Thanks for your response Justin. We certainly appreciate hyperscan for everything that it's done for the community already!<div>Kind regards,</div><div>CB<span></span><br><br>On Friday, November 11, 2016, Viiret, Justin <<a href="mailto:justin.viiret@intel.com">justin.viiret@intel.com</a>> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div lang="EN-US" link="#0563C1" vlink="#954F72">
<div>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri",sans-serif;color:#1f497d">Hi Chris,<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri",sans-serif;color:#1f497d"><u></u> <u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri",sans-serif;color:#1f497d">We’re not actively investigating OpenCL based approaches for pattern matching. We have concerns about latency (for inline applications), total performance and
the potential for “soft errors”. The amount of work it would require for us to support two patches to Suricata, one out of scope, is simply too much.<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri",sans-serif;color:#1f497d"><u></u> <u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri",sans-serif;color:#1f497d">Currently we push all our pattern-matching effort into Hyperscan itself: that way performance improvements behind the scenes in Hyperscan (algorithmic improvements,
support for new hardware features, etc) can be taken advantage of by Suricata without needing further changes to the Suricata code.<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri",sans-serif;color:#1f497d"><u></u> <u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri",sans-serif;color:#1f497d">Best regards,<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri",sans-serif;color:#1f497d"> Justin<u></u><u></u></span></p>
<p class="MsoNormal"><a name="m_8245809824689403383__MailEndCompose"><span style="font-size:11.0pt;font-family:"Calibri",sans-serif;color:#1f497d"><u></u> <u></u></span></a></p>
<p class="MsoNormal"><b><span style="font-size:11.0pt;font-family:"Calibri",sans-serif">From:</span></b><span style="font-size:11.0pt;font-family:"Calibri",sans-serif"> Oisf-users [mailto:<a href="javascript:_e(%7B%7D,'cvml','oisf-users-bounces@lists.openinfosecfoundation.org');" target="_blank">oisf-users-bounces@<wbr>lists.openinfosecfoundation.<wbr>org</a>]
<b>On Behalf Of </b>Chris Boley<br>
<b>Sent:</b> Tuesday, November 8, 2016 11:20 AM<br>
<b>To:</b> <a href="javascript:_e(%7B%7D,'cvml','oisf-users@lists.openinfosecfoundation.org');" target="_blank">oisf-users@lists.<wbr>openinfosecfoundation.org</a><br>
<b>Subject:</b> [Oisf-users] OpenCL<u></u><u></u></span></p>
<p class="MsoNormal"><u></u> <u></u></p>
<div>
<div>
<p class="MsoNormal" style="margin-bottom:12.0pt">I saw in a previous post where CUDA has gone by the wayside in the area of continuing development support. I know the Intel professionals that helped incorporate hyperscan for Suricata are monitoring this forum.
<br>
<br>
I pose this question:<br>
<br>
Given the great work that Intel has already done with hyperscan, why not work toward development incorporating even more horsepower via the OpenCL API? Given that many/most suricata appliances really are just CLI based server installations, the graphics processor
mostly is not used. I think it makes sense for Intel to encourage Suricata development for intel based graphics processors/GPU's. Maybe this is already happening and I'm simply not aware. Is this even realistic? I figured it couldn't hurt to ask.<u></u><u></u></p>
</div>
<p class="MsoNormal">CB<u></u><u></u></p>
</div>
</div>
</div>
</blockquote></div>