SQL优化-不存在自连接的地方

请提供有关如何优化此sql过程的建议。。。where not exists子选择是我的主要问题。有人告诉我，我可以获得40%的性能提升

   insert into #tmpTable (intID3,intID4,intID5,intID6, 
          StaticDate,NewDate, 
          [Description],Change,intSort) 
   select distinct 
          '' as intID3,           
          intID4,
          intID5, 
          intID6, 
          @dteDate1 as StaticDate, 
          '' as NewDate,           
          '16 character str' as [Description], 
          0 as Movement,
          0 as intSort

   from #tmpTable j 
   where not exists ( 
        select 
               1 
        from #tmpTable x 
        where
               x.intID1 = j.intID1 
               and x.intID2 = j.intID2 
               and x.[Description] = '16 character str' 
   ) 

---

尝试下面的查询

 INSERT INTO #tmpTable2 
     (intID3,intID4,intID5,intID6, 
      StaticDate,NewDate, 
      [Description],Change,intSort)
  SELECT DISTINCT 
      '' as intID3,           
      j.intID4,
      j.intID5, 
      j.intID6, 
      @dteDate1 as StaticDate, 
      '' as NewDate,           
      '16 character str' as [Description], 
      0 as Movement,
      0 as intSort
 FROM #tmpTable j 
       LEFT JOIN #tmpTable x  
             ON  x.intID1 = j.intID1 
                  AND x.intID2 = j.intID2 
                  AND x.[Description] = '16 character str' 
 WHERE x.intID1 IS NULL 

---

请发布执行计划、涉及的表的架构和它们的计数为什么不使用

WHERE j.[Description]！='16个字符的str'

？因为x和j不是同一条记录。我们将表本身连接起来，以确保没有与特定id匹配的特定文本的记录您在tmpTable上有任何唯一的id吗？Genius。。。我喜欢这种观点。。。但有一个问题……这不应该是右联接吗？@Demian这取决于您。假设您在表1中输入数据，您正试图在不输入重复记录的情况下将该数据插入表2中，根据脚本，输入表表1应该位于联接的左侧（左联接）。