<?xml version="1.0" encoding="UTF-8"?>
<rss xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:taxo="http://purl.org/rss/1.0/modules/taxonomy/" version="2.0">
  <channel>
    <title>topic Quote:Simon B. wrote: in Intel® Embree Ray Tracing Kernels</title>
    <link>https://community.intel.com/t5/Intel-Embree-Ray-Tracing-Kernels/Assert-during-structure-build/m-p/993917#M309</link>
    <description>&lt;BLOCKQUOTE&gt;Simon B. wrote:&lt;BR /&gt;&lt;P&gt;Actually could someone from the embree team comment on whether this change looks reasonable, or does an atomic_set with empty blocks at the start indicate some deeper issue?  I am happy to provide the input data to the builder.  My configuration is an AVX build using defaults.  Thanks!&lt;/P&gt;&lt;/BLOCKQUOTE&gt;
I got this fix directly from Swen: new atomic_set.h &lt;A href="http://pastebin.com/WeAcdS1C" target="_blank"&gt;http://pastebin.com/WeAcdS1C&lt;/A&gt;</description>
    <pubDate>Sat, 10 Nov 2012 11:05:40 GMT</pubDate>
    <dc:creator>Keymaster</dc:creator>
    <dc:date>2012-11-10T11:05:40Z</dc:date>
    <item>
      <title>Assert during structure build</title>
      <link>https://community.intel.com/t5/Intel-Embree-Ray-Tracing-Kernels/Assert-during-structure-build/m-p/993912#M304</link>
      <description>Hi,&lt;BR /&gt;I am currently testing the new embree (1.1 beta) in my renderer. I am really impressed by the build speeds, but I keep getting one assert fail in one of my test scenes. &lt;BR /&gt;&lt;BR /&gt;I reproduced it in your embree desktop application, it always happens in one fairly big scene (7M tris) with default settings (BVH4, triangle4). &lt;BR /&gt;&lt;BR /&gt;It is line 81 in bvh4_builder.cpp: assert(!iter);&lt;BR /&gt;&lt;BR /&gt;I've obtained permission from the author to give you the scene for debugging purposes, if you keep it confidential. If you want it, give me an email address where should it be delivered.&lt;BR /&gt;</description>
      <pubDate>Sat, 25 Aug 2012 15:36:54 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Embree-Ray-Tracing-Kernels/Assert-during-structure-build/m-p/993912#M304</guid>
      <dc:creator>Keymaster</dc:creator>
      <dc:date>2012-08-25T15:36:54Z</dc:date>
    </item>
    <item>
      <title>Assert during structure build</title>
      <link>https://community.intel.com/t5/Intel-Embree-Ray-Tracing-Kernels/Assert-during-structure-build/m-p/993913#M305</link>
      <description>Hi, I will have a look at this. Could you please send the scene (or better a link to it) to the email address in my profile. The scene will of course be kept confidential.</description>
      <pubDate>Tue, 28 Aug 2012 05:38:35 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Embree-Ray-Tracing-Kernels/Assert-during-structure-build/m-p/993913#M305</guid>
      <dc:creator>SvenW_Intel</dc:creator>
      <dc:date>2012-08-28T05:38:35Z</dc:date>
    </item>
    <item>
      <title>Assert during structure build</title>
      <link>https://community.intel.com/t5/Intel-Embree-Ray-Tracing-Kernels/Assert-during-structure-build/m-p/993914#M306</link>
      <description>Hi,&lt;BR /&gt;I unfortunately cannot find the email in your profile. Can you please post it here or contact me at keymaster( at )keymaster.cz</description>
      <pubDate>Wed, 29 Aug 2012 13:36:13 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Embree-Ray-Tracing-Kernels/Assert-during-structure-build/m-p/993914#M306</guid>
      <dc:creator>Keymaster</dc:creator>
      <dc:date>2012-08-29T13:36:13Z</dc:date>
    </item>
    <item>
      <title>I also got this assert, but</title>
      <link>https://community.intel.com/t5/Intel-Embree-Ray-Tracing-Kernels/Assert-during-structure-build/m-p/993915#M307</link>
      <description>I also got this assert, but with very simple geometry.  It seems to be caused by empty blocks at the start of an atomic set, the following change to common/atomic_set.h avoids the assert:

       typedef typename T::T Type;
     public:
       __forceinline block_iterator_unsafe (atomic_set&amp;amp; other) 
-        : root(other.root), pos(0) {}
+        : root(other.root), pos(0)
+      {
+        while (root &amp;amp;&amp;amp; 0 == root-&amp;gt;size()) {
+          root = root-&amp;gt;next;
+        }
+      }
       
       __forceinline void operator++(int)
       {

I'm not sure if empty blocks are expected here, but thought I'd post my workaround in case it helps!</description>
      <pubDate>Wed, 07 Nov 2012 22:57:48 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Embree-Ray-Tracing-Kernels/Assert-during-structure-build/m-p/993915#M307</guid>
      <dc:creator>Simon_B_4</dc:creator>
      <dc:date>2012-11-07T22:57:48Z</dc:date>
    </item>
    <item>
      <title>Actually could someone from</title>
      <link>https://community.intel.com/t5/Intel-Embree-Ray-Tracing-Kernels/Assert-during-structure-build/m-p/993916#M308</link>
      <description>Actually could someone from the embree team comment on whether this change looks reasonable, or does an atomic_set with empty blocks at the start indicate some deeper issue?  I am happy to provide the input data to the builder.  My configuration is an AVX build using defaults.  Thanks!</description>
      <pubDate>Sat, 10 Nov 2012 10:53:58 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Embree-Ray-Tracing-Kernels/Assert-during-structure-build/m-p/993916#M308</guid>
      <dc:creator>Simon_B_4</dc:creator>
      <dc:date>2012-11-10T10:53:58Z</dc:date>
    </item>
    <item>
      <title>Quote:Simon B. wrote:</title>
      <link>https://community.intel.com/t5/Intel-Embree-Ray-Tracing-Kernels/Assert-during-structure-build/m-p/993917#M309</link>
      <description>&lt;BLOCKQUOTE&gt;Simon B. wrote:&lt;BR /&gt;&lt;P&gt;Actually could someone from the embree team comment on whether this change looks reasonable, or does an atomic_set with empty blocks at the start indicate some deeper issue?  I am happy to provide the input data to the builder.  My configuration is an AVX build using defaults.  Thanks!&lt;/P&gt;&lt;/BLOCKQUOTE&gt;
I got this fix directly from Swen: new atomic_set.h &lt;A href="http://pastebin.com/WeAcdS1C" target="_blank"&gt;http://pastebin.com/WeAcdS1C&lt;/A&gt;</description>
      <pubDate>Sat, 10 Nov 2012 11:05:40 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Embree-Ray-Tracing-Kernels/Assert-during-structure-build/m-p/993917#M309</guid>
      <dc:creator>Keymaster</dc:creator>
      <dc:date>2012-11-10T11:05:40Z</dc:date>
    </item>
    <item>
      <title>Quote:Keymaster wrote:I got</title>
      <link>https://community.intel.com/t5/Intel-Embree-Ray-Tracing-Kernels/Assert-during-structure-build/m-p/993918#M310</link>
      <description>&lt;BLOCKQUOTE&gt;Keymaster wrote:&lt;BR /&gt;I got this fix directly from Swen: new atomic_set.h &lt;A href="http://pastebin.com/WeAcdS1C"&gt;http://pastebin.com/WeAcdS1C&lt;/A&gt;&lt;/BLOCKQUOTE&gt;

Thanks for that, looks equivalent to my change, I'll assume this is a valid fix then.</description>
      <pubDate>Sat, 10 Nov 2012 11:15:08 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Embree-Ray-Tracing-Kernels/Assert-during-structure-build/m-p/993918#M310</guid>
      <dc:creator>Simon_B_4</dc:creator>
      <dc:date>2012-11-10T11:15:08Z</dc:date>
    </item>
  </channel>
</rss>

