<html>#2516: Different simulation result with different number of cores setting
<table style='border-spacing: 1ex 0pt; '>
<tr><td style='text-align:right'> Reporter:</td><td></td></tr>
<tr><td style='text-align:right'> Status:</td><td>open</td></tr>
<tr><td style='text-align:right'>Milestone:</td><td></td></tr>
<tr><td style='text-align:right'> Version:</td><td>ET_2019_10</td></tr>
<tr><td style='text-align:right'> Type:</td><td>bug</td></tr>
<tr><td style='text-align:right'> Priority:</td><td>major</td></tr>
<tr><td style='text-align:right'>Component:</td><td></td></tr>
</table>
<p>Comment (by Ken Hui):</p>
<p>For all simulations I did for this test, there is no checkpoint/recovery cycle; and the time-series h+ hx data in mp_psi4 are different in (almost) every line. I mentioned that there are two groups of output:</p>
<ul>
<li>(i) number of cores = 8 / 12 / 32 / 36</li>
<li>(ii) number of cores = 16 / 17 / 24 / 48</li>
</ul>
<p>what I meant is that, among all simulations in the same group, the mp_psi4 outputs are only different after ~10 decimal places even after a long simulation time; if I compare group (i) and group (ii), for example the mp_psi4 results from 16-core or 32-core, at first they differ at roughly 1 in 10^10, but later on the discrepancy starts growing (see figure, viewed by vim -d).</p>
<p><img alt="" src="https://bitbucket.org/repo/8zBLME5/images/394861091-Screenshot%202021-03-31%20at%2012.04.54%20AM.png" />
For the OpenMP part, if I understand correctly, you mean even if I issue </p>
<div class="codehilite"><pre><span></span>export OMP_NUM_THREADS=1
</pre></div>
<p>before the mpirun command every time, there might still problem led by its operation?</p>
<p>--<br/>
Ticket URL: <a href='https://bitbucket.org/einsteintoolkit/tickets/issues/2516/different-simulation-result-with-different'>https://bitbucket.org/einsteintoolkit/tickets/issues/2516/different-simulation-result-with-different</a></p>
</html>