{"id":406234,"date":"2025-01-20T16:45:36","date_gmt":"2025-01-20T16:45:36","guid":{"rendered":"https:\/\/innovationspace.ansys.com\/forum\/forums\/topic\/rockydem-multi-gpu-solver-get-low-cuda-usage-per-gpu-compare-to-single-gpu-solve\/"},"modified":"2025-01-20T16:45:36","modified_gmt":"2025-01-20T16:45:36","slug":"rockydem-multi-gpu-solver-get-low-cuda-usage-per-gpu-compare-to-single-gpu-solve","status":"publish","type":"topic","link":"https:\/\/innovationspace.ansys.com\/forum\/forums\/topic\/rockydem-multi-gpu-solver-get-low-cuda-usage-per-gpu-compare-to-single-gpu-solve\/","title":{"rendered":"RockyDEM Multi-GPU Solver get low CUDA usage per GPU compare to Single-GPU Solve"},"content":{"rendered":"<p>&lt;p&gt;I have 2 Titan V in my workstation, and when I use multi-GPU solver I found that both two GPU take lower usage. One is 59% in average and the other is 68%. But when I use single-GPU solver that usage is 97%-98% all times. It&#8217;s calculation speed can be even faster than multi-GPU solver&lt;\/p&gt;&lt;figure class=&#8221;rte-media&#8221;&gt;<a title=\"Image from r\/ANSYS - RockyDEM Multi-GPU Solver get low CUDA usage per GPU compare to Single-GPU Solver.\" href=\"https:\/\/preview.redd.it\/rockydem-multi-gpu-solver-get-low-cuda-usage-per-gpu-v0-rylnxslke6ee1.png?width=1925&amp;format=png&amp;auto=webp&amp;s=617f3bc4d63baba624ba68e520ecde990c60f729\" target=\"_blank\" rel=\"noopener noreferrer ugc\"><img decoding=\"async\" src=\"https:\/\/preview.redd.it\/rockydem-multi-gpu-solver-get-low-cuda-usage-per-gpu-v0-rylnxslke6ee1.png?width=1925&amp;format=png&amp;auto=webp&amp;s=617f3bc4d63baba624ba68e520ecde990c60f729\" alt=\"r\/ANSYS - RockyDEM Multi-GPU Solver get low CUDA usage per GPU compare to Single-GPU Solver.\" width=\"750\" \/><\/a>&lt;\/figure&gt;&lt;figure class=&#8221;rte-media&#8221;&gt;<a title=\"Image from r\/ANSYS - RockyDEM Multi-GPU Solver get low CUDA usage per GPU compare to Single-GPU Solver.\" href=\"https:\/\/preview.redd.it\/rockydem-multi-gpu-solver-get-low-cuda-usage-per-gpu-v0-75oztg7he6ee1.png?width=849&amp;format=png&amp;auto=webp&amp;s=f73a76f2e4c42c2abdf8fd78fbf667a83a17492b\" target=\"_blank\" rel=\"noopener noreferrer ugc\"><img decoding=\"async\" src=\"https:\/\/preview.redd.it\/rockydem-multi-gpu-solver-get-low-cuda-usage-per-gpu-v0-75oztg7he6ee1.png?width=849&amp;format=png&amp;auto=webp&amp;s=f73a76f2e4c42c2abdf8fd78fbf667a83a17492b\" alt=\"r\/ANSYS - RockyDEM Multi-GPU Solver get low CUDA usage per GPU compare to Single-GPU Solver.\" width=\"750\" \/><\/a>&lt;\/figure&gt;&lt;p&gt;The problem is not about the HPC license, and no matter how many particles I settle down, the usage do not change anymore. So what makes the problem happend and how can I solve the problem?&lt;\/p&gt;<\/p>\n","protected":false},"template":"","class_list":["post-406234","topic","type-topic","status-publish","hentry","topic-tag-cfd-dem","topic-tag-rocky-dem-1","topic-tag-speedup-with-gpu"],"aioseo_notices":[],"acf":[],"custom_fields":[{"0":{"_bbp_forum_id":["27792"],"_bbp_topic_id":["406234"],"_bbp_subscription":["505256"],"_bbp_author_ip":["23.158.104.250"],"_bbp_last_reply_id":["406372"],"_bbp_last_active_id":["406372"],"_bbp_last_active_time":["2025-01-21 05:50:43"],"_bbp_reply_count":["1"],"_bbp_reply_count_hidden":["0"],"_bbp_voice_count":["1"],"_bbp_engagement":["505256"],"_btv_view_count":["275"],"_bbp_topic_status":["unanswered"],"_bbp_notification_enabled":["505256"]},"test":"zhou333708124gmail-com"}],"_links":{"self":[{"href":"https:\/\/innovationspace.ansys.com\/forum\/wp-json\/wp\/v2\/topics\/406234","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/innovationspace.ansys.com\/forum\/wp-json\/wp\/v2\/topics"}],"about":[{"href":"https:\/\/innovationspace.ansys.com\/forum\/wp-json\/wp\/v2\/types\/topic"}],"version-history":[{"count":0,"href":"https:\/\/innovationspace.ansys.com\/forum\/wp-json\/wp\/v2\/topics\/406234\/revisions"}],"wp:attachment":[{"href":"https:\/\/innovationspace.ansys.com\/forum\/wp-json\/wp\/v2\/media?parent=406234"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}