<div>Hi! Pavan Balaji</div>Thanks for your quick reply!<div>Here is the information on the error:</div><div>--Fatal error in PMPI_BARRIER:Other MPI error, error stack</div><div>--MPI_Barrier(MPI_COMM_WORLD) failed</div><div>--Failure during collective</div><div>--connect failed -- the semaphore timeout period has expired.</div><div><br></div><div>The difference of the time consuming of the two process is about several minutes.</div><div>Is it the reason that caused the connect failed?<br><div>--<br><p><span style="font-size: 14px; "><font class="Apple-style-span" face="'Times New Roman'">Email:&nbsp;&nbsp;&nbsp;&nbsp; </font><a href="mailto:yogikai@163.com"><font class="Apple-style-span" face="'Times New Roman'">yogikai@163.com</font></a><font class="Apple-style-span" face="'Times New Roman'"><br>Address: Room 921, Automation Building,<br>No. 95 Zhongguancun East Road,<br>Haidian District, Beijing 100190, China<br>Cell Phone:15210370340<br></font><span class="Apple-style-span" style="line-height: 15px; color: rgb(42, 42, 42); "><font class="Apple-style-span" style="font-size: 14px; "><font class="Apple-style-span" face="'Times New Roman'">The&nbsp;State Key Laboratory&nbsp;for Intelligent Control and Management&nbsp;of Complex Systems</font></font></span><font class="Apple-style-span" face="'Times New Roman'"><br>Institute of Automation, Chinese Academy of Sciences</font></span><br></p>        
</div><br><pre><br>At&nbsp;2011-06-29&nbsp;10:15:46£¬"Pavan&nbsp;Balaji"&nbsp;&lt;balaji@mcs.anl.gov&gt;&nbsp;wrote:

&gt;
&gt;There's&nbsp;no&nbsp;time&nbsp;limit&nbsp;in&nbsp;barrier.&nbsp;The&nbsp;problem&nbsp;might&nbsp;be&nbsp;something&nbsp;else.&nbsp;
&gt;Which&nbsp;version&nbsp;of&nbsp;MPICH2&nbsp;are&nbsp;you&nbsp;using?&nbsp;Also&nbsp;can&nbsp;you&nbsp;send&nbsp;some&nbsp;more&nbsp;
&gt;information&nbsp;on&nbsp;the&nbsp;error?
&gt;
&gt;&nbsp;&nbsp;--&nbsp;Pavan
&gt;
&gt;On&nbsp;06/28/2011&nbsp;09:11&nbsp;PM,&nbsp;Íõ¿­&nbsp;wrote:
&gt;&gt;&nbsp;Hi!
&gt;&gt;&nbsp;I&nbsp;have&nbsp;two&nbsp;computer&nbsp;in&nbsp;the&nbsp;MPI&nbsp;program,&nbsp;and&nbsp;I&nbsp;use&nbsp;the&nbsp;function
&gt;&gt;&nbsp;mpi_barrier(MPI_COMM_WORLD)&nbsp;to&nbsp;synchronize&nbsp;the&nbsp;two&nbsp;process&nbsp;on&nbsp;the&nbsp;two
&gt;&gt;&nbsp;computer.
&gt;&gt;&nbsp;One&nbsp;of&nbsp;the&nbsp;process&nbsp;is&nbsp;running&nbsp;more&nbsp;slower&nbsp;than&nbsp;the&nbsp;other&nbsp;one&nbsp;because&nbsp;two
&gt;&gt;&nbsp;different&nbsp;GPUs&nbsp;is&nbsp;on&nbsp;the&nbsp;two&nbsp;computer.&nbsp;I&nbsp;got&nbsp;"fatal&nbsp;error&nbsp;in
&gt;&gt;&nbsp;PMPI_Barrier"&nbsp;and&nbsp;"connect&nbsp;failed--the&nbsp;semaphore&nbsp;timeout&nbsp;period&nbsp;has
&gt;&gt;&nbsp;expired".
&gt;&gt;&nbsp;Is&nbsp;there&nbsp;a&nbsp;time&nbsp;limitation&nbsp;in&nbsp;the&nbsp;mpi_barrier()?
&gt;&gt;
&gt;&gt;&nbsp;--
&gt;&gt;
&gt;&gt;&nbsp;Email:&nbsp;yogikai@163.com&nbsp;&lt;mailto:yogikai@163.com&gt;
&gt;&gt;&nbsp;Address:&nbsp;Room&nbsp;921,&nbsp;Automation&nbsp;Building,
&gt;&gt;&nbsp;No.&nbsp;95&nbsp;Zhongguancun&nbsp;East&nbsp;Road,
&gt;&gt;&nbsp;Haidian&nbsp;District,&nbsp;Beijing&nbsp;100190,&nbsp;China
&gt;&gt;&nbsp;Cell&nbsp;Phone:15210370340
&gt;&gt;&nbsp;The&nbsp;State&nbsp;Key&nbsp;Laboratory&nbsp;for&nbsp;Intelligent&nbsp;Control&nbsp;and&nbsp;Management&nbsp;of
&gt;&gt;&nbsp;Complex&nbsp;Systems
&gt;&gt;&nbsp;Institute&nbsp;of&nbsp;Automation,&nbsp;Chinese&nbsp;Academy&nbsp;of&nbsp;Sciences
&gt;&gt;
&gt;&gt;
&gt;&gt;
&gt;&gt;
&gt;&gt;
&gt;&gt;&nbsp;_______________________________________________
&gt;&gt;&nbsp;mpich-discuss&nbsp;mailing&nbsp;list
&gt;&gt;&nbsp;mpich-discuss@mcs.anl.gov
&gt;&gt;&nbsp;https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss
&gt;
&gt;--&nbsp;
&gt;Pavan&nbsp;Balaji
&gt;http://www.mcs.anl.gov/~balaji
</pre></div><br><br><span title="neteasefooter"><span id="netease_mail_footer"></span></span>