https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Strange-profiling-result-with-pardiso/m-p/963073#M16009 <DIV></DIV> <P><SPAN>Hi all</SPAN><SPAN></SPAN></P> <P><SPAN></SPAN></P> <P><SPAN>Our app uses multiple solve with the same factor, when I looked at the call graph results' it seems that about 30% of the time is in the LSAME function. </SPAN></P> <P><SPAN>70% of DTRSM ( most of <SPAN></SPAN>the SOLVE <SPAN></SPAN>work) is the blas_errchk_dtrsm, that calls LSAME.</SPAN></P> <P><SPAN></SPAN></P> <P><SPAN><SPAN></SPAN>Is that reasonable? Just some profiling artifact?<SPAN> </SPAN>Or a configuration problem ?</SPAN></P> <P><SPAN></SPAN></P> <P><SPAN>The problem size in about <SPAN></SPAN>10000.</SPAN></P> <P><SPAN></SPAN></P> <P><SPAN>Thanks</SPAN></P> <DIV></DIV> Mon, 14 Nov 2005 00:13:40 GMT dshor1 2005-11-14T00:13:40Z https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Strange-profiling-result-with-pardiso/m-p/963073#M16009 <DIV></DIV> <P><SPAN>Hi all</SPAN><SPAN></SPAN></P> <P><SPAN></SPAN></P> <P><SPAN>Our app uses multiple solve with the same factor, when I looked at the call graph results' it seems that about 30% of the time is in the LSAME function. </SPAN></P> <P><SPAN>70% of DTRSM ( most of <SPAN></SPAN>the SOLVE <SPAN></SPAN>work) is the blas_errchk_dtrsm, that calls LSAME.</SPAN></P> <P><SPAN></SPAN></P> <P><SPAN><SPAN></SPAN>Is that reasonable? Just some profiling artifact?<SPAN> </SPAN>Or a configuration problem ?</SPAN></P> <P><SPAN></SPAN></P> <P><SPAN>The problem size in about <SPAN></SPAN>10000.</SPAN></P> <P><SPAN></SPAN></P> <P><SPAN>Thanks</SPAN></P> <DIV></DIV> Mon, 14 Nov 2005 00:13:40 GMT https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Strange-profiling-result-with-pardiso/m-p/963073#M16009 dshor1 2005-11-14T00:13:40Z https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Strange-profiling-result-with-pardiso/m-p/963074#M16010 <P>Hi!</P> <P>It shouldn't spend much time a LSAME - it's just a few instructions. What is theprofiler?</P> <DIV></DIV> Mon, 14 Nov 2005 22:58:21 GMT https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Strange-profiling-result-with-pardiso/m-p/963074#M16010 michael_chuvelev 2005-11-14T22:58:21Z https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Strange-profiling-result-with-pardiso/m-p/963075#M16011 <P>I know It shouldn't ... I have a felling that the problem is calling the BLAS function with very small size, and that the reson for the large overhead.</P> <P>the profiler is intel vtune</P> <DIV></DIV> Wed, 16 Nov 2005 00:18:46 GMT https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Strange-profiling-result-with-pardiso/m-p/963075#M16011 dshor1 2005-11-16T00:18:46Z