![](\"http:\/\/blogs.uct.ac.za\/gallery\/1253\/Capture.PNG\")
<\/p>\n","protected":false},"author":3,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":[],"categories":[4],"tags":[],"yoast_head":"\n![\"\"](\"https:\/\/ucthpc.uct.ac.za\/wp-content\/uploads\/2015\/07\/Ansyscomp.png\")
","protected":false},"excerpt":{"rendered":"Jonathan Bergh conducted the following benchmarks against our facilities. Tests were run on the HEX 64core AMD 6376 machine to test the
\nscaling of a standardised CFD problem over an increasing no. of cores.
\nThe test mesh was a turbomachinery blade passage and exhaust solving for
\n a low speed, incompressible, turbulent flow. A medium size mesh of
\n~2.2M cells was used and flow was solved using a pressure-based, coupled
\n solver where all flow variables are solved in a blocked-coupled fashion
\n and the additional turbulence variables were solved for in a segregated
\n fashion. The coupled approach requires x2.5-3.5 times the amount of RAM
\n than a traditional segregated approach. Total mesh RAM requirements
\nwere ~8GB which is not large but emphasis was on total solution time as
\nthe optimisation algorithm which which will be coupled to the CFD
\nproject to need (conservatively) 150-200 solutions to be calculated.<\/p>\n
The tests scaled to jobs over 16 \/ 32 \/ 48 and 64 cores on the
\n ICTS Dell (Dell C6145 rack server). Total solution time was then
\ncompared to two existing nodes (32) core based on the AMD 6276 and 6376
\nCPU’s (HP Proliant DL385p rack server). All machines use DDR3-1600MHz
\nRAM and Intel MPI. The test was not disk read-write intensive as the
\nsimulations performed were steady-state and so only minimal amounts of
\ndata were written out to disk during each iteratio.<\/p>\n
<\/p>\n","protected":false},"author":3,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":[],"categories":[4],"tags":[],"yoast_head":"\n