{"id":29798,"date":"2025-03-03T14:08:31","date_gmt":"2025-03-03T12:08:31","guid":{"rendered":"https:\/\/hgpu.org\/?p=29798"},"modified":"2025-03-03T14:08:31","modified_gmt":"2025-03-03T12:08:31","slug":"pyatf-constraint-based-auto-tuning-in-python","status":"publish","type":"post","link":"https:\/\/hgpu.org\/?p=29798","title":{"rendered":"pyATF: Constraint-Based Auto-Tuning in Python"},"content":{"rendered":"<p>We introduce pyATF &#8211; a new, language-independent, open-source auto-tuning tool that fully automatically determines optimized values of performance-critical program parameters. A major feature of pyATF is its support for constrained parameters, e.g., the value of one parameter has to divide the value of another parameter. A further major feature of pyATF is its user interface which is designed with a particular focus on expressivity and usability for real-world demands, and which is offered in the increasingly popular Python programming language. We experimentally confirm the practicality of pyATF using real-world studies from the areas of quantum chemistry, image processing, data mining, and deep learning: we show that pyATF auto-tunes the complex parallel implementations of our studies to higher performance than achieved by state-of-practice approaches, including hand-optimized vendor libraries.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>We introduce pyATF &#8211; a new, language-independent, open-source auto-tuning tool that fully automatically determines optimized values of performance-critical program parameters. A major feature of pyATF is its support for constrained parameters, e.g., the value of one parameter has to divide the value of another parameter. A further major feature of pyATF is its user interface [&hellip;]<\/p>\n","protected":false},"author":351,"featured_media":0,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_jetpack_memberships_contains_paid_content":false,"footnotes":"","jetpack_publicize_message":"","jetpack_publicize_feature_enabled":true,"jetpack_social_post_already_shared":true,"jetpack_social_options":{"image_generator_settings":{"template":"highway","default_image_id":0,"font":"","enabled":false},"version":2}},"categories":[11,89,90,3],"tags":[1856,955,1782,14,2066,1793,176,67,513],"class_list":["post-29798","post","type-post","status-publish","format-standard","hentry","category-computer-science","category-nvidia-cuda","category-opencl","category-paper","tag-auto-tuning","tag-compilers","tag-computer-science","tag-cuda","tag-nvidia-a100","tag-opencl","tag-package","tag-performance","tag-python"],"views":1548,"jetpack_publicize_connections":[],"jetpack_featured_media_url":"","jetpack_sharing_enabled":true,"_links":{"self":[{"href":"https:\/\/hgpu.org\/index.php?rest_route=\/wp\/v2\/posts\/29798","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/hgpu.org\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/hgpu.org\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/hgpu.org\/index.php?rest_route=\/wp\/v2\/users\/351"}],"replies":[{"embeddable":true,"href":"https:\/\/hgpu.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=29798"}],"version-history":[{"count":0,"href":"https:\/\/hgpu.org\/index.php?rest_route=\/wp\/v2\/posts\/29798\/revisions"}],"wp:attachment":[{"href":"https:\/\/hgpu.org\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=29798"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/hgpu.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=29798"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/hgpu.org\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=29798"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}