<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
</head>
<body>
<br>
<blockquote type="cite"
cite="mid:CAOWoSSPiTn-zteukHXYLQG7fNn+7D2PmGYu+AShE5etz2KM13Q@mail.gmail.com">
<div dir="auto">Speaking of ppcg, is there any published (or
otherwise public) benchmark of ppcg vs Davidson and/or cg? For
which cases can ppcg be expected to be faster?</div>
</blockquote>
<p>I don't know, I'm not an author of that part of code. I've just
tested it out of curiosity in this particular case (i.e. very slow
diagonalization)<br>
</p>
<p><br>
</p>
<p><br>
</p>
<blockquote type="cite"
cite="mid:CAOWoSSPiTn-zteukHXYLQG7fNn+7D2PmGYu+AShE5etz2KM13Q@mail.gmail.com">
<div dir="auto">Best regards,</div>
<div dir="auto">Michal Krompiec, Merck KGaA </div>
<div><br>
<div class="gmail_quote">
<div dir="ltr" class="gmail_attr">On Mon, Dec 14, 2020 at 2:12
PM Lorenzo Paulatto <<a href="mailto:paulatz@gmail.com"
moz-do-not-send="true">paulatz@gmail.com</a>> wrote:<br>
</div>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">p.p.s I
mean diagonalization='ppcg'<br>
<br>
On 2020-12-14 15:10, Lorenzo Paulatto wrote:<br>
> p.s. If you can use a newer version of QE that does
calculation="ppcg" <br>
> I found it to be much (i.e. 6x) faster in this case<br>
><br>
> cheers<br>
><br>
> On 2020-12-14 14:50, Lorenzo Paulatto wrote:<br>
>> Hello,<br>
>><br>
>> I've had a look at the output, and a part for the
cutoff which <br>
>> appears a bit too high (you are probably safe with
50/400Ry of <br>
>> ecutwfc/ecutrho) I only see to small problems:<br>
>><br>
>> 1. the scf calculation is using 6 pools with 10
k-points, which means <br>
>> that 4 pools have twice as much work to do as the
others. In the <br>
>> ideal case, the number of pools should be a divisor
of the number of <br>
>> k-points (i.e. 2, 5 or 10 in your case). Also, it
is recommended that <br>
>> the number of CPUs in a pool are a divisor of the
number of CPUs on <br>
>> each computing node, to avoid too much inter-node
communication. In <br>
>> your case, the best choice with 72 CPUs (on two
nodes?) could be 2 <br>
>> pools. You may gain a bit of time, but this is not
going to change a <br>
>> lot. You should consider using more CPUs if you
have the budget. For <br>
>> example, 10 pools of 12 or 18 CPUs each.<br>
>><br>
>> 2. The bands calculation runs on 12 CPUs and has a
single k-point, <br>
>> while each pool of the SCF one has up to 2
k-points. We would expect <br>
>> that the bands calculation take about half as an
scf step, i.e. about <br>
>> 50 seconds. However, the bands calculation has some
trouble <br>
>> diagonalizing the Hamiltonian, you see it writes:<br>
>><br>
>> ethr = 2.76E-12, avg # of iterations =120.0<br>
>><br>
>> while typically the very last scf diagonalization
is<br>
>><br>
>> ethr = 2.98E-12, avg # of iterations = 3.3<br>
>><br>
>> This is because, the scf calculation can start with
a very good guess <br>
>> good the wavefunction, while the bands calculation
does not. It is <br>
>> still faster than doing the entire scf procedure,
but just by a <br>
>> factor ~2.3<br>
>><br>
>> Fortunately, you do not usually need the
eigenvalues to a precision <br>
>> of 10e-12. You can set the threshold by hand using
the keyword <br>
>> diago_thr_init, I guess 1.d-6 should be tight
enough. However, double <br>
>> check what you get in output, because I am
half-suspecting that it <br>
>> may be over-written by the value in the restart
file<br>
>><br>
>> cheers<br>
>><br>
_______________________________________________<br>
Quantum ESPRESSO is supported by MaX (<a
href="http://www.max-centre.eu" rel="noreferrer"
target="_blank" moz-do-not-send="true">www.max-centre.eu</a>)<br>
users mailing list <a
href="mailto:users@lists.quantum-espresso.org"
target="_blank" moz-do-not-send="true">users@lists.quantum-espresso.org</a><br>
<a
href="https://lists.quantum-espresso.org/mailman/listinfo/users"
rel="noreferrer" target="_blank" moz-do-not-send="true">https://lists.quantum-espresso.org/mailman/listinfo/users</a></blockquote>
</div>
</div>
<br>
<fieldset class="mimeAttachmentHeader"></fieldset>
<pre class="moz-quote-pre" wrap="">_______________________________________________
Quantum ESPRESSO is supported by MaX (<a class="moz-txt-link-abbreviated" href="http://www.max-centre.eu">www.max-centre.eu</a>)
users mailing list <a class="moz-txt-link-abbreviated" href="mailto:users@lists.quantum-espresso.org">users@lists.quantum-espresso.org</a>
<a class="moz-txt-link-freetext" href="https://lists.quantum-espresso.org/mailman/listinfo/users">https://lists.quantum-espresso.org/mailman/listinfo/users</a></pre>
</blockquote>
</body>
</html>