Rodin Performances: Difference between revisions
From Event-B
Jump to navigationJump to search
imported>Tommy mNo edit summary |
imported>Nicolas |
||
Line 11: | Line 11: | ||
OS: | OS: | ||
* Ubuntu 10.10 | * Ubuntu 10.10 | ||
* linux kernel 2.6.35- | * linux kernel 2.6.35-25 | ||
* GNOME 2.32.0 | * GNOME 2.32.0 | ||
Revision as of 12:34, 7 February 2011
Performance measures for the Rodin Platform.
Configurations
Configuration 1
System:
- Intel Pentium Dual Core E2180 (2* 2 GHz)
- 2 Go RAM
OS:
- Ubuntu 10.10
- linux kernel 2.6.35-25
- GNOME 2.32.0
Java:
- Java(TM) SE Runtime Environment (build 1.6.0_22-b04)
- Java HotSpot(TM) Server VM (build 17.1-b03, mixed mode)
- VM arguments
- -Xms40m
- -Xmx1200m
Eclipse:
- Helios (3.6.1)
- JUnit 4.8.1
- tests run from a freshly launched eclipse as only running user application, in UI thread, with an empty workbench.
Benchmark Projects
Cruise Control 0.5.0
- 7 Contexts
- 49 Machines
- 2911 POs
Cars on a Bridge
Taken from Jean-Raymond Abrial's book "Modeling in Event-B"
- 3 Contexts
- 4 Machines
- 237 POs
Systerel Project
- 9 Contexts
- 14 Machines
- 927 POs
Protocol
A measurement is launched using the JUnit plug-in fr.systerel.perf.tests.
Launch Configuration must include exactly the following plug-ins:
- fr.systerel.perf.tests
- org.eventb.core
- org.eventb.core.seqprover
- org.eventb.core.ast
- org.rodinp.core
Every measurement starts with the build of a fake project intended to load all needed classes (extra time of more than 10 seconds has been observed for build). Class loading is intended to be optimised separately.
Verify that 1200 Mo are free when tests are ready for launch.
Results
Test | Applied to | Configuration | Rodin 1.3 | Rodin 2.0 | Rodin 2.1 |
---|---|---|---|---|---|
Lexer | #Configuration 1 | 19.0 s | 21.4 s | 35.6 s | |
Parser | #Configuration 1 | 17.1 s | 39.3 s | 17.2 s | |
Builder / models only | #Cruise Control 0.5.0 | #Configuration 1 | 119.6 s | 211.0 s | 30.9 s |
Builder + Auto Prover / models only | #Cruise Control 0.5.0 | #Configuration 1 | 351.5 s (1550 / 2911 POs) | 641.9 s (1550 / 2911 POs) | 484.5 s (1574 / 2911 POs) |
Builder / models + proofs | #Cruise Control 0.5.0 | #Configuration 1 | 108.6 s | 199.1 s | 123.3 s |
Builder / models only | #Cars on a Bridge | #Configuration 1 | 2060 ms | 2524 ms | 995 ms |
Builder + Auto Prover / models only | #Cars on a Bridge | #Configuration 1 | 4388 ms (237 / 237 POs) | 4320 ms (237 / 237 POs) | 3466 ms (237 / 237 POs) |
Builder / models + proofs | #Cars on a Bridge | #Configuration 1 | 2843 ms | 3360 ms | 2238 ms |
Builder / models only | #Systerel Project | #Configuration 1 | 88.7 s | 11.2 s | |
Builder + Auto Prover / models only | #Systerel Project | #Configuration 1 | 185.7 s (1218 / 1360 POs) | 162.3 s (1218 / 1360 POs) | |
Builder / models + proofs | #Systerel Project | #Configuration 1 | 149.9 s | 109.4 s |
The figure below presents and sums up the values of the table above:
Test Builder = SC + POG + POM, no proof attempt is performed.