{"id":15634805,"url":"https://github.com/nihui/ncnn-small-board","last_synced_at":"2026-01-08T03:50:13.956Z","repository":{"id":39979470,"uuid":"448444665","full_name":"nihui/ncnn-small-board","owner":"nihui","description":"ncnn benchmark on various single board computers","archived":false,"fork":false,"pushed_at":"2023-10-19T11:51:32.000Z","size":1235,"stargazers_count":157,"open_issues_count":2,"forks_count":7,"subscribers_count":12,"default_branch":"master","last_synced_at":"2025-02-04T20:35:52.111Z","etag":null,"topics":["arm","jetson","mips","ncnn","neural-network","risc-v","rockchip","sbc","single-board-computer"],"latest_commit_sha":null,"homepage":"","language":null,"has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/nihui.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null}},"created_at":"2022-01-16T03:05:48.000Z","updated_at":"2025-01-03T21:54:11.000Z","dependencies_parsed_at":"2024-10-23T03:23:24.246Z","dependency_job_id":null,"html_url":"https://github.com/nihui/ncnn-small-board","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/nihui%2Fncnn-small-board","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/nihui%2Fncnn-small-board/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/nihui%2Fncnn-small-board/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/nihui%2Fncnn-small-board/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/nihui","download_url":"https://codeload.github.com/nihui/ncnn-small-board/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":246237429,"owners_count":20745348,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["arm","jetson","mips","ncnn","neural-network","risc-v","rockchip","sbc","single-board-computer"],"created_at":"2024-10-03T10:57:46.903Z","updated_at":"2026-01-08T03:50:13.888Z","avatar_url":"https://github.com/nihui.png","language":null,"funding_links":[],"categories":["🛠️ Tools \u0026 Utilities"],"sub_categories":["🔄 Model Conversion \u0026 Editing"],"readme":"# ncnn 小板子\n\nncnn is a high-performance neural network inference framework.\n\nSee https://github.com/Tencent/ncnn for more info about ncnn.\n\n* The benchmark score is the inference time, **less is better**.\n* Runs on all cpu big cores or vulkan, and fp16 arithmetic is enabled if supported.\n* Table contents are ordered by mobilenet.\n* rpi5b data from https://github.com/Tencent/ncnn/pull/5058\n\n| small-board            | image                             | spec                                                              | squeezenet | mobilenet | shufflenet |\n| ---------------------- | --------------------------------- | ----------------------------------------------------------------- | ---------- | --------- | ---------- |\n| NVIDIA Jetson AGX Orin | ![img](/images/jetsonagxorin.jpg) | Ampere 1.3GHz x 2048\u003cbr /\u003e\u003cbr /\u003eTensor Core x 64                  | 2.13       | 2.03      | 2.57       |\n| NVIDIA Jetson AGX Orin | ![img](/images/jetsonagxorin.jpg) | A78AE 2.2GHz x 12                                                 | 3.50       | 3.49      | 5.08       |\n| Radxa Rock5B           | ![img](/images/rock5b.jpg)        | RK3588\u003cbr /\u003e\u003cbr /\u003eA76 2.4GHz x 4\u003cbr /\u003eA55 1.8GHz x 4              | 3.65       | 5.41      | 3.37       |\n| Orange Pi 5            | ![img](/images/opi5.jpg)          | RK3588S\u003cbr /\u003e\u003cbr /\u003eA76 2.4GHz x 4\u003cbr /\u003eA55 1.8GHz x 4             | 3.83       | 5.75      | 3.57       |\n| ZYSJ RK3588            |                                   | ARM Mali-G610                                                     | 7.09       | 9.16      | 5.88       |\n| ZYSJ RK3588            |                                   | RK3588\u003cbr /\u003e\u003cbr /\u003eA76 2.4GHz x 4\u003cbr /\u003eA55 1.8GHz x 4              | 7.57       | 11.01     | 7.95       |\n| Raspberry Pi 5B        |                                   | BCM2712\u003cbr /\u003e\u003cbr /\u003eA76 2.4GHz x 4                                 | 8.56       | 11.32     | 4.56       |\n| M4N DOCK               | ![img](/images/m4ndock.jpg)       | AX650\u003cbr /\u003e\u003cbr /\u003eA55 1.7Ghz x 8                                   | 10.61      | 11.86     | 10.64      |\n| NVIDIA Jetson Nano     | ![img](/images/jetsonnano.jpg)    | Maxwell 921MHz x 128                                              | 9.03       | 15.39     | 10.61      |\n| Khadas VIM4            | ![img](/images/vim4.jpg)          | ARM Mali-G52\u003cbr /\u003e\u003cbr /\u003e800MHz x MP8(8EE)                         | 12.77      | 20.17     | 12.81      |\n| Unisoc Tiger T710      | ![](/images/unisoct710.jpg)       | Unisoc Tiger T710\u003cbr /\u003e\u003cbr /\u003eA75 1.82GHz x 4\u003cbr /\u003eA55 1.82GHz x 4 | 16.25      | 20.31     | 10.59      |\n| Radxa Rock3A           | ![](/images/rock3a.jpg)           | RK3568\u003cbr /\u003e\u003cbr /\u003eA55 2.0GHz x 4                                  | 23.20      | 30.78     | 19.49      |\n| Khadas VIM3            | ![](/images/vim3.jpg)             | ARM Mali-G52\u003cbr /\u003e\u003cbr /\u003e800MHz x MP4(6EE)                         | 19.54      | 31.62     | 17.91      |\n| Panther X2             |                                   | RK3566\u003cbr /\u003e\u003cbr /\u003eA55 1.8GHz x 4                                  | 25.91      | 33.17     | 21.81      |\n| NanoPi M4              | ![](/images/nanopim4.jpg)         | ARM Mali-T864                                                     | 24.57      | 35.86     | 33.90      |\n| Radxa RockPi X         | ![](/images/rockpix.jpg)          | Intel HD Graphics (Cherry Trail) 500 MHz                          | 29.91      | 36.41     | 24.54      |\n| Khadas VIM4            | ![](/images/vim4.jpg)             | A311D2\u003cbr /\u003e\u003cbr /\u003eA73 2.2GHz x 4\u003cbr /\u003eA53 2.0GHz x 2              | 24.27      | 37.65     | 19.79      |\n| NVIDIA Jetson Nano     | ![](/images/jetsonnano.jpg)       | A57 1.43GHz x 4                                                   | 28.55      | 40.25     | 19.44      |\n| Khadas VIM3            | ![](/images/vim3.jpg)             | A311D\u003cbr /\u003e\u003cbr /\u003eA73 2.2GHz x 4\u003cbr /\u003eA53 1.8GHz x 2               | 30.98      | 42.57     | 21.62      |\n| Lichee Pi 4A           | ![](/images/lp4a.jpg)             | T-Head TH1520\u003cbr /\u003e\u003cbr /\u003eRV64GCV 2.0GHz x 4                       | 42.29      | 47.73     | 120.13     |\n| NanoPi Fire3           | ![](/images/nanopifire3.jpg)      | S5P6818\u003cbr /\u003e\u003cbr /\u003eA53 1.4GHz x 8                                 | 41.81      | 52.33     | 33.54      |\n| Debix Model A          | ![](/images/debixmodela.jpg)      | i.MX8M Plus\u003cbr /\u003e\u003cbr /\u003eA53 1.6GHz x 4                             | 40.15      | 54.01     | 29.33      |\n| Odroid XU4             | ![](/images/odroidxu4.jpg)        | Exynos 5422\u003cbr /\u003e\u003cbr /\u003eA15 2GHz x 4\u003cbr /\u003eA7 1.3GHz x 4            | 35.73      | 54.17     | 25.75      |\n| Tinker Board S R2.0    | ![](/images/tinkersr2.jpg)        | ARM Mali-T764                                                     | 41.48      | 60.28     | 55.87      |\n| Radxa Zero             | ![](/images/radxazero.jpg)        | S905Y2\u003cbr /\u003e\u003cbr /\u003eA53 1.8GHz x 4                                  | 46.82      | 60.49     | 35.18      |\n| Raspberry Pi 4B        | ![](/images/rasp4b.jpg)           | BCM2711B0\u003cbr /\u003e\u003cbr /\u003eA72 1.5GHz x 4                               | 46.28      | 60.74     | 32.91      |\n| NanoPi M4              | ![](/images/nanopim4.jpg)         | RK3399\u003cbr /\u003e\u003cbr /\u003eA72 1.8GHz x 2\u003cbr /\u003eA53 1.5GHz x 4              | 43.73      | 64.28     | 27.43      |\n| BeagleBone AI 64       | ![](/images/bbai64.jpg)           | TDA4VM\u003cbr /\u003e\u003cbr /\u003eA72 2.0GHz x 2                                  | 42.23      | 65.78     | 29.97      |\n| NanoPi K2              | ![](/images/nanopik2.jpg)         | S905\u003cbr /\u003e\u003cbr /\u003eA53 1.5GHz x 4                                    | 56.96      | 75.25     | 38.70      |\n| Radxa RockPi X         | ![](/images/rockpix.jpg)          | Atom x5-Z8350\u003cbr /\u003e\u003cbr /\u003ex86-64 1.92 GHz x 4                      | 50.22      | 80.12     | 37.96      |\n| Tinker Board S R2.0    | ![](/images/tinkersr2.jpg)        | RK3288CGW\u003cbr /\u003e\u003cbr /\u003eA17 1.8GHz x 4                               | 52.86      | 82.19     | 38.17      |\n| NanoPi R2S             | ![](/images/nanopir2s.jpg)        | RK3328\u003cbr /\u003e\u003cbr /\u003eA53 1.3GHz x 4                                  | 62.20      | 82.88     | 52.34      |\n| EAIDK 310              | ![](/images/eaidk310.jpg)         | RK3228H\u003cbr /\u003e\u003cbr /\u003eA53 1.3GHz x 4                                 | 61.87      | 83.88     | 45.83      |\n| Radxa Zero             | ![](/images/radxazero.jpg)        | ARM Mali-G31 850 MHz x 2                                          | 57.95      | 98.62     | 51.71      |\n| Raspberry Pi 3B+       | ![](/images/rasp3b.jpg)           | BCM2837B0\u003cbr /\u003e\u003cbr /\u003eA53 1.4GHz x 4                               | 84.74      | 107.84    | 58.38      |\n| Banana Pi M3           | ![](/images/bpim3.jpg)            | AllWinner A83T\u003cbr /\u003e\u003cbr /\u003eA7 1.8GHz x 8                           | 88.35      | 109.72    | 80.91      |\n| MangoPI MQ-Quad        | ![](/images/mq-quad.jpg)          | AllWinner H616\u003cbr /\u003e\u003cbr /\u003eA53 1.5GHz x 4                          | 99.60      | 129.70    | 66.90      |\n| Raspberry Pi Zero 2W   | ![](/images/raspz2.jpg)           | BCM2710A1\u003cbr /\u003e\u003cbr /\u003eA53 1.0GHz x 4                               | 119.52     | 162.60    | 72.72      |\n| EASY EAI Nano          | ![](/images/easyeainano.jpg)      | RV1126\u003cbr /\u003e\u003cbr /\u003eA7 1.5GHz x 4                                   | 105.16     | 163.82    | 68.79      |\n| Radxa RockPi S         | ![](/images/rockpis.jpg)          | RK3308\u003cbr /\u003e\u003cbr /\u003eA35 1.3GHz x 4                                  | 142.02     | 214.56    | 102.44     |\n| StarFive VisionFive V2 | ![](/images/visionfivev2.jpg)     | SiFive’s U74\u003cbr /\u003e\u003cbr /\u003eRV64GFC 1.5GHz x 4                        | 150.07     | 256.21    | 91.14      |\n| RVB-ICE                | ![](/images/ice.jpg)              | T-Head C910\u003cbr /\u003e\u003cbr /\u003eRV64GC 1.2GHz x 2                          | 161.10     | 274.39    | 118.61     |\n| Loongson Pie Lite      | ![](/images/ls2k.jpg)             | Loongson 2K1000\u003cbr /\u003e\u003cbr /\u003eGS264 1.0GHz x 2                       | 185.28     | 276.03    | 124.39     |\n| Ingenic X2000          |                                   | Ingenic X2000\u003cbr /\u003e\u003cbr /\u003eXBurst2 1.2GHz x 2                       | 175.99     | 282.25    | 118.4      |\n| Raspberry Pi 2B        | ![](/images/rasp2b.jpg)           | BCM2836\u003cbr /\u003e\u003cbr /\u003eA7 900MHz x 4                                  | 179.50     | 296.97    | 132.49     |\n| Banana Pi M2 Zero 2    | ![](/images/bananaz2.jpg)         | AllWinner H2+\u003cbr /\u003e\u003cbr /\u003eA7 1.2GHz x 4                            | 218.73     | 303.92    | 138.43     |\n| Raspberry Pi 4B        | ![](/images/rasp4b.jpg)           | Broadcom VideoCore VI                                             | 286.71     | 365.04    | 162.14     |\n| PICO-PI-IMX7           | ![](/images/pico.jpg)             | i.MX7D\u003cbr /\u003e\u003cbr /\u003eA7 1.0GHz x 2                                   | 220.10     | 366.92    | 137.02     |\n| Microphase Z7-Lite     | ![](/images/z7lite.jpg)           | XC7Z020CLG400\u003cbr /\u003e\u003cbr /\u003eA9 766MHz x 2                            | 389.18     | 623.71    | 212.20     |\n| MangoPI MQ-Pro         | ![](/images/mq-pro.jpg)           | D1 C906\u003cbr /\u003e\u003cbr /\u003eRV64GCV 1GHz x 1                               | 385.21     | 691.47    | 487.67     |\n| NeZha D1               | ![](/images/nezhad1.jpg)          | D1 C906\u003cbr /\u003e\u003cbr /\u003eRV64GCV 1GHz x 1                               | 396.18     | 727.97    | 495.87     |\n| StarFive VisionFive V1 | ![](/images/visionfivev1.jpg)     | SiFive’s U74\u003cbr /\u003e\u003cbr /\u003eRV64GFC 1.5GHz x 2                        | 609.41     | 993.87    | 358.98     |\n| Raspberry Pi B+        | ![](/images/raspbp.jpg)           | ARM1176JZF-S\u003cbr /\u003e\u003cbr /\u003e700MHz x 1                                | 4285.24    | 7684.05   | 2102.22    |\n\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fnihui%2Fncnn-small-board","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fnihui%2Fncnn-small-board","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fnihui%2Fncnn-small-board/lists"}