<html><head><meta http-equiv="Content-Type" content="text/html; charset=us-ascii"></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; line-break: after-white-space;" class="">It depends on the solver used. What solver are you using?<br class=""><div><br class=""><blockquote type="cite" class=""><div class="">On Jul 11, 2022, at 5:33 PM, Ce Qin <<a href="mailto:qince168@gmail.com" class="">qince168@gmail.com</a>> wrote:</div><br class="Apple-interchange-newline"><div class=""><div dir="ltr" class=""><div dir="ltr" class=""><div dir="ltr" class="">Dear all,<div class=""><br class=""></div><div class="">I want to analyze the strong scaling of our in-house FEM code.</div><div class="">The test problem has about 20M DoFs. I ran the problem using</div><div class="">various settings. The speedups for the assembly and solving</div><div class="">procedures are as follows:</div><div class=""><div class=""><font face="monospace" class="">                                   Assembly     Solving</font></div><div class=""><font face="monospace" class="">NProcessors NNodes CoresPerNode                        </font></div><div class=""><font face="monospace" class="">1           1      1                    1.0         1.0</font></div><div class=""><font face="monospace" class="">2           1      2               1.995246    1.898756</font></div><div class=""><font face="monospace" class="">            2      1               2.121401    2.436149</font></div><div class=""><font face="monospace" class="">4           1      4               4.658187    6.004539</font></div><div class=""><font face="monospace" class="">            2      2               4.666667    5.942085</font></div><div class=""><font face="monospace" class="">            4      1                4.65272    6.101214</font></div><div class=""><font face="monospace" class="">8           2      4               9.380985   16.581135</font></div><div class=""><font face="monospace" class="">            4      2               9.308575   17.258891</font></div><div class=""><font face="monospace" class="">            8      1               9.314449   17.380612</font></div><div class=""><font face="monospace" class="">16          2      8              18.575953   34.483058</font></div><div class=""><font face="monospace" class="">            4      4              18.745129   34.854409</font></div><div class=""><font face="monospace" class="">            8      2              18.828393    36.45509</font></div><div class=""><font face="monospace" class="">32          4      8              37.140626   70.175879</font></div><div class=""><font face="monospace" class="">            8      4              37.166421   71.533865</font></div></div><div class=""><font face="monospace" class=""><br class=""></font></div><div class=""><font face="arial, sans-serif" class="">I don't quite understand this result. Why we can achieve</font><font face="arial, sans-serif" class=""> a speedup of</font></div><div class=""><font face="arial, sans-serif" class="">about </font><span style="font-family:arial,sans-serif" class="">70+ using 32 processors? Could you please help me explain this?</span></div><div class=""><br class=""></div><div class=""><font face="arial, sans-serif" class="">Thank you in advance.</font></div><div class=""><font face="arial, sans-serif" class=""><br class=""></font></div><div class=""><font face="arial, sans-serif" class="">Best,</font></div><div class=""><font face="arial, sans-serif" class="">Ce</font></div><div class=""><font face="monospace" class=""><br class=""></font></div><div class=""><font face="monospace" class=""><br class=""></font></div></div></div></div>
</div></blockquote></div><br class=""></body></html>