<?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 TLB miss in Software Tuning, Performance Optimization &amp; Platform Monitoring</title>
    <link>https://community.intel.com/t5/Software-Tuning-Performance/TLB-miss/m-p/1126715#M6323</link>
    <description>&lt;P&gt;Hello.&lt;/P&gt;

&lt;P&gt;I am working on IvyTown processor, and&amp;nbsp;&lt;SPAN style="font-size: 1em;"&gt;want to measure the number of &lt;/SPAN&gt;dTLB&lt;SPAN style="font-size: 1em;"&gt; misses.&lt;/SPAN&gt;&lt;/P&gt;

&lt;P&gt;&lt;SPAN style="font-size: 1em;"&gt;I have three questions.&lt;/SPAN&gt;&lt;/P&gt;

&lt;P&gt;1. Can I measure the number of L1 dTLB misses by the sum of DTLB_LOAD_MISSES.STLB_HIT, DTLB_STORE_MISSES.STLB_HIT, DTLB_LOAD_MISSES.MISS_CUASES_A_WALK, and DTLB_STORE_MISSES.MISS_CAUSES_A_WALK?&lt;/P&gt;

&lt;P&gt;2. Does it also contain the TLB miss of huge pages?&lt;/P&gt;

&lt;P&gt;3. Does LLC miss count contain DTLB miss? (I measured the number of LLC misses by OFFCORE_RESPONSES.ALL_READS.LLC_MISS.ANY_RESPONSE_0)&lt;/P&gt;

&lt;P&gt;Thank you.&lt;/P&gt;</description>
    <pubDate>Tue, 11 Sep 2018 06:08:31 GMT</pubDate>
    <dc:creator>Yukyoung_L_</dc:creator>
    <dc:date>2018-09-11T06:08:31Z</dc:date>
    <item>
      <title>TLB miss</title>
      <link>https://community.intel.com/t5/Software-Tuning-Performance/TLB-miss/m-p/1126715#M6323</link>
      <description>&lt;P&gt;Hello.&lt;/P&gt;

&lt;P&gt;I am working on IvyTown processor, and&amp;nbsp;&lt;SPAN style="font-size: 1em;"&gt;want to measure the number of &lt;/SPAN&gt;dTLB&lt;SPAN style="font-size: 1em;"&gt; misses.&lt;/SPAN&gt;&lt;/P&gt;

&lt;P&gt;&lt;SPAN style="font-size: 1em;"&gt;I have three questions.&lt;/SPAN&gt;&lt;/P&gt;

&lt;P&gt;1. Can I measure the number of L1 dTLB misses by the sum of DTLB_LOAD_MISSES.STLB_HIT, DTLB_STORE_MISSES.STLB_HIT, DTLB_LOAD_MISSES.MISS_CUASES_A_WALK, and DTLB_STORE_MISSES.MISS_CAUSES_A_WALK?&lt;/P&gt;

&lt;P&gt;2. Does it also contain the TLB miss of huge pages?&lt;/P&gt;

&lt;P&gt;3. Does LLC miss count contain DTLB miss? (I measured the number of LLC misses by OFFCORE_RESPONSES.ALL_READS.LLC_MISS.ANY_RESPONSE_0)&lt;/P&gt;

&lt;P&gt;Thank you.&lt;/P&gt;</description>
      <pubDate>Tue, 11 Sep 2018 06:08:31 GMT</pubDate>
      <guid>https://community.intel.com/t5/Software-Tuning-Performance/TLB-miss/m-p/1126715#M6323</guid>
      <dc:creator>Yukyoung_L_</dc:creator>
      <dc:date>2018-09-11T06:08:31Z</dc:date>
    </item>
    <item>
      <title>Hi,</title>
      <link>https://community.intel.com/t5/Software-Tuning-Performance/TLB-miss/m-p/1126716#M6324</link>
      <description>&lt;P&gt;Hi,&lt;/P&gt;

&lt;P&gt;1. Yes the event set&amp;nbsp;looks correct. Note that IVT also has DTLB_LOAD_MISSES.WALK_DURATION and DTLB_STORE_MISSES.WALK_DURATION&amp;nbsp;events - they will be more&amp;nbsp;useful if you want to estimate performance impact rather than raw count of misses.&lt;/P&gt;

&lt;P&gt;2. Yes according to events documentation all page sizes should be covered by these events&lt;/P&gt;

&lt;P&gt;3. No I think the situation when you have DTLB miss but LLC (or L2 or even L1)&amp;nbsp;hit for actual data is quite&amp;nbsp;possible.&lt;/P&gt;</description>
      <pubDate>Mon, 17 Sep 2018 08:47:13 GMT</pubDate>
      <guid>https://community.intel.com/t5/Software-Tuning-Performance/TLB-miss/m-p/1126716#M6324</guid>
      <dc:creator>Dmitry_R_Intel1</dc:creator>
      <dc:date>2018-09-17T08:47:13Z</dc:date>
    </item>
  </channel>
</rss>

