<?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 Re:  new idea about jpeg  encode and decode  in Intel® Integrated Performance Primitives</title>
    <link>https://community.intel.com/t5/Intel-Integrated-Performance/new-idea-about-jpeg-encode-and-decode/m-p/898057#M12381</link>
    <description>&lt;DIV style="margin:0px;"&gt;
&lt;DIV id="quote_reply" style="margin-top: 5px; width: 100%;"&gt;
&lt;DIV style="margin-left:2px;margin-right:2px;"&gt;Quoting - &lt;A href="https://community.intel.com/en-us/profile/336498"&gt;Vladimir Dudnik (Intel)&lt;/A&gt;&lt;/DIV&gt;
&lt;DIV style="background-color:#E5E5E5; padding:5px;border: 1px; border-style: inset;margin-left:2px;margin-right:2px;"&gt;&lt;EM&gt;I think with future Intel Larrabee architecture you will be able to implement huffman encoding natively on GPU. In this case you may put whole JPEG algorithm to GPU. Stay tuned.&lt;BR /&gt;&lt;BR /&gt;Regards,&lt;BR /&gt; Vladimir&lt;BR /&gt;&lt;/EM&gt;&lt;/DIV&gt;
&lt;/DIV&gt;
&lt;/DIV&gt;
&lt;BR /&gt;Thanks a lot, If It can be able to implement huffman encoding on GPU , the jpeg encode and decode time-consuming can be reduce.</description>
    <pubDate>Thu, 27 Aug 2009 11:29:38 GMT</pubDate>
    <dc:creator>qianqianzhutianfang</dc:creator>
    <dc:date>2009-08-27T11:29:38Z</dc:date>
    <item>
      <title>new idea about jpeg  encode and decode</title>
      <link>https://community.intel.com/t5/Intel-Integrated-Performance/new-idea-about-jpeg-encode-and-decode/m-p/898051#M12375</link>
      <description>To reduce the use of cpu, Can I use cpu and gpu together to achivejpeg encode and decode , can this ideafeasible? IS jpeg encode and decode time-consuming can be reduce??</description>
      <pubDate>Tue, 11 Aug 2009 07:46:51 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Integrated-Performance/new-idea-about-jpeg-encode-and-decode/m-p/898051#M12375</guid>
      <dc:creator>qianqianzhutianfang</dc:creator>
      <dc:date>2009-08-11T07:46:51Z</dc:date>
    </item>
    <item>
      <title>Re:  new idea about jpeg  encode and decode</title>
      <link>https://community.intel.com/t5/Intel-Integrated-Performance/new-idea-about-jpeg-encode-and-decode/m-p/898052#M12376</link>
      <description>&lt;DIV style="margin:0px;"&gt;&lt;/DIV&gt;
When you consider heterogeneous model (when part of your algorithm run on cpu and other part is running on gpu) you need to take into account the overhead of exchanging data between those two parts. In gpu case you have to send data through PCIe bus which may cost too much in overall application performance.&lt;BR /&gt;For such a simple algorithms like JPEG it may give no benefits while for some other, computationally extensive application it may provide benefits.&lt;BR /&gt;&lt;BR /&gt;Regards,&lt;BR /&gt; Vladimir&lt;BR /&gt;</description>
      <pubDate>Tue, 11 Aug 2009 16:22:50 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Integrated-Performance/new-idea-about-jpeg-encode-and-decode/m-p/898052#M12376</guid>
      <dc:creator>Vladimir_Dudnik</dc:creator>
      <dc:date>2009-08-11T16:22:50Z</dc:date>
    </item>
    <item>
      <title>Re:  new idea about jpeg  encode and decode</title>
      <link>https://community.intel.com/t5/Intel-Integrated-Performance/new-idea-about-jpeg-encode-and-decode/m-p/898053#M12377</link>
      <description>&lt;DIV style="margin:0px;"&gt;
&lt;DIV id="quote_reply" style="margin-top: 5px; width: 100%;"&gt;
&lt;DIV style="margin-left:2px;margin-right:2px;"&gt;Quoting - &lt;A href="https://community.intel.com/en-us/profile/336498"&gt;Vladimir Dudnik (Intel)&lt;/A&gt;&lt;/DIV&gt;
&lt;DIV style="background-color:#E5E5E5; padding:5px;border: 1px; border-style: inset;margin-left:2px;margin-right:2px;"&gt;&lt;EM&gt;When you consider heterogeneous model (when part of your algorithm run on cpu and other part is running on gpu) you need to take into account the overhead of exchanging data between those two parts. In gpu case you have to send data through PCIe bus which may cost too much in overall application performance.&lt;BR /&gt;For such a simple algorithms like JPEG it may give no benefits while for some other, computationally extensive application it may provide benefits.&lt;BR /&gt;&lt;BR /&gt;Regards,&lt;BR /&gt; Vladimir&lt;BR /&gt;&lt;/EM&gt;&lt;/DIV&gt;
&lt;/DIV&gt;
&lt;/DIV&gt;
&lt;BR /&gt;Do you mean this method can'tsave time for JPEG Encode or Decode? I have made a sample to do this, but it cost too much time to send data between GPU and CPU, I don't known whetheror not to continue research? or give up this idea?</description>
      <pubDate>Thu, 13 Aug 2009 12:58:55 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Integrated-Performance/new-idea-about-jpeg-encode-and-decode/m-p/898053#M12377</guid>
      <dc:creator>qianqianzhutianfang</dc:creator>
      <dc:date>2009-08-13T12:58:55Z</dc:date>
    </item>
    <item>
      <title>Re:  new idea about jpeg  encode and decode</title>
      <link>https://community.intel.com/t5/Intel-Integrated-Performance/new-idea-about-jpeg-encode-and-decode/m-p/898054#M12378</link>
      <description>&lt;DIV style="margin:0px;"&gt;&lt;/DIV&gt;
Well you to decide to continue with this or not. I can guess the only way which probably might make sense with existing GPU (which can't implement whole JPEG algorithm) is something like doing huffman decoding on cpu and send decoded coefficients to GPU where you may try to do IDCT, color conversion and display.&lt;BR /&gt;&lt;BR /&gt;Vladimir&lt;BR /&gt;</description>
      <pubDate>Thu, 13 Aug 2009 16:56:37 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Integrated-Performance/new-idea-about-jpeg-encode-and-decode/m-p/898054#M12378</guid>
      <dc:creator>Vladimir_Dudnik</dc:creator>
      <dc:date>2009-08-13T16:56:37Z</dc:date>
    </item>
    <item>
      <title>Re:  new idea about jpeg  encode and decode</title>
      <link>https://community.intel.com/t5/Intel-Integrated-Performance/new-idea-about-jpeg-encode-and-decode/m-p/898055#M12379</link>
      <description>&lt;DIV style="margin:0px;"&gt;
&lt;DIV id="quote_reply" style="margin-top: 5px; width: 100%;"&gt;
&lt;DIV style="margin-left:2px;margin-right:2px;"&gt;Quoting - &lt;A href="https://community.intel.com/en-us/profile/336498"&gt;Vladimir Dudnik (Intel)&lt;/A&gt;&lt;/DIV&gt;
&lt;DIV style="background-color:#E5E5E5; padding:5px;border: 1px; border-style: inset;margin-left:2px;margin-right:2px;"&gt;&lt;EM&gt;Well you to decide to continue with this or not. I can guess the only way which probably might make sense with existing GPU (which can't implement whole JPEG algorithm) is something like doing huffman decoding on cpu and send decoded coefficients to GPU where you may try to do IDCT, color conversion and display.&lt;BR /&gt;&lt;BR /&gt;Vladimir&lt;BR /&gt;&lt;/EM&gt;&lt;/DIV&gt;
&lt;/DIV&gt;
&lt;/DIV&gt;
&lt;BR /&gt;Yes,you are right. but this is only parts of my work,I also need encode JPEG,include doing color conversion,DCT,quantitative onGPU and send encoded data to CPU to do huffman encode. Which need exchange data two times, it cost too much time.&lt;BR /&gt;&lt;BR /&gt;aslo,I can't make GPU and CPUdo work Synchronizationly,so .... can'tsave time .</description>
      <pubDate>Fri, 14 Aug 2009 11:52:37 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Integrated-Performance/new-idea-about-jpeg-encode-and-decode/m-p/898055#M12379</guid>
      <dc:creator>qianqianzhutianfang</dc:creator>
      <dc:date>2009-08-14T11:52:37Z</dc:date>
    </item>
    <item>
      <title>Re:  new idea about jpeg  encode and decode</title>
      <link>https://community.intel.com/t5/Intel-Integrated-Performance/new-idea-about-jpeg-encode-and-decode/m-p/898056#M12380</link>
      <description>I think with future Intel Larrabee architecture you will be able to implement huffman encoding natively on GPU. In this case you may put whole JPEG algorithm to GPU. Stay tuned.&lt;BR /&gt;&lt;BR /&gt;Regards,&lt;BR /&gt; Vladimir&lt;BR /&gt;</description>
      <pubDate>Fri, 14 Aug 2009 15:25:13 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Integrated-Performance/new-idea-about-jpeg-encode-and-decode/m-p/898056#M12380</guid>
      <dc:creator>Vladimir_Dudnik</dc:creator>
      <dc:date>2009-08-14T15:25:13Z</dc:date>
    </item>
    <item>
      <title>Re:  new idea about jpeg  encode and decode</title>
      <link>https://community.intel.com/t5/Intel-Integrated-Performance/new-idea-about-jpeg-encode-and-decode/m-p/898057#M12381</link>
      <description>&lt;DIV style="margin:0px;"&gt;
&lt;DIV id="quote_reply" style="margin-top: 5px; width: 100%;"&gt;
&lt;DIV style="margin-left:2px;margin-right:2px;"&gt;Quoting - &lt;A href="https://community.intel.com/en-us/profile/336498"&gt;Vladimir Dudnik (Intel)&lt;/A&gt;&lt;/DIV&gt;
&lt;DIV style="background-color:#E5E5E5; padding:5px;border: 1px; border-style: inset;margin-left:2px;margin-right:2px;"&gt;&lt;EM&gt;I think with future Intel Larrabee architecture you will be able to implement huffman encoding natively on GPU. In this case you may put whole JPEG algorithm to GPU. Stay tuned.&lt;BR /&gt;&lt;BR /&gt;Regards,&lt;BR /&gt; Vladimir&lt;BR /&gt;&lt;/EM&gt;&lt;/DIV&gt;
&lt;/DIV&gt;
&lt;/DIV&gt;
&lt;BR /&gt;Thanks a lot, If It can be able to implement huffman encoding on GPU , the jpeg encode and decode time-consuming can be reduce.</description>
      <pubDate>Thu, 27 Aug 2009 11:29:38 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Integrated-Performance/new-idea-about-jpeg-encode-and-decode/m-p/898057#M12381</guid>
      <dc:creator>qianqianzhutianfang</dc:creator>
      <dc:date>2009-08-27T11:29:38Z</dc:date>
    </item>
  </channel>
</rss>

