<?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 32 bit outperforming 64 bit by factor of 2-3 in QlikView</title>
    <link>https://community.qlik.com/t5/QlikView/32-bit-outperforming-64-bit-by-factor-of-2-3/m-p/228068#M80129</link>
    <description>&lt;HTML&gt;&lt;HEAD&gt;&lt;/HEAD&gt;&lt;BODY&gt;&lt;P&gt;I have been doing some benchmarking between the 32 and 64 bit versions of QV, and have found a significant discrepancy in performance for a Pivot Table calc times, which I think may force us to use the 32 bit server if a better solution cannot be found. I am running QV 9 SR3 Build 7440&lt;/P&gt;&lt;P&gt;I suspect that we're doing something funky that most other QV users aren't doing (i.e. large expressions with conditional logic containing set expressions. Interestingly I've separately tested with two different expressions: one dealing with averages (i.e. floating point operations) and the second with only counts and sums (i.e. integer operations).&lt;/P&gt;&lt;P&gt;I should also point out that until SR3 came out, this code didn't work at all in the 64 bit version, so I wonder if my suboptimal performance is related to this patched bug.&lt;/P&gt;&lt;P&gt;I've pasted the Pivot Table dimension definitions below (the first 5 dimensions are primary [i.e. rows], and the 6th dimension is secondary [i.e. columns]).&lt;/P&gt;&lt;P&gt;If anyone has any ideas I'm all ears&lt;/P&gt;&lt;P&gt;Dimensions:&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;PRE ___default_attr="plain" class="jive_text_macro jive_macro_code" jivemacro="code"&gt;&lt;BR /&gt;ORG1_MemberName1&lt;BR /&gt;=if($(CurOrgLevel) &amp;gt;= 2, ORG1_MemberName2)&lt;BR /&gt;=if($(CurOrgLevel) &amp;gt;= 3, ORG1_MemberName3)&lt;BR /&gt;=if($(CurOrgLevel) &amp;gt;= 4, ORG1_MemberName4)&lt;BR /&gt;=if($(CurOrgLevel) &amp;gt;= 5, ORG1_MemberName5)&lt;BR /&gt;=$(PdRollupField)&lt;BR /&gt;&lt;/PRE&gt;&lt;BR /&gt;&lt;BR /&gt; &lt;P&gt;Variables:&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;PRE ___default_attr="plain" class="jive_text_macro jive_macro_code" jivemacro="code"&gt;&lt;BR /&gt;CurOrgLevel = 2&lt;BR /&gt;PdRollupField = PD1_PeriodName2&lt;BR /&gt;PdResp1 = {&amp;lt;SG1_path=,ORG1_path=,ORG1_level={1},ORG1_MemberName1={'Acme Co'}&amp;gt;}&lt;BR /&gt;PdResp2 = {&amp;lt;SG1_path=,ORG1_path=,ORG1_level={1},ORG1_MemberName2={'Montana','Idaho','Nevada'}&amp;gt;}&lt;BR /&gt;PdResp3 =&lt;BR /&gt;PdResp4 =&lt;BR /&gt;PdResp5 =&lt;BR /&gt;&lt;/PRE&gt;&lt;BR /&gt;&lt;BR /&gt; &lt;P&gt;Current Selections:&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;PRE ___default_attr="plain" class="jive_text_macro jive_macro_code" jivemacro="code"&gt;&lt;BR /&gt;SG1_path = 'Question 1'&lt;BR /&gt;PD1_path = 96 of 296&lt;BR /&gt;ORG1_path = NOT 'Acme Co'&lt;BR /&gt;&lt;/PRE&gt;&lt;BR /&gt;&lt;BR /&gt; &lt;P&gt;&lt;/P&gt;&lt;P&gt;Chart Expression:&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;PRE ___default_attr="plain" class="jive_text_macro jive_macro_code" jivemacro="code"&gt;&lt;BR /&gt;&lt;BR /&gt;If(dimensionality()=5 AND $(CurOrgLevel) &amp;gt;= 5,&lt;BR /&gt;count($(PdResp5) distinct FT_ResponseId)&lt;BR /&gt;,&lt;BR /&gt; if(dimensionality()=4 AND $(CurOrgLevel) &amp;gt;= 4,&lt;BR /&gt;sum($(PdResp4) aggr(count($(PdResp4) distinct FT_ResponseId), ORG1_MemberName4, $(PdRollupField)))&lt;BR /&gt;,&lt;BR /&gt; if(dimensionality()=3 AND $(CurOrgLevel) &amp;gt;= 3,&lt;BR /&gt;sum($(PdResp3) aggr(count($(PdResp3) distinct FT_ResponseId), ORG1_MemberName3, $(PdRollupField)))&lt;BR /&gt;,&lt;BR /&gt; if(dimensionality()=2 AND $(CurOrgLevel) &amp;gt;= 2,&lt;BR /&gt;sum($(PdResp2) aggr(count($(PdResp2) distinct FT_ResponseId), ORG1_MemberName2, $(PdRollupField)))&lt;BR /&gt;,&lt;BR /&gt; if (dimensionality()=1 AND $(CurOrgLevel) &amp;gt;= 1,&lt;BR /&gt;sum($(PdResp1) aggr(count($(PdResp1) distinct FT_ResponseId), ORG1_MemberName2, $(PdRollupField)))&lt;BR /&gt; )&lt;BR /&gt; )&lt;BR /&gt; )&lt;BR /&gt; )&lt;BR /&gt;)&lt;BR /&gt;&lt;BR /&gt;&lt;/PRE&gt;&lt;BR /&gt;&lt;BR /&gt;&lt;/BODY&gt;&lt;/HTML&gt;</description>
    <pubDate>Thu, 29 Apr 2010 16:25:36 GMT</pubDate>
    <dc:creator />
    <dc:date>2010-04-29T16:25:36Z</dc:date>
    <item>
      <title>32 bit outperforming 64 bit by factor of 2-3</title>
      <link>https://community.qlik.com/t5/QlikView/32-bit-outperforming-64-bit-by-factor-of-2-3/m-p/228068#M80129</link>
      <description>&lt;HTML&gt;&lt;HEAD&gt;&lt;/HEAD&gt;&lt;BODY&gt;&lt;P&gt;I have been doing some benchmarking between the 32 and 64 bit versions of QV, and have found a significant discrepancy in performance for a Pivot Table calc times, which I think may force us to use the 32 bit server if a better solution cannot be found. I am running QV 9 SR3 Build 7440&lt;/P&gt;&lt;P&gt;I suspect that we're doing something funky that most other QV users aren't doing (i.e. large expressions with conditional logic containing set expressions. Interestingly I've separately tested with two different expressions: one dealing with averages (i.e. floating point operations) and the second with only counts and sums (i.e. integer operations).&lt;/P&gt;&lt;P&gt;I should also point out that until SR3 came out, this code didn't work at all in the 64 bit version, so I wonder if my suboptimal performance is related to this patched bug.&lt;/P&gt;&lt;P&gt;I've pasted the Pivot Table dimension definitions below (the first 5 dimensions are primary [i.e. rows], and the 6th dimension is secondary [i.e. columns]).&lt;/P&gt;&lt;P&gt;If anyone has any ideas I'm all ears&lt;/P&gt;&lt;P&gt;Dimensions:&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;PRE ___default_attr="plain" class="jive_text_macro jive_macro_code" jivemacro="code"&gt;&lt;BR /&gt;ORG1_MemberName1&lt;BR /&gt;=if($(CurOrgLevel) &amp;gt;= 2, ORG1_MemberName2)&lt;BR /&gt;=if($(CurOrgLevel) &amp;gt;= 3, ORG1_MemberName3)&lt;BR /&gt;=if($(CurOrgLevel) &amp;gt;= 4, ORG1_MemberName4)&lt;BR /&gt;=if($(CurOrgLevel) &amp;gt;= 5, ORG1_MemberName5)&lt;BR /&gt;=$(PdRollupField)&lt;BR /&gt;&lt;/PRE&gt;&lt;BR /&gt;&lt;BR /&gt; &lt;P&gt;Variables:&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;PRE ___default_attr="plain" class="jive_text_macro jive_macro_code" jivemacro="code"&gt;&lt;BR /&gt;CurOrgLevel = 2&lt;BR /&gt;PdRollupField = PD1_PeriodName2&lt;BR /&gt;PdResp1 = {&amp;lt;SG1_path=,ORG1_path=,ORG1_level={1},ORG1_MemberName1={'Acme Co'}&amp;gt;}&lt;BR /&gt;PdResp2 = {&amp;lt;SG1_path=,ORG1_path=,ORG1_level={1},ORG1_MemberName2={'Montana','Idaho','Nevada'}&amp;gt;}&lt;BR /&gt;PdResp3 =&lt;BR /&gt;PdResp4 =&lt;BR /&gt;PdResp5 =&lt;BR /&gt;&lt;/PRE&gt;&lt;BR /&gt;&lt;BR /&gt; &lt;P&gt;Current Selections:&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;PRE ___default_attr="plain" class="jive_text_macro jive_macro_code" jivemacro="code"&gt;&lt;BR /&gt;SG1_path = 'Question 1'&lt;BR /&gt;PD1_path = 96 of 296&lt;BR /&gt;ORG1_path = NOT 'Acme Co'&lt;BR /&gt;&lt;/PRE&gt;&lt;BR /&gt;&lt;BR /&gt; &lt;P&gt;&lt;/P&gt;&lt;P&gt;Chart Expression:&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;PRE ___default_attr="plain" class="jive_text_macro jive_macro_code" jivemacro="code"&gt;&lt;BR /&gt;&lt;BR /&gt;If(dimensionality()=5 AND $(CurOrgLevel) &amp;gt;= 5,&lt;BR /&gt;count($(PdResp5) distinct FT_ResponseId)&lt;BR /&gt;,&lt;BR /&gt; if(dimensionality()=4 AND $(CurOrgLevel) &amp;gt;= 4,&lt;BR /&gt;sum($(PdResp4) aggr(count($(PdResp4) distinct FT_ResponseId), ORG1_MemberName4, $(PdRollupField)))&lt;BR /&gt;,&lt;BR /&gt; if(dimensionality()=3 AND $(CurOrgLevel) &amp;gt;= 3,&lt;BR /&gt;sum($(PdResp3) aggr(count($(PdResp3) distinct FT_ResponseId), ORG1_MemberName3, $(PdRollupField)))&lt;BR /&gt;,&lt;BR /&gt; if(dimensionality()=2 AND $(CurOrgLevel) &amp;gt;= 2,&lt;BR /&gt;sum($(PdResp2) aggr(count($(PdResp2) distinct FT_ResponseId), ORG1_MemberName2, $(PdRollupField)))&lt;BR /&gt;,&lt;BR /&gt; if (dimensionality()=1 AND $(CurOrgLevel) &amp;gt;= 1,&lt;BR /&gt;sum($(PdResp1) aggr(count($(PdResp1) distinct FT_ResponseId), ORG1_MemberName2, $(PdRollupField)))&lt;BR /&gt; )&lt;BR /&gt; )&lt;BR /&gt; )&lt;BR /&gt; )&lt;BR /&gt;)&lt;BR /&gt;&lt;BR /&gt;&lt;/PRE&gt;&lt;BR /&gt;&lt;BR /&gt;&lt;/BODY&gt;&lt;/HTML&gt;</description>
      <pubDate>Thu, 29 Apr 2010 16:25:36 GMT</pubDate>
      <guid>https://community.qlik.com/t5/QlikView/32-bit-outperforming-64-bit-by-factor-of-2-3/m-p/228068#M80129</guid>
      <dc:creator />
      <dc:date>2010-04-29T16:25:36Z</dc:date>
    </item>
  </channel>
</rss>

