Documentation
Rate This Document
Findability
Accuracy
Completeness
Readability
SubmitClose

Using the Training Result

The tuning result processing tool facilitates report display, single-round value setting, retests, and importance analysis in interactive mode.

Command Function

Follows the tuning result to display performance results, retest performance using parameters of a round, assign values to application parameters, reset parameters, and analyze parameter importance.

Syntax

1
devkit kat use [-h] -i <dir> [-l {0,1,2,3}]

Parameter Description

Table 1 Parameter description

Parameter

Option

Description

-h/--help

-

Obtains help information. This parameter is optional.

-l/--log-level

0/1/2/3
Log level, which defaults to 2. This parameter is optional.
NOTE:

The default level is 2 (WARNING).

  • 0: DEBUG
  • 1: INFO
  • 2: WARNING
  • 3: ERROR

-i/--input

-

Sets the result directory of automatic tuning. This parameter is mandatory.
Table 2 Subcommand description

Subcommand

Option

Description

example

-

Displays examples.

exit

-

Exits the interactive user interface.

help

-

Obtains help information.

ibesttest

-

Uses the optimal parameters inferred from parameter importance analysis to perform tests.

NOTE:

If you are uncertain about the optimal parameters, you can run this command to infer the optimal parameters for testing.

iexample

-

Displays example subcommands of parameter importance.

iglobal

-

Displays the importance of all parameters for tuning.

NOTE:

You can run this command to check the parameters that can improve performance during multiple rounds of tuning. In the command output, a higher value in the importance(%) column indicates a stronger correlation between the parameter and the tuning effect.

iround

-

Displays the importance of local parameters in the nth round.

NOTE:

If you are uncertain about the performance data of a round, you can run this command to check how the parameters of that round contribute to performance improvement.

log

0/1/2/3
Log level, which defaults to 2.
  • 0: DEBUG
  • 1: INFO
  • 2: WARNING
  • 3: ERROR

reset

-

Restores the application parameters and system parameters to the original values before value assignment.

set

-

Applies the parameter values of a tuning round to the application and system.

NOTE:
  • This subcommand is suitable for the scenario where users are satisfied with the performance data of a certain tuning round.
  • The set subcommand is used to change the parameter values. Exercise caution when running this subcommand.

show

-

Displays the performance test data of some rounds.

test

-

Performs the tests again using the same parameters of some rounds.

top

-

Displays the data of the top N rounds with the best performance results.

train

-

Resumes automatic tuning. The Kunpeng AutoTuner automatically tunes task parameters based on service scenario metrics.
  • You can append r plus a number to a subcommand to specify the round of ranking by performance. For example:

    show r1: Display the result data of the round whose performance ranks first.

    set r1: Apply the task parameters of the round whose performance ranks first to the application and system.

    test r1: Use the parameters of the round whose performance ranks first to perform the performance test again.

  • You can append a single number to a subcommand to indicate a round. For example:

    show 5: Display the data of the fifth round of automatic tuning.

    set 5: Apply the task parameters of the fifth round to the application and system.

    test 5: Use the task parameters of the fifth round to perform the performance test again.

Table 3 train subcommand parameters

Parameter

Description

-h/--help

Obtains help information. This parameter is optional.

-r/--round

Number of training rounds, which defaults to 50. This parameter is optional.

--algorithm

Training algorithm, which must be HPO. This parameter is optional.

-i/--initial-random-ratio

Initial random ratio. A larger value indicates more random training rounds. The value ranges from 0 to 1 and defaults to 0. This parameter is optional.

Example

View the automatic tuning result report. (The last command output in Enabling Automatic Tuning shows the result directory.)

The parameter set generated after automatic tuning is located in the /opt/template/template_Postgresql_20250119_073054/train-20250225-112832 directory.

1
devkit kat use -i /opt/template/template_Postgresql_20250119_073054/train-20250225-112832

Command output:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
Available Commands:

    example     - Show examples for use
    exit        - Exit the program
    help        - Show this help message
    ibesttest   - test best parameters inferenced by parameter importance analysis
    iexample    - Show examples for iglobal and iround
    iglobal     - show global parameter importance
    iround      - show local parameter importance of round <n>
    log         - change the log level
    reset       - reset params
    set         - set the config of round n.
    show        - show round <n> information.
    test        - Retest round sequence. test <n> times.
    top         - show top N performance
    train       - Additional training after original result

For more usage, please refer to "help <command>". e.g., help test
  • Viewing the example

    Type example and press Enter.

    The following is a simple example of the command output:

     1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
    To understand the use function, see the following example report of automatic
training.

Example report of an automatic training task (including multiple rounds of
performance test results and performance improvement data):

Total round : 10
Baseline performance: 5.0
--------------------------------------------------------------------------------
| Rank |     Round     |  Performance   |  Improvement  (%) |
--------------------------------------------------------------------------------
|   1  |   round    5  |      10.0      |      100.00       |
|   2  |   round    3  |       9.0      |       80.00       |
|   3  |   round    2  |       8.0      |       60.00       |
|   4  |   round    4  |       7.0      |       40.00       |
|   5  |   round    1  |       6.0      |       20.00       |
--------------------------------------------------------------------------------

Usage:
    The show, set, and test subcommands display the data of each round, assign
    a value to the task parameter of each round, and use the task parameter of
    each round for retesting.

    (1) You can add the letter r plus a number to a subcommand to specify which
    round to be ranked by performance. For example, r1 indicates the first row
    of the training result. You can use:

    show r1      # Display the result data of the round whose performance ranks
                 # first.
    set r1       # Assign a value to the task parameter of the first round of
                 # performance task.
    test r1      # Use the parameter of the first round of performance test to
                 # perform the performance test again.

    (2) You can add a number to a subcommand to specify a specific round. For
    example, perform operations on the fifth round result (the first row in
    the training result table):

    show 5       # Display the result data of the fifth round of automatic
                 # training.
    set 5        # Assign a value to the task parameter of the fifth round of
                 # task.
    test 5       # Use the parameter of the fifth round of task to perform the
                 # performance test again.

    (3) When using the test subcommand, you can specify multiple rounds of data
    and use the round parameters to perform the performance test again. Separate
    the rounds using commas (,).
    For example:

    test 1,2,r1,r2    # Use the parameters of the first and second training rounds
                      # and the first and second rounds with the best performance
                      # data to perform the performance test again (corresponding
                      # to rows 1, 2, 5, and 3 in the table).

    You can add top or n-m to the test command to specify multiple rounds.
    For example:
    test 1-3,top3     # Rounds 1, 2, and 3 (corresponding to rows 2, 3, and 5 in
                      # the table) and the three rounds with the best performance
                      # data (corresponding to rows 1, 2, and 3 in the table)

    You can add the specific rounds to the test command and then a space and a
    number after the specific rounds to specify the number of repeated performance
    tests for these rounds. For example:
    test 1,2 3        # Retest rounds: 1,2,1,2,1,2
  • Obtaining help information

    Type help and press Enter. You can type a subcommand next to help to obtain the help information about the subcommand.

     1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
    Available Commands:

    example     - Show examples for use
    exit        - Exit the program
    help        - Show this help message
    
    iexample    - Show examples for iglobal and iround
    iglobal     - show global parameter importance
    iround      - show local parameter importance of round <n>
    log         - change the log level
    reset       - reset params
    set         - set the config of round n.
    show        - show round <n> information.
    test        - Retest round sequence. test <n> times.
    top         - show top N performance

For more usage, please refer to "help <command>". e.g., help test
  • Viewing the performance test data

    Type show 1 and press Enter to view the parameters and performance data of the first round of automatic tuning. Type show and press Enter to view the performance data of all rounds.

    Command output:

     1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
    [2025-02-25 09:52:11 UTC] [KAT] [message] - ======================== info of Round   1   ===================================
[2025-02-25 09:52:11 UTC] [KAT] [message] - # Performance: 91.07
[2025-02-25 09:52:11 UTC] [KAT] [message] - --------------------------------------------------------------------------------
[2025-02-25 09:52:11 UTC] [KAT] [message] - ----------------------          Params         ---------------------------------
[2025-02-25 09:52:11 UTC] [KAT] [message] - --------------------------------------------------------------------------------
[2025-02-25 09:52:11 UTC] [KAT] [message] -     Postgresql:
[2025-02-25 09:52:11 UTC] [KAT] [message] -         autovacuum_vacuum_cost_delay                                 = 96
[2025-02-25 09:52:11 UTC] [KAT] [message] -         autovacuum_vacuum_cost_limit                                 = 1994
[2025-02-25 09:52:11 UTC] [KAT] [message] -         autovacuum_vacuum_scale_factor                               = 0.07744139964226633
[2025-02-25 09:52:11 UTC] [KAT] [message] -         autovacuum_vacuum_threshold                                  = 714
...
...
...
[2025-02-25 09:52:11 UTC] [KAT] [message] -         parallel_leader_participation                                = on
[2025-02-25 09:52:11 UTC] [KAT] [message] -         vacuum_cleanup_index_scale_factor                            = 0.8960460592061281
[2025-02-25 09:52:11 UTC] [KAT] [message] - --------------------------------------------------------------------------------
[2025-02-25 09:52:11 UTC] [KAT] [message] - ----------------------          Run Info        --------------------------------
[2025-02-25 09:52:11 UTC] [KAT] [message] - --------------------------------------------------------------------------------
[2025-02-25 09:52:11 UTC] [KAT] [message] - Run record: 1
[2025-02-25 09:52:11 UTC] [KAT] [message] - Result: success
[2025-02-25 09:52:11 UTC] [KAT] [message] - Performance: 91.07
[2025-02-25 09:52:11 UTC] [KAT] [message] - Run time: 2025-02-25 09:38:24.265750
[2025-02-25 09:52:11 UTC] [KAT] [message] - Finish time: P2025-02-25 09:38:24.265759
[2025-02-25 09:52:11 UTC] [KAT] [message] -
[2025-02-25 09:52:11 UTC] [KAT] [message] - ================================================================================
  • Viewing the data of the three rounds with the highest performance

    Type top 3 and press Enter.

    Command output:

     1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
    [2025-02-25 09:53:05 UTC] [KAT] [message] - ============================== Auto Tuning Report ==============================
[2025-02-25 09:53:05 UTC] [KAT] [message] - Total round             : 5
[2025-02-25 09:53:05 UTC] [KAT] [message] - Total run               : 6
[2025-02-25 09:53:05 UTC] [KAT] [message] - Fail times              : 0
[2025-02-25 09:53:05 UTC] [KAT] [message] - Application             : Postgresql
[2025-02-25 09:53:05 UTC] [KAT] [message] - Application version     : 11.3
[2025-02-25 09:53:05 UTC] [KAT] [message] - Test tool               : benchmarksql
[2025-02-25 09:53:05 UTC] [KAT] [message] - Tuning direction        : high
[2025-02-25 09:53:05 UTC] [KAT] [message] - Algorithm               : HPO
[2025-02-25 09:53:05 UTC] [KAT] [message] - Random initial ratio    : 0.5
[2025-02-25 09:53:05 UTC] [KAT] [message] - Performance description : tpmC, transactions per minute
[2025-02-25 09:53:05 UTC] [KAT] [message] - Parameter groups        : Postgresql
[2025-02-25 09:53:05 UTC] [KAT] [message] - Baseline performance    : 97.03
[2025-02-25 09:53:05 UTC] [KAT] [message] - Top 3 performance :
[2025-02-25 09:53:05 UTC] [KAT] [message] - --------------------------------------------------------------------------------
[2025-02-25 09:53:05 UTC] [KAT] [message] - |  Rank  |     Round     |  Performance   |  Improvement  (%) |
[2025-02-25 09:53:05 UTC] [KAT] [message] - --------------------------------------------------------------------------------
[2025-02-25 09:53:05 UTC] [KAT] [message] - |   1    |   round  1    |     91.07      |       -6.14       |
[2025-02-25 09:53:05 UTC] [KAT] [message] - |   2    |   round  4    |     86.92      |      -10.42       |
[2025-02-25 09:53:05 UTC] [KAT] [message] - |   3    |   round  3    |     79.44      |      -18.13       |
[2025-02-25 09:53:05 UTC] [KAT] [message] - --------------------------------------------------------------------------------
[2025-02-25 09:53:05 UTC] [KAT] [message] - Note:
[2025-02-25 09:53:05 UTC] [KAT] [message] - The performance value is the return value of the run test step,
[2025-02-25 09:53:05 UTC] [KAT] [message] - Performance improvement (%) =
[2025-02-25 09:53:05 UTC] [KAT] [message] -     (round performance - baseline performance) / baseline performance * 100
[2025-02-25 09:53:05 UTC] [KAT] [message] - Case package locate: /opt/template/template_Postgresql_20250119_073054/train-20250225-112832
[2025-02-25 09:53:05 UTC] [KAT] [message] - ================================================================================
  • Performing the performance test again for a specified round

    Type test r1 and press Enter.

    Before assigning values to parameters, you can retest the rounds with good performance improvement for multiple times to check whether the tuned parameters are stable. A smaller value of CV(%) indicates higher performance data consistency and more stable parameters in a round. That means you can consider using the parameters of that round to the actual application.

    In a normal test scenario, if the value of CV(%) is less than 1%, the parameter tuning effect is considered stable. If the value is within the range of 1% to 5%, the effect is less stable. If the value is within the range of 5% to 10%, the effect is unstable and you need to determine whether to accept it. If the value exceeds 10%, the tuning effect of the current round is unstable and you are advised to solve the stability problem (caused by environmental or human factors).

    • When using the test subcommand, you can specify multiple rounds of data and use the parameters of these rounds to perform the performance test again. Separate the rounds using commas (,). For example:

      test 1,2,r1,r2: Use the first and second rounds of tuning and the two rounds whose performance ranks first and second to perform the performance test again.

    • You can append top or n-m to the test command to specify multiple rounds. For example:

      test 1-3,top3: The first, second, and third rounds of automatic tuning and the three rounds with the best performance data

    • You can append the specific rounds to the test command and then a space and a number after the specific rounds to specify the number of repeated performance tests for these rounds. For example:

      test 1,2 3: Repeat the first and second rounds of automatic tuning for three times.

    • Generally, A/B testing is used to ensure stable test results. For example, to retest the result of the round (assume it is round A) whose performance ranks first, you can run test 0,A 5 to alternately test five groups of the baseline round and highest-performing for 10 times.

    Command output:

     1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
    [2025-02-25 09:53:48 UTC] [KAT] [message] - ================================================================================
[2025-02-25 09:53:48 UTC] [KAT] [message] -           kunpeng automatic tuning - task set up
[2025-02-25 09:53:48 UTC] [KAT] [message] -           start time: 2025-02-25 09:53:48
[2025-02-25 09:53:48 UTC] [KAT] [message] - --------------------------------------------------------------------------------
[2025-02-25 09:53:51 UTC] [KAT] [message] - --------- kunpeng automatic tuning - Step   <params prepare>   start -----------
[2025-02-25 09:53:51 UTC] [KAT] [message] - ================================================================================
[2025-02-25 09:53:51 UTC] [KAT] [message] -           kunpeng automatic tuning - test 1
[2025-02-25 09:53:51 UTC] [KAT] [message] - --------- kunpeng automatic tuning - Step    <round set up>    start -----------
[2025-02-25 09:53:51 UTC] [KAT] [message] - --------- kunpeng automatic tuning - Step   <assign params>    start -----------
[2025-02-25 09:53:53 UTC] [KAT] [message] - --------- kunpeng automatic tuning - Step      <run test>      start -----------
[2025-02-25 09:54:01 UTC] [KAT] [message] - --------- Run test successfully.
[2025-02-25 09:54:01 UTC] [KAT] [message] - --------- Performance     :       99.85
[2025-02-25 09:54:01 UTC] [KAT] [message] - --------- kunpeng automatic tuning - Step  <round tear down>   start -----------
[2025-02-25 09:54:02 UTC] [KAT] [message] - ================================ Detail Report =================================
[2025-02-25 09:54:02 UTC] [KAT] [message] - Actual test times : 1
[2025-02-25 09:54:02 UTC] [KAT] [message] - Fail times        : 0
[2025-02-25 09:54:02 UTC] [KAT] [message] - --------------------------------------------------------------------------------
[2025-02-25 09:54:02 UTC] [KAT] [message] - |  No.  |  Round  | Status | Performance | Performance(before) |
[2025-02-25 09:54:02 UTC] [KAT] [message] - --------------------------------------------------------------------------------
[2025-02-25 09:54:02 UTC] [KAT] [message] - |   1   |    1    |   S    |    99.85    |        91.07        |
[2025-02-25 09:54:02 UTC] [KAT] [message] - --------------------------------------------------------------------------------
[2025-02-25 09:54:02 UTC] [KAT] [message] - ================================  Total Report =================================
[2025-02-25 09:54:02 UTC] [KAT] [message] - Tested rounds     : 1
[2025-02-25 09:54:02 UTC] [KAT] [message] - Tuning Direction : high
[2025-02-25 09:54:02 UTC] [KAT] [message] - Performance description: tpmC, transactions per minute
[2025-02-25 09:54:02 UTC] [KAT] [message] - Baseline performance  : 97.03
[2025-02-25 09:54:02 UTC] [KAT] [message] - --------------------------------------------------------------------------------
[2025-02-25 09:54:02 UTC] [KAT] [message] - |  Round  | Times | Performance | Performance(before) | Change (%) |  CV(%) |
[2025-02-25 09:54:02 UTC] [KAT] [message] - --------------------------------------------------------------------------------
[2025-02-25 09:54:02 UTC] [KAT] [message] - |    1    |   1   |    99.85    |        91.07        |   +9.64    |  0.00  |
[2025-02-25 09:54:02 UTC] [KAT] [message] - --------------------------------------------------------------------------------
[2025-02-25 09:54:02 UTC] [KAT] [message] - ================================================================================
  • Changing the output log level

    Type log 1 and press Enter. The output log level is changed to INFO, which is valid only for the current login on the interactive user interface.

  • Viewing the importance of all parameters for tuning

    Type iglobal and press Enter.

    A higher value in the importance(%) column indicates the parameter in that row is more strongly associated with the tuning result (greater impact on performance). You can pay more attention to that row.

    Command output:

     1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
       Param Group     |     Parameter      | importance(%) |       range        |            Top 5
-------------------------------------------------------------------------------------------------------------
rocksdb             |max_background_flush|12.79%         |[1, 64]             |['31', '18', '7', '18', '35']
                    |es                  |               |                    |
-------------------------------------------------------------------------------------------------------------
rocksdb             |target_file_size_bas|10.79%         |[2, 268435456]      |['3144477', '18231017', '58331
                    |e                   |               |                    |577', '18231017', '123520730']
-------------------------------------------------------------------------------------------------------------
rocksdb             |subcompactions      |10.54%         |[1, 32]             |['14', '22', '26', '26', '19']
-------------------------------------------------------------------------------------------------------------
...
...
...
-------------------------------------------------------------------------------------------------------------
rocksdb             |pin_l0_filter_and_in|0.31%          |('true', 'false')   |['false', 'false', 'false', 'f
                    |dex_blocks_in_cache |               |                    |alse', 'false']
-------------------------------------------------------------------------------------------------------------
  • Viewing the importance of parameters in a round of tuning

    Type iround 1 and press Enter.

    Baseline performance: baseline performance data obtained from the tests.

    Round N performance: performance data obtained in the nth round of test.

    Expected performance: performance data calculated by the model based on the training history when the gains of all parameters are zero.

    Round N estimated performance: estimated performance data of the nth round, which is calculated by the model based on the training history.

    Estimated total benefit: difference between the expected performance and estimated performance of the nth round, which indicates the total benefit obtained by all parameters.

    Command output:

     1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
    Baseline performance: 19516955.00
Round 1 performance: 19878245.0
Expected performance: 19747731.20
Round 1 estimated performance: 19836562.31
Estimated total benefit: 88831.11

    Param Group     |     Parameter      |  Param Value  | Estimated Benefit
------------------------------------------------------------------------------
rocksdb             |max_background_flush|7              |19063.97
                    |es                  |               |
------------------------------------------------------------------------------
rocksdb             |subcompactions      |26             |16060.71
------------------------------------------------------------------------------
rocksdb             |target_file_size_bas|18231017       |15296.90
                    |e                   |               |
------------------------------------------------------------------------------
...
...
...
------------------------------------------------------------------------------
rocksdb             |pin_l0_filter_and_in|false          |371.09
                    |dex_blocks_in_cache |               |
------------------------------------------------------------------------------
rocksdb             |delayed_write_rate  |33446408       |112.15
------------------------------------------------------------------------------
  • Using the optimal parameters inferred from parameter importance analysis to perform tests

    Type ibesttest and press Enter.

    Command output:

     1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
    Baseline performance: 24592683.0
Best parameters inferenced:
  Param Group  |          Parameter           |    Param Value
-------------------------------------------------------------------
rocksdb        |batch_size                    |109
-------------------------------------------------------------------
rocksdb        |block_size                    |142561
-------------------------------------------------------------------
rocksdb        |bloom_bits                    |77
-------------------------------------------------------------------
...
...
...
-------------------------------------------------------------------
rocksdb        |target_file_size_base         |34462869
-------------------------------------------------------------------
rocksdb        |write_buffer_size             |21101528
-------------------------------------------------------------------
[2025-02-18 07:24:41 UTC] [KAT] [message] - --------- kunpeng automatic tuning - Step    <round set up>    start -----------------
[2025-02-18 07:24:41 UTC] [KAT] [message] - --------- kunpeng automatic tuning - Step   <assign params>    start -----------------
[2025-02-18 07:24:42 UTC] [KAT] [message] - --------- kunpeng automatic tuning - Step      <run test>      start -----------------
[2025-02-18 07:24:42 UTC] [KAT] [message] -  ### Run test successfully. Performance of this round is [28985507] ###
[2025-02-18 07:24:42 UTC] [KAT] [message] - --------- kunpeng automatic tuning - Step  <round tear down>   start -----------------
  • Resuming the tuning process

    Type train -r 2 and press Enter to perform the next two rounds of automatic tuning on the use screen.

    The final report is displayed together with the previous automatic tuning data.

    Command output:

     1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
    [2025-02-25 11:12:21 UTC] [KAT] [message] - Algorithm               : HPO
[2025-02-25 11:12:21 UTC] [KAT] [message] - ================================================================================
[2025-02-25 11:12:21 UTC] [KAT] [message] -           kunpeng automatic tuning - round 6
[2025-02-25 11:12:21 UTC] [KAT] [message] -           start time: 2025-02-25 11:12:21
[2025-02-25 11:12:21 UTC] [KAT] [message] - --------------------------------------------------------------------------------
[2025-02-25 11:12:21 UTC] [KAT] [message] - Due to the presence of historical data and the random iterations being set to 0, the inference tuning phase begins.
[2025-02-25 11:12:33 UTC] [KAT] [message] - --------- kunpeng automatic tuning - Step    <round set up>    start -----------
[2025-02-25 11:12:33 UTC] [KAT] [message] - --------- kunpeng automatic tuning - Step   <assign params>    start -----------
[2025-02-25 11:12:35 UTC] [KAT] [message] - --------- kunpeng automatic tuning - Step      <run test>      start -----------
[2025-02-25 11:12:42 UTC] [KAT] [message] - --------- Run test successfully.
[2025-02-25 11:12:42 UTC] [KAT] [message] - --------- Performance     :       83.74
[2025-02-25 11:12:42 UTC] [KAT] [message] - --------- kunpeng automatic tuning - Step  <round tear down>   start -----------
[2025-02-25 11:12:42 UTC] [KAT] [message] - ================================================================================
[2025-02-25 11:12:42 UTC] [KAT] [message] -           kunpeng automatic tuning - round 7
[2025-02-25 11:12:42 UTC] [KAT] [message] -           start time: 2025-02-25 11:12:42
[2025-02-25 11:12:42 UTC] [KAT] [message] - --------------------------------------------------------------------------------
[2025-02-25 11:12:42 UTC] [KAT] [message] - --------- kunpeng automatic tuning - Step    <round set up>    start -----------
[2025-02-25 11:12:42 UTC] [KAT] [message] - --------- kunpeng automatic tuning - Step   <assign params>    start -----------
[2025-02-25 11:12:44 UTC] [KAT] [message] - --------- kunpeng automatic tuning - Step      <run test>      start -----------
[2025-02-25 11:12:52 UTC] [KAT] [message] - --------- Run test successfully.
[2025-02-25 11:12:52 UTC] [KAT] [message] - --------- Performance     :       81.41
[2025-02-25 11:12:52 UTC] [KAT] [message] - --------- kunpeng automatic tuning - Step  <round tear down>   start -----------
[2025-02-25 11:12:52 UTC] [KAT] [message] - ============================== Auto Tuning Report ==============================
[2025-02-25 11:12:52 UTC] [KAT] [message] - Total round             : 7
[2025-02-25 11:12:52 UTC] [KAT] [message] - Total run               : 12
[2025-02-25 11:12:52 UTC] [KAT] [message] - Fail times              : 0
[2025-02-25 11:12:52 UTC] [KAT] [message] - Application             : Postgresql
[2025-02-25 11:12:52 UTC] [KAT] [message] - Application version     : 11.3
[2025-02-25 11:12:52 UTC] [KAT] [message] - Test tool               : benchmarksql
[2025-02-25 11:12:52 UTC] [KAT] [message] - Tuning direction        : high
[2025-02-25 11:12:52 UTC] [KAT] [message] - Algorithm               : HPO
[2025-02-25 11:12:52 UTC] [KAT] [message] - Random initial ratio    : 0.5
[2025-02-25 11:12:52 UTC] [KAT] [message] - Performance description : tpmC, transactions per minute
[2025-02-25 11:12:52 UTC] [KAT] [message] - Parameter groups        : Postgresql
[2025-02-25 11:12:52 UTC] [KAT] [message] - Baseline performance    : 97.03
[2025-02-25 11:12:52 UTC] [KAT] [message] - Top 10 performance :
[2025-02-25 11:12:52 UTC] [KAT] [message] - --------------------------------------------------------------------------------
[2025-02-25 11:12:52 UTC] [KAT] [message] - |  Rank  |     Round     |  Performance   |  Improvement  (%) |
[2025-02-25 11:12:52 UTC] [KAT] [message] - --------------------------------------------------------------------------------
[2025-02-25 11:12:52 UTC] [KAT] [message] - |   1    |   round  4    |     86.92      |      -10.42       |
[2025-02-25 11:12:52 UTC] [KAT] [message] - |   2    |   round  6    |     83.74      |      -13.70       |
[2025-02-25 11:12:52 UTC] [KAT] [message] - |   3    |   round  7    |     81.41      |      -16.10       |
[2025-02-25 11:12:52 UTC] [KAT] [message] - |   4    |   round  1    |     80.32      |      -17.22       |
[2025-02-25 11:12:52 UTC] [KAT] [message] - |   5    |   round  3    |     79.44      |      -18.13       |
[2025-02-25 11:12:52 UTC] [KAT] [message] - |   6    |   round  5    |     70.05      |      -27.81       |
[2025-02-25 11:12:52 UTC] [KAT] [message] - |   7    |   round  2    |     66.37      |      -31.60       |
[2025-02-25 11:12:52 UTC] [KAT] [message] - --------------------------------------------------------------------------------
[2025-02-25 11:12:52 UTC] [KAT] [message] - Note:
[2025-02-25 11:12:52 UTC] [KAT] [message] - The performance value is the return value of the run test step,
[2025-02-25 11:12:52 UTC] [KAT] [message] - Performance improvement (%) =
[2025-02-25 11:12:52 UTC] [KAT] [message] -     (round performance - baseline performance) / baseline performance * 100
[2025-02-25 11:12:52 UTC] [KAT] [message] - Case package locate: /opt/template/template_Postgresql_20250119_073054/train-20250225-112832
[2025-02-25 11:12:52 UTC] [KAT] [message] - ================================================================================
  • If you accept the performance tuning data of the round whose performance ranks first, you can apply the parameters of that round to the application and system.

    Type set r1 and press Enter.

    The following information is returned if the parameters are set successfully:

    1
    [2025-02-25 11:14:31 UTC] [KAT] [message] - set params of r1 success!
  • Exiting from the interactive user interface

    Type exit and press Enter.

Parent topic: Kunpeng AutoTuner