https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/MKL-Operations-in-TensorFlow/m-p/1181805#M29371 <P>&nbsp;</P> <P><SPAN style="font-size: 1em;">Hi all,</SPAN></P> <P>I run a TensorFlow model Inception with Intel MKL-DNN support.&nbsp;</P> <P>The execution time of MKL operations in MKL (e.g. <SPAN data-sheets-userformat="{&quot;2&quot;:6659,&quot;3&quot;:[null,0],&quot;4&quot;:[null,2,16776960],&quot;12&quot;:0,&quot;14&quot;:[null,2,0],&quot;15&quot;:&quot;-webkit-standard&quot;}" data-sheets-value="{&quot;1&quot;:2,&quot;2&quot;:&quot;_MklConv2DBackpropFilter&quot;}" style="caret-color: rgb(0, 0, 0); color: rgb(0, 0, 0); font-size: 10pt; font-family: -webkit-standard, Arial;">_MklConv2DBackpropFilter,&nbsp;</SPAN><SPAN data-sheets-userformat="{&quot;2&quot;:2563,&quot;3&quot;:[null,0],&quot;4&quot;:[null,2,16776960],&quot;12&quot;:0,&quot;14&quot;:[null,2,0]}" data-sheets-value="{&quot;1&quot;:2,&quot;2&quot;:&quot;_MklConv2D&quot;}" style="caret-color: rgb(0, 0, 0); color: rgb(0, 0, 0); font-size: 10pt; font-family: Arial;">_MklConv2D,&nbsp;</SPAN><SPAN data-sheets-userformat="{&quot;2&quot;:2563,&quot;3&quot;:[null,0],&quot;4&quot;:[null,2,16776960],&quot;12&quot;:0,&quot;14&quot;:[null,2,0]}" data-sheets-value="{&quot;1&quot;:2,&quot;2&quot;:&quot;_MklConv2DBackpropInput&quot;}" style="caret-color: rgb(0, 0, 0); color: rgb(0, 0, 0); font-size: 10pt; font-family: Arial;">_MklConv2DBackpropInput, etc</SPAN>) have no change when I change the number of intra-threads (means the number of threads running one operation). The results are as below. Does anyone know why the MKL the performance of MKL operations will not change when changing the number of threads running this operation? Or does anyone explain the implementation of MKL-DNN support for TensorFlow for this situation? Thank you!</P> <P><SPAN style="font-size: 1em;">Name&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;Intra-threads &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Time</SPAN></P> <P>_MklConv2DBackpropFilter &nbsp; &nbsp; 8&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;470<BR /> &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;17&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;466<BR /> &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;34&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;468<BR /> &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;68&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;467</P> <P>_MklConv2DBackpropInput &nbsp; &nbsp; 8&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;354<BR /> &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;17&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;344<BR /> &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;34&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;347<BR /> &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;68&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;347<BR /> &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;<BR /> _MklConv2D &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;8&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;300<BR /> &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;17&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;304<BR /> &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;34&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;311<BR /> &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;68&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;311</P> <P>Kevin</P> Fri, 11 May 2018 21:04:56 GMT Liu__Jiawen 2018-05-11T21:04:56Z https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/MKL-Operations-in-TensorFlow/m-p/1181805#M29371 <P>&nbsp;</P> <P><SPAN style="font-size: 1em;">Hi all,</SPAN></P> <P>I run a TensorFlow model Inception with Intel MKL-DNN support.&nbsp;</P> <P>The execution time of MKL operations in MKL (e.g. <SPAN data-sheets-userformat="{&quot;2&quot;:6659,&quot;3&quot;:[null,0],&quot;4&quot;:[null,2,16776960],&quot;12&quot;:0,&quot;14&quot;:[null,2,0],&quot;15&quot;:&quot;-webkit-standard&quot;}" data-sheets-value="{&quot;1&quot;:2,&quot;2&quot;:&quot;_MklConv2DBackpropFilter&quot;}" style="caret-color: rgb(0, 0, 0); color: rgb(0, 0, 0); font-size: 10pt; font-family: -webkit-standard, Arial;">_MklConv2DBackpropFilter,&nbsp;</SPAN><SPAN data-sheets-userformat="{&quot;2&quot;:2563,&quot;3&quot;:[null,0],&quot;4&quot;:[null,2,16776960],&quot;12&quot;:0,&quot;14&quot;:[null,2,0]}" data-sheets-value="{&quot;1&quot;:2,&quot;2&quot;:&quot;_MklConv2D&quot;}" style="caret-color: rgb(0, 0, 0); color: rgb(0, 0, 0); font-size: 10pt; font-family: Arial;">_MklConv2D,&nbsp;</SPAN><SPAN data-sheets-userformat="{&quot;2&quot;:2563,&quot;3&quot;:[null,0],&quot;4&quot;:[null,2,16776960],&quot;12&quot;:0,&quot;14&quot;:[null,2,0]}" data-sheets-value="{&quot;1&quot;:2,&quot;2&quot;:&quot;_MklConv2DBackpropInput&quot;}" style="caret-color: rgb(0, 0, 0); color: rgb(0, 0, 0); font-size: 10pt; font-family: Arial;">_MklConv2DBackpropInput, etc</SPAN>) have no change when I change the number of intra-threads (means the number of threads running one operation). The results are as below. Does anyone know why the MKL the performance of MKL operations will not change when changing the number of threads running this operation? Or does anyone explain the implementation of MKL-DNN support for TensorFlow for this situation? Thank you!</P> <P><SPAN style="font-size: 1em;">Name&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;Intra-threads &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Time</SPAN></P> <P>_MklConv2DBackpropFilter &nbsp; &nbsp; 8&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;470<BR /> &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;17&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;466<BR /> &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;34&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;468<BR /> &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;68&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;467</P> <P>_MklConv2DBackpropInput &nbsp; &nbsp; 8&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;354<BR /> &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;17&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;344<BR /> &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;34&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;347<BR /> &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;68&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;347<BR /> &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;<BR /> _MklConv2D &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;8&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;300<BR /> &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;17&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;304<BR /> &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;34&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;311<BR /> &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;68&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;311</P> <P>Kevin</P> Fri, 11 May 2018 21:04:56 GMT https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/MKL-Operations-in-TensorFlow/m-p/1181805#M29371 Liu__Jiawen 2018-05-11T21:04:56Z https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/MKL-Operations-in-TensorFlow/m-p/1181806#M29372 <P>Hi Kevin,</P> <P>What is Hardware (processor) you running?<BR /> and Also like&nbsp; <A href="https://github.com/intel/mkl-dnn/issues/234"><U><FONT color="#0066cc"></FONT></U></A><A href="https://github.com/intel/mkl-dnn/issues/234" target="_blank">https://github.com/intel/mkl-dnn/issues/234</A>&nbsp; ..&nbsp; , &nbsp;i recommend you&nbsp; submit your issue to&nbsp; <A href="https://github.com/intel/mkl-dnn/issues">https://github.com/intel/mkl-dnn/issues</A>&nbsp;&nbsp;where our developer will help you directly.<BR /> <BR /> Best Regards,</P> <P>Ying</P> Mon, 14 May 2018 02:05:12 GMT https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/MKL-Operations-in-TensorFlow/m-p/1181806#M29372 Ying_H_Intel 2018-05-14T02:05:12Z