<div dir="ltr">Hi, Michael<div><br><div>Thank you for your advice.<br><br></div><div>As you suggested, the wrong values were generated because I run tests on a single core.</div><div>In the output file, there <span id="fd919541-c17b-4a9c-a45f-20036cd4f328" ginger_software_uiphraseguid="8b765057-271c-4cad-b6c6-5b3af92929db" class="GINGER_SOFTWARE_mark">exists</span> warnings:</div><div><div>*** 13:04:46 WARNING in tmc_setup<span id="dca39f94-fc22-43ac-95f3-5cacefb46908" ginger_software_uiphraseguid="dbcf1199-768b-4b99-91fa-9559a8332e39" class="GINGER_SOFTWARE_mark">:</span>tmc_redistributing_cores<span id="01982630-fb9e-432f-b3d7-80c42cb4f71e" ginger_software_uiphraseguid="dbcf1199-768b-4b99-91fa-9559a8332e39" class="GINGER_SOFTWARE_mark"> :</span>: TMC need at ***</div><div> *** least 2 cores (one for master, one for <span id="f5d94686-bd4a-47ae-9984-4daa8f95685d" ginger_software_uiphraseguid="0fcdb3dd-3fc1-4b7b-a63e-c3599ffc87b1" class="GINGER_SOFTWARE_mark">worker</span>) ***</div><div><br></div><div>TMC|NOTenoughProcessorsX= -999</div><div>TMC|NOTcalculatedTotal energy: -999</div></div><div><br></div><div>When I run <span id="524b4f8a-f8d3-4ee6-b1a0-f37261ee732d" ginger_software_uiphraseguid="079e7c6a-b040-4902-8beb-a1aaccefd6b9" class="GINGER_SOFTWARE_mark">test</span> again on multiple core, these wrong results <span id="808262bb-2a9b-447b-85af-7612f54f1cb2" ginger_software_uiphraseguid="079e7c6a-b040-4902-8beb-a1aaccefd6b9" class="GINGER_SOFTWARE_mark">disappered</span>.</div><div><br></div><div>Thanks for your help!.</div><div><br></div><div>Yusuke</div><div><br><br>On Monday, 14 April 2014 20:23:07 UTC+9, Michael wrote<span id="82c6f67f-779c-4c92-8327-101550cc7a44" ginger_software_uiphraseguid="1e124267-cbdc-4485-bf75-1129a710df78" class="GINGER_SOFTWARE_mark">:</span><blockquote class="gmail_quote" style="margin: 0;margin-left: 0.8ex;border-left: 1px #ccc solid;padding-left: 1ex;">Hi,
<br>
<br>On Sun, Apr 13, 2014 at 10:00:00PM -0700, Kaneta Yuske wrote:
<br>> I recently compiled <span id="7c8c2726-5936-4db4-a619-67e5bf9a32b7" ginger_software_uiphraseguid="f53aee90-e33b-4b34-8652-b67653424061" class="GINGER_SOFTWARE_mark">popt</span> version of CP2K 2.5.1 with <span id="09ed3fcd-e3be-4029-b1ff-03f97ca5e669" ginger_software_uiphraseguid="f53aee90-e33b-4b34-8652-b67653424061" class="GINGER_SOFTWARE_mark">gfortran</span> 4.8.2 and open
<br>> MPI 1.8 on TSUBAME 2.5 (arch file attached)
<br>>
<br>> After compilation, I run <span id="6c3391b1-0d6b-4d9d-846c-e1da301b358c" ginger_software_uiphraseguid="f176c5b9-8a47-4592-a55e-b881c0d5331d" class="GINGER_SOFTWARE_mark">regtest</span> with reference to CP2k 2.5 branch uploaded
<br>> <span id="47532c7b-947f-49c9-a25e-34b85810dd2c" ginger_software_uiphraseguid="cfd022de-2c3b-4640-9fc5-f068692669ee" class="GINGER_SOFTWARE_mark">to</span> <span id="8723c96c-f498-4fc0-90a6-d9f360006324" ginger_software_uiphraseguid="cfd022de-2c3b-4640-9fc5-f068692669ee" class="GINGER_SOFTWARE_mark">sourceforge</span>, but I got 98 wrong results out of 2469 tests.
<br>>
<br>> Most of the wrong results seemed to come from math library difference.
<br>> However, TMC tests, which account for 23 out of 98 wrong results, showed
<br>> <span id="c7bcb37c-5afc-49ac-9fa1-9736a353c1de" ginger_software_uiphraseguid="c4506e8f-9873-4e0e-ab84-0eefa8b1e01f" class="GINGER_SOFTWARE_mark">totally</span> different <span id="5b373af8-80bf-4dfe-94b3-77511e46f76b" ginger_software_uiphraseguid="c4506e8f-9873-4e0e-ab84-0eefa8b1e01f" class="GINGER_SOFTWARE_mark">value</span> like -999. Besides, Most of the test related to TMC
<br>> (23 out of 27) ended with wrong results.
<br>>
<br>> I don??<span id="59cb28c2-ac34-4cc7-914d-3aa289458675" ginger_software_uiphraseguid="55be9572-3910-42a9-b019-95cc4f21e651" class="GINGER_SOFTWARE_mark">?</span>t plan to use Monte Carlo, but I want to know whether these wrong
<br>> <span id="8d035b6a-ad18-463e-83dc-543c532bb220" ginger_software_uiphraseguid="4bee548e-ca80-45af-8c43-b9b7d292acfe" class="GINGER_SOFTWARE_mark">results</span> are related to my compiling procedure. Does anyone help me?
<br>
<br>In my experience, this happens at least partially if you run the
<br><span id="db95854f-013d-49a8-adc7-c2a1f7dcfbcb" ginger_software_uiphraseguid="efdc6f79-60b8-4ecd-a8d2-6c1afb9cd7b0" class="GINGER_SOFTWARE_mark">regtests</span> on a single core, likely due to this code in tmc_setup<span id="29139ce6-3462-40e5-83f4-ccd72c73f331" ginger_software_uiphraseguid="efdc6f79-60b8-4ecd-a8d2-6c1afb9cd7b0" class="GINGER_SOFTWARE_mark">.</span>F:
<br>
<br> CALL tmc_redistributing_cores<span id="3556fccf-cb40-4d46-9fee-11324dad5178" ginger_software_uiphraseguid="7f599d26-e5e4-46e1-8e93-939e925f7c04" class="GINGER_SOFTWARE_mark">(</span>tmc_<wbr>env%tmc_comp_set, para_env, &
<br> ana_on_the_fly=tmc_env%tmc_<wbr>comp_set%ana_on_the_fly,
<br>&
<br> success=success, error=error)
<br>
<br> IF<span id="af7245cd-1b68-46f2-96d4-08643a13e03c" ginger_software_uiphraseguid="7c2c3c68-9b05-4334-a460-4b9e3dcb4766" class="GINGER_SOFTWARE_mark">(</span>success) THEN
<br>[...]
<br> ELSE
<br> IF<span id="1eae6a89-1ac8-4a2f-ae53-be059dced483" ginger_software_uiphraseguid="54aec5e4-9396-49f6-80aa-f1f8835eaaa4" class="GINGER_SOFTWARE_mark">(</span>tmc_env%params%print_test_<wbr>output) THEN
<br> WRITE<span id="9b469ff9-4797-4339-a04b-b213dedbeb27" ginger_software_uiphraseguid="e108e19e-8955-4e24-8bd3-8ed2cfb8a734" class="GINGER_SOFTWARE_mark">(</span>output_unit<span id="f82749ed-d8e3-45f4-afff-0357e4cbd067" ginger_software_uiphraseguid="e108e19e-8955-4e24-8bd3-8ed2cfb8a734" class="GINGER_SOFTWARE_mark">,</span>*)"TMC|<wbr>NOTenoughProcessorsX= -999"
<br> WRITE<span id="15325be4-bb38-47cb-8641-60251202cfd2" ginger_software_uiphraseguid="23806899-8487-4ae5-a3fc-af72b6540acd" class="GINGER_SOFTWARE_mark">(</span>output_unit<span id="7146f339-a05e-4e96-a4a8-54f9af9ddf8b" ginger_software_uiphraseguid="23806899-8487-4ae5-a3fc-af72b6540acd" class="GINGER_SOFTWARE_mark">,</span>*)"TMC|<wbr>NOTcalculatedTotal energy: -999"
<br> END IF
<br> END IF
<br>
<br><span id="05f0f8b7-89b0-4084-81a9-45ab009c56fb" ginger_software_uiphraseguid="6a83afb8-7b7d-4937-af09-b296dbb07c24" class="GINGER_SOFTWARE_mark">which</span> apparently gets parsed by the <span id="f4de6bc8-5b1b-438a-a37d-ac35a1506473" ginger_software_uiphraseguid="6a83afb8-7b7d-4937-af09-b296dbb07c24" class="GINGER_SOFTWARE_mark">regtest</span> code for comparison. So I
<br><span id="060f1534-5182-4f10-84ff-0986e3f4c5fc" ginger_software_uiphraseguid="f0711a9b-cd20-4d76-8835-f3ee93a1bd9c" class="GINGER_SOFTWARE_mark">suggest</span> to check the output of the failing tests.
<br>
<br>Also, I believe setting a bogus energy on failure and exiting without
<br><span id="8e4903f8-5c5d-4fce-bc7b-3f47fcecde29" ginger_software_uiphraseguid="702ba616-403f-4e47-a900-998dd8ecf207" class="GINGER_SOFTWARE_mark">error</span> is not in line with the rest of CP2K code and should possibly be
<br><span id="7b781265-aa7a-4815-b0ef-44bfee6b1dc8" ginger_software_uiphraseguid="594e647c-3702-47d4-9570-b5f9ca042a3d" class="GINGER_SOFTWARE_mark">changed</span> (if that is indeed the culprit).
<br>
<br>
<br>Michael
<br></blockquote></div></div></div>