{"id":26384338,"url":"https://github.com/apws25/accelmoe","last_synced_at":"2026-05-11T16:37:29.941Z","repository":{"id":280108100,"uuid":"938046013","full_name":"APWS25/AccelMoE","owner":"APWS25","description":"This repository is for CUDA kernel re-implementation of CPU-based MoE model.","archived":false,"fork":false,"pushed_at":"2025-03-16T09:10:12.000Z","size":703,"stargazers_count":0,"open_issues_count":0,"forks_count":1,"subscribers_count":0,"default_branch":"master","last_synced_at":"2025-03-16T10:20:39.448Z","etag":null,"topics":["cpp","cuda","mixture-of-experts"],"latest_commit_sha":null,"homepage":"","language":"C++","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/APWS25.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":"2025-02-24T10:36:25.000Z","updated_at":"2025-03-16T09:10:15.000Z","dependencies_parsed_at":null,"dependency_job_id":"67d7534f-5e6c-4306-a756-ac18c743bf68","html_url":"https://github.com/APWS25/AccelMoE","commit_stats":null,"previous_names":["apws25/accelmoe"],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/APWS25%2FAccelMoE","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/APWS25%2FAccelMoE/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/APWS25%2FAccelMoE/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/APWS25%2FAccelMoE/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/APWS25","download_url":"https://codeload.github.com/APWS25/AccelMoE/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":243989726,"owners_count":20379649,"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":["cpp","cuda","mixture-of-experts"],"created_at":"2025-03-17T07:23:01.326Z","updated_at":"2026-05-11T16:37:29.891Z","avatar_url":"https://github.com/APWS25.png","language":"C++","funding_links":[],"categories":[],"sub_categories":[],"readme":"# AccelMoE: Accelerated Mixture-of-Expert model \nAccelMoE is a project that optimizes a CPU-based mixture-of-experts architecture into GPU-based accelerated code. The project utilizes CUDA kernel programming to effectively execute computations on the GPU. The project was awarded **3rd Place** at the Accelerator Programming School competition.\n\n\u003e [!NOTE]\n\u003e This work is a project conducted as part of the Accelerator Programming School at Seoul National University.\n\n## Optimization Overview\n![Archeture](https://github.com/user-attachments/assets/d47bdb8a-34e3-4c66-b8c4-47382f1de415)\n\n\n## Optimization Techniques\n- GPU formatting using CUDA kernel programming\n- Kernel fusion to combine Conv1D or Linear and ReLU operations\n- CUDA streaming for efficient parallel processing\n- Batch processing to maximize throughput\n- Warp occupancy optimization\n\n## Improved Performance\nAchieved a 650× speedup when executed on the GPU.\n### CPU version\n```bash\nInitializing inputs and parameters...Done!\nPredicting sentiment...Done!\nElapsed time: 1.467701 (sec)\nThroughput: 0.681338 (sentences/sec)\nFinalizing...Done!\nSaving outputs to ./data/outputs.bin...Done!\nValidating...PASSED!\n```\n\n### GPU version\n```bash\nInitializing inputs and parameters...Done!\nPredicting sentiment...Done!\nElapsed time: 0.074036 (sec)\nThroughput: 432.224966 (sentences/sec)\nFinalizing...Done!\nSaving outputs to ./data/outputs.bin...Done!\nValidating...PASSED!\n```\n\n## Contributors\n|       |       |\n| :---: | :---: |\n|\u003cimg src=\"https://avatars.githubusercontent.com/u/89632139?v=4\" style=\"width: 150px;\"\u003e|\u003cimg src=\"https://avatars.githubusercontent.com/u/113115154?v=4\" style=\"width: 150px;\"\u003e|\n|[Haeseung Jeon](https://github.com/JeonHaeseung)|[Suyeon Jo](https://github.com/Suyeonnie)|\n|@Ewha Womans Univ.|@Myongji Univ.|\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fapws25%2Faccelmoe","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fapws25%2Faccelmoe","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fapws25%2Faccelmoe/lists"}