{"id":205,"date":"2024-01-04T19:17:17","date_gmt":"2024-01-04T19:17:17","guid":{"rendered":"https:\/\/mvermeulen.org\/perf\/?page_id=205"},"modified":"2024-01-05T10:11:24","modified_gmt":"2024-01-05T10:11:24","slug":"blender","status":"publish","type":"page","link":"https:\/\/mvermeulen.org\/perf\/workloads\/phoronix\/blender\/","title":{"rendered":"blender"},"content":{"rendered":"\n<p>Performance of blender open source modeling. This one has a longer runtime, so tested the CPU versions of both barbershop and BMW27. You can see the slightly different profiles below though the overall ordering is similar.<\/p>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"1280\" height=\"960\" src=\"https:\/\/mvermeulen.org\/perf\/wp-content\/uploads\/sites\/7\/2024\/01\/amdtopdown-11.png\" alt=\"\" class=\"wp-image-206\" srcset=\"https:\/\/mvermeulen.org\/perf\/wp-content\/uploads\/sites\/7\/2024\/01\/amdtopdown-11.png 1280w, https:\/\/mvermeulen.org\/perf\/wp-content\/uploads\/sites\/7\/2024\/01\/amdtopdown-11-1024x768.png 1024w, https:\/\/mvermeulen.org\/perf\/wp-content\/uploads\/sites\/7\/2024\/01\/amdtopdown-11-768x576.png 768w\" sizes=\"auto, (max-width: 1280px) 100vw, 1280px\" \/><\/figure>\n\n\n\n<p>AMD metrics show a relatively high branch misprediction rate. The code has a fair amount of floating point and a medium level of branches.<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>elapsed              4301.794\non_cpu               0.988          # 15.81 \/ 16 cores\nutime                67988.943\nstime                40.657\nnvcsw                93218          # 14.22%\nnivcsw               562276         # 85.78%\ninblock              23304          # 5.42\/sec\nonblock              14896          # 3.46\/sec\ncpu-clock            68033985522749 # 68033.986 seconds\ntask-clock           68034524981502 # 68034.525 seconds\npage faults          15706060       # 230.854\/sec\ncontext switches     676776         # 9.948\/sec\ncpu migrations       7545           # 0.111\/sec\nmajor page faults    188            # 0.003\/sec\nminor page faults    15705872       # 230.851\/sec\nalignment faults     0              # 0.000\/sec\nemulation faults     0              # 0.000\/sec\nbranches             35293093083193 # 110.727 branches per 1000 inst\nbranch misses        774791487242   # 2.20% branch miss\nconditional          23242796360354 # 72.921 conditional branches per 1000 inst\nindirect             3186917069241  # 9.999 indirect branches per 1000 inst\ncpu-cycles           277679476584227 # 4.03 GHz\ninstructions         318780360725662 # 1.15 IPC\nslots                555266778052170 #\nretiring             112176312261701 # 20.2% (27.1%)\n-- ucode             590561413008   #     0.1%\n-- fastpath          111585750848693 #    20.1%\nfrontend             110144593394679 # 19.8% (26.6%)\n-- latency           73569126550464 #    13.2%\n-- bandwidth         36575466844215 #     6.6%\nbackend              175308374563980 # 31.6% (42.3%)\n-- cpu               68402426527667 #    12.3%\n-- memory            106905948036313 #    19.3%\nspeculation          16699046042329 #  3.0% ( 4.0%)\n-- branch mispredict 16262790880833 #     2.9%\n-- pipeline restart  436255161496   #     0.1%\nsmt-contention       140938216831180 # 25.4% ( 0.0%)\ncpu-cycles           278774223790977 # 4.03 GHz\ninstructions         318786777948843 # 1.14 IPC\ninstructions         106255663700201 # 63.306 l2 access per 1000 inst\nl2 hit from l1       6330267339905  # 8.18% l2 miss\nl2 miss from l1      354154464303   #\nl2 hit from l2 pf    200240252262   #\nl3 hit from l2 pf    144019200939   #\nl3 miss from l2 pf   52074843805    #\ninstructions         106229525090169 # 350.889 float per 1000 inst\nfloat 512            47             # 0.000 AVX-512 per 1000 inst\nfloat 256            1188           # 0.000 AVX-256 per 1000 inst\nfloat 128            37274753429769 # 350.889 AVX-128 per 1000 inst\nfloat MMX            0              # 0.000 MMX per 1000 inst\nfloat scalar         188            # 0.000 scalar per 1000 inst\n<\/code><\/pre>\n\n\n\n<p>The corresponding Intel metrics show an even higher level of branch misprediction.<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>elapsed              6447.702\non_cpu               0.989          # 15.83 \/ 16 cores\nutime                102048.975\nstime                30.174\nnvcsw                86509          # 12.32%\nnivcsw               615573         # 87.68%\ninblock              81288          # 12.61\/sec\nonblock              15128          # 2.35\/sec\ncpu-clock            102082685570860 # 102082.686 seconds\ntask-clock           102083155121057 # 102083.155 seconds\npage faults          15755958       # 154.344\/sec\ncontext switches     734082         # 7.191\/sec\ncpu migrations       20384          # 0.200\/sec\nmajor page faults    1398           # 0.014\/sec\nminor page faults    15754560       # 154.331\/sec\nalignment faults     0              # 0.000\/sec\nemulation faults     0              # 0.000\/sec\nbranches             35287152996927 # 110.702 branches per 1000 inst\nbranch misses        1227056009226  # 3.48% branch miss\nconditional          35287153019519 # 110.702 conditional branches per 1000 inst\nindirect             10658847038465 # 33.439 indirect branches per 1000 inst\nslots                450860864981234 #\nretiring             184934949459959 # 41.0% (41.0%)\n-- ucode             14363091429244 #     3.2%\n-- fastpath          170571858030715 #    37.8%\nfrontend             133446042742152 # 29.6% (29.6%)\n-- latency           70249860801628 #    15.6%\n-- bandwidth         63196181940524 #    14.0%\nbackend              69647195865492 # 15.4% (15.4%)\n-- cpu               32408832374285 #     7.2%\n-- memory            37238363491207 #     8.3%\nspeculation          59237600057642 # 13.1% (13.1%)\n-- branch mispredict 58474019541106 #    13.0%\n-- pipeline restart  763580516536   #     0.2%\nsmt-contention       0              #  0.0% ( 0.0%)\ncpu-cycles           289398136086046 # 2.79 GHz\ninstructions         342048750860200 # 1.18 IPC\nl2 access            11054117043179 # 62.596 l2 access per 1000 inst\nl2 miss              1604086759113  # 14.51% l2 miss\n<\/code><\/pre>\n\n\n\n<p>We had a crash towards end of the process tree, but overall both blender and jemalloc_bg_thd take most of the time<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>583 processes\n\t283 blender              2261250.75  1204.25\n\t 20 jemalloc_bg_thd      190773.20   104.68\n\t 19 vulkaninfo               0.38     0.57\n\t  2 vulkani:disk$0           0.04     0.06\n\t  6 clang                    0.03     0.04\n\t  1 llvmpipe-0               0.02     0.03\n\t  1 llvmpipe-1               0.02     0.03\n\t  1 llvmpipe-10              0.02     0.03\n\t  1 llvmpipe-11              0.02     0.03\n\t  1 llvmpipe-12              0.02     0.03\n\t  1 llvmpipe-13              0.02     0.03\n\t  1 llvmpipe-14              0.02     0.03\n\t  1 llvmpipe-15              0.02     0.03\n\t  1 llvmpipe-2               0.02     0.03\n\t  1 llvmpipe-3               0.02     0.03\n\t  1 llvmpipe-4               0.02     0.03\n\t  1 llvmpipe-5               0.02     0.03\n\t  1 llvmpipe-6               0.02     0.03\n\t  1 llvmpipe-7               0.02     0.03\n\t  1 llvmpipe-8               0.02     0.03\n\t  1 llvmpipe-9               0.02     0.03\n\t 72 sh                       0.00     0.00\n\t 12 gcc                      0.00     0.00\n\t  9 stty                     0.00     0.00\n\t  8 systemd-detect-          0.00     0.00\n\t  7 gsettings                0.00     0.00\n\t  7 stat                     0.00     0.00\n\t  6 llvm-link                0.00     0.00\n\t  6 xdg-user-dir             0.00     0.00\n\t  5 gmain                    0.00     0.00\n\t  5 rm                       0.00     0.00\n\t  4 phoronix-test-s          0.00     0.00\n\t  3 dconf worker             0.00     0.00\n\t  3 glxinfo                  0.00     0.00\n\t  2 which                    0.00     0.00\n\t  1 date                     0.00     0.00\n\t  1 dirname                  0.00     0.00\n\t  1 grep                     0.00     0.00\n\t  1 ifconfig                 0.00     0.00\n\t  1 ip                       0.00     0.00\n\t  1 lscpu                    0.00     0.00\n\t  1 mktemp                   0.00     0.00\n\t  1 ps                       0.00     0.00\n\t  1 readlink                 0.00     0.00\n\t  1 realpath                 0.00     0.00\n\t  1 sed                      0.00     0.00\n\t  1 setterm                  0.00     0.00\n\t  1 sort                     0.00     0.00\n\t  1 systemctl                0.00     0.00\n\t  1 template.sh              0.00     0.00\n\t  1 wc                       0.00     0.00\n<\/code><\/pre>\n\n\n\n<p>The core parts of the computation have a repeating pattern with several processes started, one per core.<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>      561138) blender start=285.39 finish=421.43\n        561139) blender start=285.39 finish=421.41\n          561140) jemalloc_bg_thd start=285.44 finish=421.41\n            561162) jemalloc_bg_thd start=285.58 finish=421.41\n            561166) jemalloc_bg_thd start=285.58 finish=421.41\n            561171) jemalloc_bg_thd start=285.58 finish=421.41\n          561141) blender start=285.46 finish=421.39\n          561142) blender start=285.46 finish=421.39\n          561143) blender start=285.46 finish=421.39\n          561144) blender start=285.46 finish=421.39\n          561145) blender start=285.46 finish=421.39\n          561146) blender start=285.46 finish=421.39\n          561147) blender start=285.46 finish=421.39\n          561148) blender start=285.46 finish=421.39\n          561149) blender start=285.46 finish=421.39\n          561150) blender start=285.46 finish=421.39\n          561151) blender start=285.46 finish=421.39\n          561152) blender start=285.46 finish=421.39\n          561153) blender start=285.46 finish=421.39\n          561154) blender start=285.46 finish=421.39\n          561155) blender start=285.46 finish=421.39\n          561156) blender start=285.46 finish=421.39\n          561157) sh start=285.48 finish=285.48\n            561158) xdg-user-dir start=285.48 finish=285.48\n          561159) blender start=285.58 finish=421.41\n            561161) blender start=285.58 finish=421.41\n              561165) blender start=285.58 finish=421.41\n              561170) blender start=285.58 finish=421.41\n                561175) blender start=285.60 finish=421.41\n                561176) blender start=285.60 finish=421.41\n            561164) blender start=285.58 finish=421.41\n              561169) ?? start=285.58 finish=0.00 \n                561174) blender start=285.60 finish=421.41\n              561173) blender start=285.58 finish=421.41\n          561160) blender start=285.58 finish=421.41\n            561163) blender start=285.58 finish=421.41\n              561168) blender start=285.58 finish=421.41\n              561172) blender start=285.58 finish=421.41\n            561167) blender start=285.58 finish=421.41\n              561178) blender start=285.81 finish=421.40\n              561179) blender start=285.81 finish=421.40\n              561180) blender start=285.81 finish=421.40\n              561181) blender start=285.81 finish=421.40\n              561182) blender start=285.81 finish=421.40\n              561183) blender start=285.81 finish=421.40\n              561184) blender start=285.81 finish=421.40\n              561185) blender start=285.82 finish=421.40\n              561186) blender start=285.82 finish=421.40\n              561187) blender start=285.82 finish=421.40\n              561188) blender start=285.82 finish=421.40\n              561189) blender start=285.82 finish=421.40\n              561190) blender start=285.82 finish=421.40\n              561191) blender start=285.82 finish=421.40\n              561192) blender start=285.82 finish=421.40\n          561177) blender start=285.77 finish=421.12\n          561195) blender start=421.13 finish=421.37\n          561196) blender start=421.13 finish=421.37\n          561197) blender start=421.13 finish=421.37\n          561198) blender start=421.13 finish=421.37\n          561199) blender start=421.13 finish=421.37\n          561200) blender start=421.13 finish=421.37\n          561201) blender start=421.13 finish=421.37\n          561202) blender start=421.13 finish=421.37\n          561203) blender start=421.13 finish=421.37\n          561204) blender start=421.13 finish=421.37\n          561205) blender start=421.13 finish=421.37\n          561206) blender start=421.13 finish=421.37\n          561207) blender start=421.13 finish=421.37\n          561208) blender start=421.13 finish=421.37\n          561209) blender start=421.13 finish=421.37\n          561210) blender start=421.13 finish=421.37\n        561211) rm start=421.43 finish=421.43\n      561212) sh start=421.44 finish=421.44\n        561213) sh start=421.44 finish=421.44<\/code><\/pre>\n","protected":false},"excerpt":{"rendered":"<p>Performance of blender open source modeling. This one has a longer runtime, so tested the CPU versions of both barbershop and BMW27. You can see the slightly different profiles below though the overall ordering is similar. AMD metrics show a <span class=\"excerpt-dots\">&hellip;<\/span> <a class=\"more-link\" href=\"https:\/\/mvermeulen.org\/perf\/workloads\/phoronix\/blender\/\"><span class=\"more-msg\">Continue reading &rarr;<\/span><\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"parent":58,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"footnotes":""},"class_list":["post-205","page","type-page","status-publish","hentry"],"_links":{"self":[{"href":"https:\/\/mvermeulen.org\/perf\/wp-json\/wp\/v2\/pages\/205","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/mvermeulen.org\/perf\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/mvermeulen.org\/perf\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/mvermeulen.org\/perf\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/mvermeulen.org\/perf\/wp-json\/wp\/v2\/comments?post=205"}],"version-history":[{"count":1,"href":"https:\/\/mvermeulen.org\/perf\/wp-json\/wp\/v2\/pages\/205\/revisions"}],"predecessor-version":[{"id":207,"href":"https:\/\/mvermeulen.org\/perf\/wp-json\/wp\/v2\/pages\/205\/revisions\/207"}],"up":[{"embeddable":true,"href":"https:\/\/mvermeulen.org\/perf\/wp-json\/wp\/v2\/pages\/58"}],"wp:attachment":[{"href":"https:\/\/mvermeulen.org\/perf\/wp-json\/wp\/v2\/media?parent=205"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}