<html dir="ltr">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<style type="text/css" id="owaParaStyle"></style>
</head>
<body fpstyle="1" ocsi="0">
<div style="direction: ltr;font-family: Tahoma;color: #000000;font-size: 10pt;">
<div>Hi Matthias,</div>
<div><br>
</div>
<div>> There are failing tests for me as well, but the failures are in very special</div>
<div>features. Fixing those requires a lot of work. Most of the recent failures I</div>
<div>fixed were ICC bugs.</div>
<div><br>
</div>
<div>We see significant differences in tracking results with Vc-0.99 and Vc-0.7.2 using ICC-13.1.0. For example number of broken fast tracks (clones for ref set) is factor of 7(!) different. (see numbers below)</div>
<div>Meanwhile with VC_IMPL=Scalar both Vc-0.99 and Vc-0.7.2 give same result.</div>
<div><br>
</div>
<div>Do you have any suggestions how to proceed?</div>
<div><br>
</div>
<div>Regards,</div>
<div>Igor</div>
<div><br>
</div>
<div>Vc-0.99:</div>
<div> ---- Global Performance 10 events Statistic ---- </div>
<div>Track category : Eff / Clones | All Reco | All MC</div>
<div>Allset efficiency : 0.920 / 0.290 | 7119<span class="Apple-tab-span" style="white-space:pre">
</span> | 7740</div>
<div>Refset efficiency : 0.982 / 0.136 | 910<span class="Apple-tab-span" style="white-space:pre">
</span> | 927</div>
<div>Extra efficiency : 0.911 / 0.310 | 6209<span class="Apple-tab-span" style="white-space:pre">
</span> | 6813</div>
<div>Rest efficiency : 0.898 / 0.727 | 79<span class="Apple-tab-span" style="white-space:pre">
</span> | 88</div>
<div>Ghost probability : 0.129 | 917</div>
<div>All reco tracks/ev : 711.900</div>
<div>Reconstruction Time Real = 767.606 ms, CPU = 767.000 ms, parallelization speedup: 0.998</div>
<div><br>
</div>
<div>Vc-0.7.2:</div>
<div> ---- Global Performance 10 events Statistic ---- </div>
<div>Track category : Eff / Clones | All Reco | All MC</div>
<div>Allset efficiency : 0.944 / 0.113 | 7304<span class="Apple-tab-span" style="white-space:pre">
</span> | 7740</div>
<div>Refset efficiency : 0.980 / 0.020 | 908<span class="Apple-tab-span" style="white-space:pre">
</span> | 927</div>
<div>Extra efficiency : 0.939 / 0.126 | 6396<span class="Apple-tab-span" style="white-space:pre">
</span> | 6813</div>
<div>Rest efficiency : 0.909 / 0.773 | 80<span class="Apple-tab-span" style="white-space:pre">
</span> | 88</div>
<div>Ghost probability : 0.127 | 929</div>
<div>All reco tracks/ev : 730.400</div>
<div>Reconstruction Time Real = 146.283 ms, CPU = 145.000 ms, parallelization speedup: 0.981</div>
<div><br>
</div>
<div>________________________________________</div>
<div>From: vc-bounces@compeng.uni-frankfurt.de [vc-bounces@compeng.uni-frankfurt.de] on behalf of Matthias Kretz [kretz@compeng.uni-frankfurt.de]</div>
<div>Sent: Friday, August 16, 2013 10:33 AM</div>
<div>To: vc@compeng.uni-frankfurt.de</div>
<div>Subject: Re: Vc at MIC</div>
<div><br>
</div>
<div>On Tuesday 13 August 2013 20:19:52 Zyzak, Maksym wrote:</div>
<div>> I have compiled Vc-0.99.71 on the computer with MIC cards. And the test are</div>
<div>> failed. For the compilation I used icc 13.1.0.</div>
<div><br>
</div>
<div>I recommend to always update ICC to the latest version. I use ICC 13.1.2</div>
<div>20130514. There were some really strange miscompiles with the older compilers.</div>
<div><br>
</div>
<div>> The CPU is Intel(R) Xeon(R)</div>
<div>> CPU E5-2680. The log of test is given below. Can you please suggest what</div>
<div>> can be the problem? Should I use another version of Vc?</div>
<div><br>
</div>
<div>There are failing tests for me as well, but the failures are in very special</div>
<div>features. Fixing those requires a lot of work. Most of the recent failures I</div>
<div>fixed were ICC bugs.</div>
<div><br>
</div>
<div>--</div>
<div>Dipl.-Phys. Matthias Kretz</div>
<div><br>
</div>
<div>Web: http://compeng.uni-frankfurt.de/?mkretz</div>
<div><br>
</div>
<div>SIMD easy and portable: http://compeng.uni-frankfurt.de/?vc</div>
<div><br>
</div>
<div>_______________________________________________</div>
<div>Vc mailing list</div>
<div>Vc@compeng.uni-frankfurt.de</div>
<div>https://compeng.uni-frankfurt.de/mailman/listinfo/vc</div>
<div><br>
</div>
</div>
</body>
</html>