Difference between revisions of "Version benchmarking"

From Heureka Wiki
Jump to navigation Jump to search
 
(6 intermediate revisions by one other user not shown)
Line 1: Line 1:
 
[[Category:Release]]
 
[[Category:Release]]
  
Here we present a comparison between different versions for a benchmark problem run in PlanWise. We call this problem Hexagons1000, it can be downloaded from [https://www.heurekaslu.se/filerepository/PlanwiseBenchmark/ here]. The problem consists of one thousand simulated treatment units. The results refer to an optimization with simple even-flow constraints. The TPG-simulation includes nine forest domains and the somewhat time-consuming breeding model is activated. For each treatment unit managed with even-aged management a nature conservation retention patch is created resulting in a total of ? units. The forest is a simulated but realistic landscape with respect to age classes, site fertilities and tree species distribution.  
+
Here we present a comparison between different versions for a benchmark problem run in PlanWise. We call this problem Hexagons1000, it can be downloaded from [https://www.heurekaslu.se/filerepository/PlanwiseBenchmark/ here]. The problem consists of one thousand simulated treatment units. The results refer to an optimization with simple even-flow constraints. The TPG-simulation includes nine forest domains and the somewhat time-consuming breeding model is activated. For each treatment unit managed with even-aged management a nature conservation retention patch is created. The forest is a synthetic but is a realistic landscape with respect to age classes, site fertilities and tree species distribution.  
  
The comparison is intended to present to users a simple form of quality and status check and give some insigth into what can be expected. Net present value is an excellent indicator for whether there has been any model changes, since it basically depends on everything calculated by the growth-and-yield simulator.  
+
The comparison is intended to present to users a simple form of quality and status check and give some insight into what simulation times that can be expected. Please note that if more result variables are added (such as dead wood, carbon and nitrogen etc.) simulation times might be quite significantly longer.
 +
Net present value is an good indicator for whether there has been any large model changes, since it basically depends on everything calculated by the growth-and-yield simulator.  
  
 
This page was/is first published with version 2.11. We intend to run the benchmark problem for each major version released.
 
This page was/is first published with version 2.11. We intend to run the benchmark problem for each major version released.
Line 9: Line 10:
 
{| {{table}} border = 1 style="text-align: center; width: 100%; border-collapse: collapse; "
 
{| {{table}} border = 1 style="text-align: center; width: 100%; border-collapse: collapse; "
 
| style="background:#f0f0f0;width:8%;"|'''Version'''
 
| style="background:#f0f0f0;width:8%;"|'''Version'''
| style="background:#f0f0f0;width:5%;"|'''TPG-time, 1 CPU-core (s)'''
+
| style="background:#f0f0f0;width:5%;"|'''TPG-time, 1 CPU-thread(s)'''
| style="background:#f0f0f0;width:5%;"|'''TPG-time, 8 CPU-cores (s)'''
+
| style="background:#f0f0f0;width:5%;"|'''TPG-time, 16 CPU-threads(s)'''
 
| style="background:#f0f0f0;width:5%;;"|'''Total NPV'''
 
| style="background:#f0f0f0;width:5%;;"|'''Total NPV'''
 
| style="background:#f0f0f0;width:5%;;"|'''NPV/ha within managed area'''
 
| style="background:#f0f0f0;width:5%;;"|'''NPV/ha within managed area'''
Line 23: Line 24:
 
| style="background:#f0f0f0;width:10%;;"|'''Release notes'''
 
| style="background:#f0f0f0;width:10%;;"|'''Release notes'''
 
|-style = "vertical-align: top; text-align: center;"
 
|-style = "vertical-align: top; text-align: center;"
|[[Version 2.16.0|2.16.0]]||    ||1336<sup>d</sup> ||20 307 509 ||37 883 ||2 528 || 4 574 ||6.76|| 799 || 266||14.7||6.5||||[[Version 2.16.0]]
+
|[[Version 2.19|2.19]]||||710<sup>d</sup> ||21 224 113 ||39 670 ||2 661 || 4 990 ||7.08||1 000 || 282||16.7||7.0||||[[Version 2.19]]
 +
|-style = "vertical-align: top; text-align: center;"
 +
|[[Version 2.18.3|2.18.3]]||    ||2153<sup>d</sup> ||21 192 183 ||39 608 ||2 669 || 4 959 ||7.11||1 002 || 293||16.9||7.0||||[[Version 2.18.3]]
 +
|-style = "vertical-align: top; text-align: center;"
 +
|[[Version 2.18.2|2.18.2]]||      ||2067<sup>d</sup> ||21 192 183 ||39 608 ||2 669 || 4 959 ||7.11||1 002 || 293||16.9||7.0||||[[Version 2.18.2]]
 +
|-style = "vertical-align: top; text-align: center;"
 +
|[[Version 2.18|2.18]]||    ||1425<sup>d</sup> ||21 298 040 ||39 814 ||2 659 || 4 932 ||7.12||1 003 || 293||16.9||7.0||||[[Version 2.18]]
 +
|-style = "vertical-align: top; text-align: center;"
 +
|[[Version 2.17|2.17]]||    ||1297<sup>d</sup> ||20 308 979 ||37 886 ||2 542 || 4 574 ||6.80|| 799 || 266||14.7||6.5||||
 +
[[Version 2.17]]
 +
|-style = "vertical-align: top; text-align: center;"
 +
|[[Version 2.16|2.16]]||    ||1336<sup>d</sup> ||20 307 509 ||37 883 ||2 528 || 4 574 ||6.76|| 799 || 266||14.7||6.5||||[[Version 2.16]]
 
|-style = "vertical-align: top; text-align: center;"
 
|-style = "vertical-align: top; text-align: center;"
 
|[[Version 2.16|2.16 RC]]||                                ||1228<sup>a</sup>||20 114 643||37 515||2 612||2 607||6.92||765||302||13.7||6.31||||[[Version 2.16]]
 
|[[Version 2.16|2.16 RC]]||                                ||1228<sup>a</sup>||20 114 643||37 515||2 612||2 607||6.92||765||302||13.7||6.31||||[[Version 2.16]]
Line 55: Line 67:
 
|}
 
|}
  
<sup>a</sup>Windows 7, Processor Intel I7-4770, 8-core, 32GB RAM. Samsung SSD 840 EVO. SQL Server 2012 Prof. <br>
+
<sup>a</sup>Windows 7, Processor Intel I7-4770, 4-core, 32GB RAM. Samsung SSD 840 EVO. SQL Server 2012 Prof. <br>
 
<!--<sup>b</sup>With Thinning Period Delay Max = 1 and 0 in generation 1 and 2, respectively.-->
 
<!--<sup>b</sup>With Thinning Period Delay Max = 1 and 0 in generation 1 and 2, respectively.-->
 
<!--<sup>b</sup>With Thinning Period Delay Max = 2 and 1 in generation 1 and 2, respectively.-->
 
<!--<sup>b</sup>With Thinning Period Delay Max = 2 and 1 in generation 1 and 2, respectively.-->
 
<sup>b</sup>With Thinning Period Delay Max = 2 and 0 in generation 1 and 2, respectively. <br>
 
<sup>b</sup>With Thinning Period Delay Max = 2 and 0 in generation 1 and 2, respectively. <br>
 
<sup>c</sup>With understorey cleaning inactivated. <br>
 
<sup>c</sup>With understorey cleaning inactivated. <br>
 +
<sup>d</sup>Windows 10, AMD Ryzen 7 4750U, 8-core, 16GB RAM. <br>

Latest revision as of 15:10, 15 September 2022


Here we present a comparison between different versions for a benchmark problem run in PlanWise. We call this problem Hexagons1000, it can be downloaded from here. The problem consists of one thousand simulated treatment units. The results refer to an optimization with simple even-flow constraints. The TPG-simulation includes nine forest domains and the somewhat time-consuming breeding model is activated. For each treatment unit managed with even-aged management a nature conservation retention patch is created. The forest is a synthetic but is a realistic landscape with respect to age classes, site fertilities and tree species distribution.

The comparison is intended to present to users a simple form of quality and status check and give some insight into what simulation times that can be expected. Please note that if more result variables are added (such as dead wood, carbon and nitrogen etc.) simulation times might be quite significantly longer. Net present value is an good indicator for whether there has been any large model changes, since it basically depends on everything calculated by the growth-and-yield simulator.

This page was/is first published with version 2.11. We intend to run the benchmark problem for each major version released.

Version TPG-time, 1 CPU-thread(s) TPG-time, 16 CPU-threads(s) Total NPV NPV/ha within managed area Final felling volume per year, all periods (m3sk/ha,yr) Final felling volume period 1 (m3sk/ha,yr) Final felling area per year, all periods (m3sk/ha,yr) Thinning volume per year, all periods (m3sk/ha,yr) Thinning volume per year, period 1 (m3sk/ha,yr) Thinning area per year, all periods (m3sk/ha,yr) Total production (m3sk/ha,yr) Comment Release notes
2.19 710d 21 224 113 39 670 2 661 4 990 7.08 1 000 282 16.7 7.0 Version 2.19
2.18.3 2153d 21 192 183 39 608 2 669 4 959 7.11 1 002 293 16.9 7.0 Version 2.18.3
2.18.2 2067d 21 192 183 39 608 2 669 4 959 7.11 1 002 293 16.9 7.0 Version 2.18.2
2.18 1425d 21 298 040 39 814 2 659 4 932 7.12 1 003 293 16.9 7.0 Version 2.18
2.17 1297d 20 308 979 37 886 2 542 4 574 6.80 799 266 14.7 6.5

Version 2.17

2.16 1336d 20 307 509 37 883 2 528 4 574 6.76 799 266 14.7 6.5 Version 2.16
2.16 RC 1228a 20 114 643 37 515 2 612 2 607 6.92 765 302 13.7 6.31 Version 2.16
2.15.4 2961
(49m 21s)a
1247a 20 114 643 37 515 2 612 2 607 6.92 765 302 13.7 6.31 Version 2.15
2.14 2861
(48m 41s)a
1132a 20 060 196 37 412 2 685 2 633 7.33 748 340 13.7 6.25 Version 2.14
2.13 2894
(48m 14s)a
1038a 19 997 049 37 365 2 685 2 633 7.33 748 340 13.7 6.25 Version 2.13
2.12.3 1010a 20 081 186 37 527 2 681 2 622 7.35 748 331 13.6 6.23 Increased npv: New bucking algorithm, bug fixed Version 2.12
2.12.3c 983a 19 915 687 37 207 2 713 2 637 7.36 718 376 13.3 6.25 Understorey cleaning inactivated Version 2.12
2.11.1b 2483s
(41m 23s)a
19 707 063 36 824 2 680 2 610 7.34 756 326 13.8 6.24 Version 2.11
2.10 2532s
(42m 12s)a
- 19 606 195 36 624 2 711 2 652 7.40 697 403 11.5 6.20 Version 2.10
2.8.0 2360s
(39m 20s)a
- 19 652 164 36 712 2 724 2 652 7.40 713 389 11.7 6.29 Version 2.8.0
2.6.0 2370s
(40m 30s)a
- 19 634 926 36 679 2 724 2 652 7.40 713 389 11.7 6.29 Version 2.6.0

aWindows 7, Processor Intel I7-4770, 4-core, 32GB RAM. Samsung SSD 840 EVO. SQL Server 2012 Prof.
bWith Thinning Period Delay Max = 2 and 0 in generation 1 and 2, respectively.
cWith understorey cleaning inactivated.
dWindows 10, AMD Ryzen 7 4750U, 8-core, 16GB RAM.