====================================================================================================
SYSBENCH BENCHMARK REPORT 
====================================================================================================
MACHINE:  r815-test2
RUN ID:   183
RUN DATE: 2010-11-08T23:05:48.672479
WORKLOAD: innodb_1000K_readwrite
SERVER:   drizzled
VERSION:  staging
REVISION: 1915
COMMENT:  1915: Brian Aker 2010-11-08 [merge] Merge in Stewart's change to blob 
field size.
====================================================================================================

TRENDING OVER LAST 5 runs 
Conc   TPS     % Diff from Avg   Diff       Min        Max        Avg        
STD       
====================================================================================================
16     1756.87       +2.25%      38.58    1679.63    1775.14    1718.29      
29.29
32     2081.78       +0.09%       1.84    1954.58    2156.79    2079.94      
62.59
64     1452.89       +1.21%      17.37    1376.34    1476.27    1435.52       
29.6
128    1055.47       -4.29%     -47.26    1047.15    1144.71    1102.74      
32.19
256    885.51       +0.99%       8.70     862.51     900.78     876.81       
9.42
512    628.62       +1.85%      11.41     606.58     632.47     617.21       
8.31
1024   396.19       +0.89%       3.49     383.34     401.73     392.69       
7.16
====================================================================================================

TRENDING OVER Last 20 runs 

Conc   TPS     % Diff from Avg   Diff       Min        Max        Avg        
STD       
====================================================================================================
16     1756.87       +2.34%      40.12    1654.62    1775.14    1716.75      
28.83
32     2081.78       -0.42%      -8.86    1954.58    2241.22    2090.64      
55.78
64     1452.89       -0.32%      -4.62    1376.34    1630.25    1457.51      
45.03
128    1055.47       -4.85%     -53.85    1045.19    1229.75    1109.32      
38.53
256    885.51       +0.66%       5.82     847.09     924.15     879.69      
18.33
512    628.62       +0.94%       5.83     605.91     739.79     622.79      
25.65
1024   396.19       -1.15%      -4.60     383.34     477.96     400.79      
17.53
====================================================================================================

TRENDING OVER ALL runs 

Conc   TPS     % Diff from Avg   Diff       Min        Max        Avg        
STD       
====================================================================================================
16     1756.87       +3.59%      60.88     922.18    1869.76    1695.99     
174.78
32     2081.78       -1.12%     -23.48    1262.41    2395.61    2105.26     
208.38
64     1452.89       -4.01%     -60.76    1375.75    1721.72    1513.65      
86.39
128    1055.47       -8.24%     -94.76    1045.19    1345.78    1150.23      
72.74
256    885.51       -1.37%     -12.26     541.64    1034.59     897.77      
77.04
512    628.62       -1.85%     -11.85     375.56     781.01     640.47      
72.82
1024   396.19       -5.54%     -23.24     383.34     487.49     419.43      
32.47
====================================================================================================
FULL REVISION COMMENTARY:

  1267.1.8: Stewart Smith 2010-11-08 change dynamic_cast in field_conv.cc back 
to static_cast for runtime performance
  1267.1.7: Stewart Smith 2010-11-08 in field_conv.cc, get_copy_func should not 
try to cast Field_blob to Field_varstring as Field_ blob does not inherit from 
field_varstring. Previously this kinda worked out due to the first member of 
field_blob kinda making sense in this comparison. Instead, we just shortcut to 
that decision if we are converting from BLOB.
  1267.1.6: Stewart Smith 2010-11-08 change cast in field_conv.cc that was 
producing valgrind warning to dynamic_cast to crash on invalid cast. this shows 
up in several tests: e.g. func_gconcat
  1267.1.5: Stewart Smith 2010-11-07 [merge] merge trunk
  1267.1.4: Stewart Smith 2009-12-09 remove last bits of packlength from 
Field_blob, leaving only support for 4 byte length for blobs (not 1, not 2 and 
certainly not 3).
  1267.1.3: Stewart Smith 2009-12-09 remove Field_blob constructor that had 
set_packflag argument being called in Item::tmp_table_field_from_field_type(). 
This would cause a Field_blob to be created that used 3 bytes instead of 4 for 
length... potentially causing all sorts of trouble
  1267.1.2: Stewart Smith 2009-12-09 remove Field_blob constructor that only 
takes packlength as argument. It's never even called
  1267.1.1: Stewart Smith 2009-12-09 remove blob_pack_length argument to 
Field_blob constructor.

_______________________________________________
Mailing list: https://launchpad.net/~drizzle-benchmark
Post to     : [email protected]
Unsubscribe : https://launchpad.net/~drizzle-benchmark
More help   : https://help.launchpad.net/ListHelp

Reply via email to