{"id":18930,"date":"2019-06-09T12:33:16","date_gmt":"2019-06-09T09:33:16","guid":{"rendered":"https:\/\/hgpu.org\/?p=18930"},"modified":"2019-06-09T12:33:16","modified_gmt":"2019-06-09T09:33:16","slug":"ppopencl-a-performance-portable-opencl-compiler-with-host-and-kernel-thread-code-fusion","status":"publish","type":"post","link":"https:\/\/hgpu.org\/?p=18930","title":{"rendered":"PPOpenCL: a performance-portable OpenCL compiler with host and kernel thread code fusion"},"content":{"rendered":"<p>OpenCL offers code portability but no performance portability. Given an OpenCL program X specifically written for one platform P, existing OpenCL compilers, which usually optimize its host and kernel codes individually, often yield poor performance for another platform Q. Instead of obtaining a performance-improved version of X for Q via manual tuning, we aim to achieve this automatically by a source-to-source OpenCL compiler framework, PPOpenCL. By fusing X&#8217;s host and kernel thread codes (with the operations in different work-items in the same work-group represented explicitly), we are able to apply data flow analyses, and subsequently, performance-enhancing optimizations on a fused control flow graph specifically for platformQ. Validation against OpenCL benchmarks shows that PPOpenCL (implemented in Clang 3.9.1) can achieve significantly improved portable performance on seven platforms considered.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>OpenCL offers code portability but no performance portability. Given an OpenCL program X specifically written for one platform P, existing OpenCL compilers, which usually optimize its host and kernel codes individually, often yield poor performance for another platform Q. Instead of obtaining a performance-improved version of X for Q via manual tuning, we aim to [&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":[11,90,3],"tags":[215,955,1782,1793],"class_list":["post-18930","post","type-post","status-publish","format-standard","hentry","category-computer-science","category-opencl","category-paper","tag-code-generation","tag-compilers","tag-computer-science","tag-opencl"],"views":2105,"jetpack_publicize_connections":[],"jetpack_featured_media_url":"","jetpack_sharing_enabled":true,"_links":{"self":[{"href":"https:\/\/hgpu.org\/index.php?rest_route=\/wp\/v2\/posts\/18930","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=18930"}],"version-history":[{"count":0,"href":"https:\/\/hgpu.org\/index.php?rest_route=\/wp\/v2\/posts\/18930\/revisions"}],"wp:attachment":[{"href":"https:\/\/hgpu.org\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=18930"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/hgpu.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=18930"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/hgpu.org\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=18930"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}