<DIV><includetail>
<DIV>Hello , everyone</DIV>
<DIV>Thanks for your kind advice.</DIV>
<DIV>Of course, I have not ignored those three posts.And I have accepted their suggestions.</DIV>
<DIV>And I'm very sorry I bother everybody again.</DIV>
<DIV>I think I have compiled both float and double precision for fftw.</DIV>
<DIV>And I just choose to compile double precision for gromacs.</DIV>
<DIV>Do Mark's means that I must compile gromacs in both float and double precision ?</DIV>
<DIV>But I don't think that is essential. Because I have successfully compiled(both parallel version and single thread version) gromacs on another computer(it's a cluster) in this way.</DIV>
<DIV> </DIV>
<DIV>So, I think this is a problem of this computer.</DIV>
<DIV>Altix450 have 8 compute nodes ,each node have two <FONT size=2>Itanium2 </FONT>64bit Dual Cores CPU.</DIV>
<DIV>And it is a machine of shared memery.</DIV>
<DIV> </DIV>
<DIV>Now, I am trying Carsten's advice, I hope it can work normally.</DIV>
<DIV> </DIV>
<DIV>Thanks everyone's good advice again.</DIV>
<DIV>Best wishes.</DIV>
<DIV> </DIV>
<DIV>Wenping Lv</DIV></DIV></includetail>