<?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 RAPL interface for Sandy Bridge MSR update rate in Software Tuning, Performance Optimization &amp; Platform Monitoring</title>
    <link>https://community.intel.com/t5/Software-Tuning-Performance/RAPL-interface-for-Sandy-Bridge-MSR-update-rate/m-p/810326#M850</link>
    <description>Hi All,&lt;DIV&gt;&lt;/DIV&gt;&lt;DIV&gt;The ia 64 32 system developer mentions the following:&lt;/DIV&gt;&lt;DIV&gt;&lt;/DIV&gt;&lt;DIV&gt;&lt;DIV id="_mcePaste"&gt;"MSR_PKG_ENERGY_STATUS is a read-only MSR. It reports the actual energy use forthe package domain. This MSR is updated every ~1msec."&lt;/DIV&gt;&lt;/DIV&gt;&lt;DIV&gt;Is there any way for one to increase the rate at which MSRs are updated? I'm interested in checking the energy consumed by core sections which last as short as a microsecond. In other words can I increase the update rate by a 100-1000 fold without causing a high overhead?&lt;/DIV&gt;&lt;DIV&gt;&lt;/DIV&gt;&lt;DIV&gt;Many Thanks&lt;/DIV&gt;&lt;DIV&gt;&lt;/DIV&gt;&lt;DIV&gt;Akshay&lt;/DIV&gt;</description>
    <pubDate>Sat, 02 Jun 2012 10:38:05 GMT</pubDate>
    <dc:creator>Akshay_Venkatesh</dc:creator>
    <dc:date>2012-06-02T10:38:05Z</dc:date>
    <item>
      <title>RAPL interface for Sandy Bridge MSR update rate</title>
      <link>https://community.intel.com/t5/Software-Tuning-Performance/RAPL-interface-for-Sandy-Bridge-MSR-update-rate/m-p/810326#M850</link>
      <description>Hi All,&lt;DIV&gt;&lt;/DIV&gt;&lt;DIV&gt;The ia 64 32 system developer mentions the following:&lt;/DIV&gt;&lt;DIV&gt;&lt;/DIV&gt;&lt;DIV&gt;&lt;DIV id="_mcePaste"&gt;"MSR_PKG_ENERGY_STATUS is a read-only MSR. It reports the actual energy use forthe package domain. This MSR is updated every ~1msec."&lt;/DIV&gt;&lt;/DIV&gt;&lt;DIV&gt;Is there any way for one to increase the rate at which MSRs are updated? I'm interested in checking the energy consumed by core sections which last as short as a microsecond. In other words can I increase the update rate by a 100-1000 fold without causing a high overhead?&lt;/DIV&gt;&lt;DIV&gt;&lt;/DIV&gt;&lt;DIV&gt;Many Thanks&lt;/DIV&gt;&lt;DIV&gt;&lt;/DIV&gt;&lt;DIV&gt;Akshay&lt;/DIV&gt;</description>
      <pubDate>Sat, 02 Jun 2012 10:38:05 GMT</pubDate>
      <guid>https://community.intel.com/t5/Software-Tuning-Performance/RAPL-interface-for-Sandy-Bridge-MSR-update-rate/m-p/810326#M850</guid>
      <dc:creator>Akshay_Venkatesh</dc:creator>
      <dc:date>2012-06-02T10:38:05Z</dc:date>
    </item>
    <item>
      <title>RAPL interface for Sandy Bridge MSR update rate</title>
      <link>https://community.intel.com/t5/Software-Tuning-Performance/RAPL-interface-for-Sandy-Bridge-MSR-update-rate/m-p/810327#M851</link>
      <description>Hi Akshay,&lt;BR /&gt;&lt;BR /&gt;The count reported from this MSR is for the entire processor -- including all the cores. It is not directly possible to get per core readings for energy. &lt;BR /&gt;&lt;BR /&gt;The access latency for reading this register is also fairly sizable, so reading it at a frequency higher than 1000 times per sec may have significant overhead.&lt;BR /&gt;&lt;BR /&gt;If you need to isolate the consumption of a specific program or thread, you will need to use a script to allow it to run continously and read the value over a reasonable duration, and then normalize appropriately.&lt;BR /&gt;&lt;BR /&gt;Hope this helps,&lt;BR /&gt;Hussam</description>
      <pubDate>Thu, 12 Jul 2012 16:41:33 GMT</pubDate>
      <guid>https://community.intel.com/t5/Software-Tuning-Performance/RAPL-interface-for-Sandy-Bridge-MSR-update-rate/m-p/810327#M851</guid>
      <dc:creator>Hussam_Mousa__Intel_</dc:creator>
      <dc:date>2012-07-12T16:41:33Z</dc:date>
    </item>
  </channel>
</rss>

