bzr commit into mysql-trunk-bugfixing branch (tor.didriksen:3259) WL#1393

[Tor Didriksen]

#At file:///export/home/didrik/repo/next-mr-opt-team-wl1393-merge/ based on revid:tor.didriksen@stripped

 3259 Tor Didriksen	2010-12-10
      WL#1393 Optimizing filesort with small limit
     @ sql/filesort.cc
        Adjust, and comment on, cost function.

    modified:
      mysql-test/r/order_by_sortkey.result
      mysql-test/t/order_by_sortkey.test
      sql/filesort.cc
=== modified file 'mysql-test/r/order_by_sortkey.result'
--- a/mysql-test/r/order_by_sortkey.result	2010-11-30 11:54:00 +0000
+++ b/mysql-test/r/order_by_sortkey.result	2010-12-10 15:06:03 +0000
@@ -40,8 +40,6 @@ INSERT INTO tmp SELECT f1,f2 FROM t1;
 INSERT INTO t1(f1,f2) SELECT * FROM tmp;
 INSERT INTO tmp SELECT f1,f2 FROM t1;
 INSERT INTO t1(f1,f2) SELECT * FROM tmp;
-INSERT INTO tmp SELECT f1,f2 FROM t1;
-INSERT INTO t1(f1,f2) SELECT * FROM tmp;
 set sort_buffer_size= 32768;
 FLUSH STATUS;
 SHOW SESSION STATUS LIKE 'Sort%';

=== modified file 'mysql-test/t/order_by_sortkey.test'
--- a/mysql-test/t/order_by_sortkey.test	2010-11-30 11:54:00 +0000
+++ b/mysql-test/t/order_by_sortkey.test	2010-12-10 15:06:03 +0000
@@ -50,8 +50,6 @@ INSERT INTO tmp SELECT f1,f2 FROM t1;
 INSERT INTO t1(f1,f2) SELECT * FROM tmp; 
 INSERT INTO tmp SELECT f1,f2 FROM t1;
 INSERT INTO t1(f1,f2) SELECT * FROM tmp; 
-INSERT INTO tmp SELECT f1,f2 FROM t1;
-INSERT INTO t1(f1,f2) SELECT * FROM tmp;
 
 # Test when only sortkeys fits to memory
 set sort_buffer_size= 32768; 

=== modified file 'sql/filesort.cc'
--- a/sql/filesort.cc	2010-12-09 11:54:39 +0000
+++ b/sql/filesort.cc	2010-12-10 15:06:03 +0000
@@ -218,7 +218,8 @@ ha_rows filesort(THD *thd, TABLE *table,
       !(param.tmp_buffer= (char*) my_malloc(param.sort_length,MYF(MY_WME))))
     goto err;
 
-  if (check_if_pq_applicable(&param, &table_sort,
+  if (param.max_rows != HA_POS_ERROR &&
+      check_if_pq_applicable(&param, &table_sort,
                              table, num_rows, memory_available))
   {
     DBUG_PRINT("info", ("filesort PQ is applicable"));
@@ -1174,8 +1175,7 @@ bool check_if_pq_applicable(Sort_param *
     {
       make_char_array(filesort_info,
                       param->max_keys_per_buffer, param->rec_length);
-      if (filesort_info->sort_keys)
-        DBUG_RETURN(true);
+      DBUG_RETURN(filesort_info->sort_keys != NULL);
     }
     else
     {
@@ -1189,8 +1189,7 @@ bool check_if_pq_applicable(Sort_param *
   {
     make_char_array(filesort_info,
                     param->max_keys_per_buffer, param->rec_length);
-    if (filesort_info->sort_keys)
-      DBUG_RETURN(true);
+    DBUG_RETURN(filesort_info->sort_keys != NULL);
   }
 
   // Try to strip off addon fields.
@@ -1211,11 +1210,17 @@ bool check_if_pq_applicable(Sort_param *
         PQ has cost:
         (insert + qsort) * log(queue size) / TIME_FOR_COMPARE_ROWID +
         cost of file lookup afterwards.
+        The lookup cost is a bit pessimistic: we take scan_time and assume
+        that on average we find the row after scanning half of the file.
+        A better estimate would be lookup cost, but note that we are doing
+        random lookups here, rather than sequential scan.
       */
-      const double pq_cost=
+      const double pq_cpu_cost= 
         (PQ_slowness * num_rows + param->max_keys_per_buffer) *
-        log((double) param->max_keys_per_buffer) / TIME_FOR_COMPARE_ROWID +
-        param->max_rows * table->file->scan_time();
+        log((double) param->max_keys_per_buffer) / TIME_FOR_COMPARE_ROWID;
+      const double pq_io_cost=
+        param->max_rows * table->file->scan_time() / 2.0;
+      const double pq_cost= pq_cpu_cost + pq_io_cost;
 
       if (sort_merge_cost < pq_cost)
         DBUG_RETURN(false);


Attachment: [text/bzr-bundle] bzr/[e-mail]