{"id":8138,"date":"2012-09-01T01:05:25","date_gmt":"2012-08-31T22:05:25","guid":{"rendered":"http:\/\/hgpu.org\/?p=8138"},"modified":"2012-09-01T01:05:25","modified_gmt":"2012-08-31T22:05:25","slug":"a-gpu-support-for-large-scale-quantum-chemistry-applications","status":"publish","type":"post","link":"https:\/\/hgpu.org\/?p=8138","title":{"rendered":"A GPU Support for Large Scale Quantum Chemistry Applications"},"content":{"rendered":"<p>GPU\/GPGPU computing has been used widely in scientific simulation to improve the performance on hybrid architectures. The quantum chemistry field has benefited greatly from using GPUs, including tasks such as visualization of molecular orbitals and computation of electronic structures. To gain significant success in using GPUs, a large amount of code rewriting and restructuring is required, which is done primarily by those who understand the algorithm in great detail. In this paper, two widely used quantum chemistry packages are investigated to identify the hot spots that can benefit most from GPUs, as well as be the least intrusive to the existing code base. The paper uses an experimental approach to integrate GPU capability without restructuring the application. Experimental results show that the bottleneck is in CPU-GPU data transmission. Additionally, a GPU-based DFTFOCK method is implemented in GAMESS\/NWChem and a GPU-based eigensolver is integrated with NWChem successfully. Further performance tuning is ongoing.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>GPU\/GPGPU computing has been used widely in scientific simulation to improve the performance on hybrid architectures. The quantum chemistry field has benefited greatly from using GPUs, including tasks such as visualization of molecular orbitals and computation of electronic structures. To gain significant success in using GPUs, a large amount of code rewriting and restructuring is [&hellip;]<\/p>\n","protected":false},"author":351,"featured_media":0,"comment_status":"open","ping_status":"open","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":false,"jetpack_social_options":{"image_generator_settings":{"template":"highway","default_image_id":0,"font":"","enabled":false},"version":2}},"categories":[66,89,3],"tags":[1790,14,20,1006,134],"class_list":["post-8138","post","type-post","status-publish","format-standard","hentry","category-chemistry","category-nvidia-cuda","category-paper","tag-chemistry","tag-cuda","tag-nvidia","tag-tesla-c2070","tag-visualization"],"views":3052,"jetpack_publicize_connections":[],"jetpack_featured_media_url":"","jetpack_sharing_enabled":true,"_links":{"self":[{"href":"https:\/\/hgpu.org\/index.php?rest_route=\/wp\/v2\/posts\/8138","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=8138"}],"version-history":[{"count":0,"href":"https:\/\/hgpu.org\/index.php?rest_route=\/wp\/v2\/posts\/8138\/revisions"}],"wp:attachment":[{"href":"https:\/\/hgpu.org\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=8138"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/hgpu.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=8138"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/hgpu.org\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=8138"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}