<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<p>A teljes thread itt érhető el:
<a class="moz-txt-link-freetext" href="http://mails.dpdk.org/archives/users/2019-January/003877.html">http://mails.dpdk.org/archives/users/2019-January/003877.html</a></p>
<p>Üdv.</p>
<p>S<br>
</p>
<div class="moz-cite-prefix">2019.01.18. 14:18 keltezéssel, Sándor
Laki írta:<br>
</div>
<blockquote type="cite"
cite="mid:355909c9-c960-1ed6-fcac-7a12e06b0c0e@elte.hu">
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
<p>Ez relevánsnak tűnik a crypto szempontjából és érdemes
figyelembe venni a szívások elkerülése végett.</p>
<p>Üdv.<br>
S<br>
</p>
<div class="moz-forward-container"><br>
<br>
-------- Továbbított üzenet --------
<table class="moz-email-headers-table" cellspacing="0"
cellpadding="0" border="0">
<tbody>
<tr>
<th valign="BASELINE" nowrap="nowrap" align="RIGHT">Tárgy:
</th>
<td>Re: [dpdk-users] Run-to-completion or Pipe-line for
QAT PMD in DPDK</td>
</tr>
<tr>
<th valign="BASELINE" nowrap="nowrap" align="RIGHT">Dátum:
</th>
<td>Fri, 18 Jan 2019 13:13:43 +0000</td>
</tr>
<tr>
<th valign="BASELINE" nowrap="nowrap" align="RIGHT">Feladó:
</th>
<td>Trahe, Fiona <a class="moz-txt-link-rfc2396E"
href="mailto:fiona.trahe@intel.com"
moz-do-not-send="true"><fiona.trahe@intel.com></a></td>
</tr>
<tr>
<th valign="BASELINE" nowrap="nowrap" align="RIGHT">Címzett:
</th>
<td>Changchun Zhang <a class="moz-txt-link-rfc2396E"
href="mailto:changchun.zhang@oracle.com"
moz-do-not-send="true"><changchun.zhang@oracle.com></a>,
<a class="moz-txt-link-abbreviated"
href="mailto:users@dpdk.org" moz-do-not-send="true">users@dpdk.org</a>
<a class="moz-txt-link-rfc2396E"
href="mailto:users@dpdk.org" moz-do-not-send="true"><users@dpdk.org></a></td>
</tr>
<tr>
<th valign="BASELINE" nowrap="nowrap" align="RIGHT">CC: </th>
<td>Trahe, Fiona <a class="moz-txt-link-rfc2396E"
href="mailto:fiona.trahe@intel.com"
moz-do-not-send="true"><fiona.trahe@intel.com></a></td>
</tr>
</tbody>
</table>
<br>
<br>
Hi Alex,<br>
<br>
<blockquote type="cite">-----Original Message-----<br>
From: users [<a class="moz-txt-link-freetext"
href="mailto:users-bounces@dpdk.org" moz-do-not-send="true">mailto:users-bounces@dpdk.org</a>]
On Behalf Of Changchun Zhang<br>
Sent: Thursday, January 17, 2019 11:01 PM<br>
To: <a class="moz-txt-link-abbreviated"
href="mailto:users@dpdk.org" moz-do-not-send="true">users@dpdk.org</a><br>
Subject: [dpdk-users] Run-to-completion or Pipe-line for QAT
PMD in DPDK<br>
<br>
Hi,<br>
<br>
<br>
<br>
I have user question on using the QAT device in the DPDK.<br>
<br>
In the real design, after calling enqueuer_burst() on the
specified queue pair at one of the lcore,<br>
usually which one is usually done?<br>
<br>
1. should we do run-to-completion to call dequeuer_burst()
waiting for the device finishing the<br>
crypto operation,<br>
<br>
2. or should we do pipe-line, in which we return right after
enqueuer_burst() and release the CPU.<br>
And call dequeuer_burst() on other thread function?<br>
<br>
Option 1 is more like synchronous and can be seen on all the
DPDK crypto examples, while option 2 is<br>
asynchronous which I have never seen in any reference design
if I missed anything.<br>
</blockquote>
[Fiona] Option 2 is not possible with QAT - the dequeue must be
called in the same thread as the enqueue. This is<br>
optimised without atomics for best performance - if this is a
problem let us know. However best performance is not quite using
option 1 and not a synchronous blocking method. If you enqueue
and then go straight to dequeue, you're not getting the best
advantage from the<br>
cycles freed up by offloading. i.e. best to enqueue a burst,
then go do some other work, like maybe collecting more requests
for next enqueue or other processing, then dequeue. Take and
process whatever ops are dequeued - this<br>
will not necessarily match up with the number you've enqueued -
depends on how quickly you call the dequeue.<br>
Don't wait until all the enqueued ops are dequeued before
enqueuing the next batch.<br>
SO it's asynchronous. But in the same thread.<br>
You'll get best throughput when you keep the input filled up so
the device has operations to work on and<br>
regularly dequeue a burst. Dequeuing too often will waste cycles
in the overhead calling the API, dequeuing too<br>
slowly will cause the device to back up. Ideally tune for your
application to find the sweet spot in<br>
between these 2 extremes. </div>
<div id="DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2"><br>
<table style="border-top: 1px solid #D3D4DE;">
<tbody>
<tr>
<td style="width: 55px; padding-top: 18px;"><a
href="https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient"
target="_blank" moz-do-not-send="true"><img
src="https://ipmcdn.avast.com/images/icons/icon-envelope-tick-round-orange-animated-no-repeat-v1.gif"
alt="" style="width: 46px; height: 29px;"
moz-do-not-send="true" width="46" height="29"></a></td>
<td style="width: 470px; padding-top: 17px; color:
#41424e; font-size: 13px; font-family: Arial, Helvetica,
sans-serif; line-height: 18px;">Mentes a vírusoktól. <a
href="https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient"
target="_blank" style="color: #4453ea;"
moz-do-not-send="true">www.avast.com</a> </td>
</tr>
</tbody>
</table>
<a href="#DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2" width="1"
height="1" moz-do-not-send="true"> </a></div>
<br>
<fieldset class="mimeAttachmentHeader"></fieldset>
<pre class="moz-quote-pre" wrap="">_______________________________________________
P4 mailing list
<a class="moz-txt-link-abbreviated" href="mailto:P4@plc.inf.elte.hu">P4@plc.inf.elte.hu</a>
<a class="moz-txt-link-freetext" href="https://plc.inf.elte.hu/mailman/listinfo/p4">https://plc.inf.elte.hu/mailman/listinfo/p4</a>
</pre>
</blockquote>
<pre class="moz-signature" cols="72">--
Sándor Laki, PhD
Assistant professor
Department of Information Systems
Eötvös Loránd University
Pázmány Péter stny. 1/C
H-1117, Budapest, Hungary
Room 2.506
Web: <a class="moz-txt-link-freetext" href="http://lakis.web.elte.hu">http://lakis.web.elte.hu</a>
Phone: +36 1 372 2869 / 8477
Cell: +36 70 374 2646</pre>
</body>
</html>