"value":"2026-06-20","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}