You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 8 Next »

We run tests with different releases of LemonTree against different test-models to examine the improvement of the memory usage and performance.

This page gives an overview of the performance and memory usage and describes the test-models.

Performance and memory usage are continually improved in LemonTree.


Win10, 16GB RAM i7-6700HQ CPU @2.60 GHz, 64 BIT

Test description

We have different test setups which run against the recent releases of LemonTree.

1: Small model, with few changes in 3-way comparison

These models are the example models which come with the LemonTree installation.

2: Medium model with several changes in 3-way comparison

A larger model which is diffed in a 3-way comparison. There are changes in A (respective to Base) and B (respective to Base). This test-case can be regarded as standard use-case for LemonTree, when used with models that are under version control (GIT, SVN).

3: Large model with a lot of changes in 2-way comparison

A really large model with a lot changes and even more elements to be checked by LemonTree. This is quite challenging for LemonTree. In a 2-way comparison Model A is also regarded as Base model.

4: EAExample, compared with empty model

When comparing an empty model with another model the diff results in a lot of new objects which leads to special challenges for LemonTree. In older versions of LemonTree (< 1.5) this was not successful within the test environment.

5: Tagged Value model with a lot of TaggedValues in 2-way comparison

TaggedValues needed a lot of RAM in older versions of LemonTree (<1.8). 


Test Name


File size

Total packages

Total diagrams

Total elements

Total connectors

Total TaggedValues*


Small ModelBase3 MB617202152242

A3 MB617205154242

B3 MB617203153242

Medium ModelBase128 MB7582.03821.92226.07588.326

A128 MB7391.98821.68525.59488.810

B133 MB7391.98721.67025.56888.331

Large ModelBase233 MB8.4906.733114.96269.65232.788same as A

A233 MB8.4906.733114.96269.65232.788same as Base

B195 MB8.0596.252107.11164.34529.157

EaExampleBase1,4 MB10000empty model

A1,4 MB10000empty model

B20 MB4275644.6562.62329.460

TV Model
Base142 MB33890946.38022.819316.458same as A

A142 MB33890946.38022.819316.458same as Base

B152 MB34798250.81027.289336.759

* Total Tagged Values are calculated via SQLs in EA:

select count(*) from t_attributetag UNION
select count(*) from t_connectortag UNION
select count(*) from t_operationtag UNION
select count(*) from t_taggedvalue UNION
select count(*) from t_objectproperties

Test results

Test name

LT Version

1) Loading models

2) Diff & Merge calculation

3) Writing merge file

Max RAM usage

Small Model1.5.5.022s0.5s3s250 MB

1.7.220s1s2s284 MB


257 MB

2.1.413s1s1s263 MB

2.2.018s1s1s292 MB

2.2.220s1s1s297 MB

LT Version1) Loading models2) Diff & Merge calculation3) Writing merge fileMax RAM usage
Medium model1.5.5.02m15s4m42s6.500 MB

1.7.22m18s40s40s7.200 MB

2.0.02m16s20s1m5.454 MB

2.1.41m57s32s12s5.595 MB

2.2.02m23s13s20s6.184 MB

2.2.22m17s13s14s5.928 MB

1) Loading models2) Diff & Merge calculation3) Writing merge fileMax RAM usage
Large model1.5.5.02m30s37m68m9.100 MB

1.7.22m45s25m37m10.100 MB

2.0.02m30s15m49m9.525 MB

2.1.42m22s17m58s22m38s10.011 MB

2.2.02m09s7m55s21m53s10.486 MB

2.2.23m13s18m13s29m14s110.548 MB

1) Loading models2) Diff & Merge calculation3) Writing merge fileMax RAM usage
EAExample1.5.5.025s25s15m40s770 MB

1.7.226s60s6m980 MB

2.0.023s10s7m13s764 MB

2.1.413s32s4m44s789 MB

2.2.019s14s4m53s805 MB

2.2.222s16s4m57s841 MB

1) Loading models2) Diff & Merge calculation3) Writing merge fileMax RAM usage
TV Model
1.5.52m1m10s13m10s5.500 MB

1.7.23m10s11m11s6m10s6.708 MB

2.0.01m52s1m5m54s4.219 MB

2.1.41m19s59s3m54s4.359 MB

2.2.01m46s31s3m57s4.541 MB

2.2.21m47s33s4m10s4.073 MB

1: The test environment got a disc encryption since the last performance tests. We assume that's why the large models take longer on writing/loading. Older versions of LemonTree now also need longer with that models on writing.

  • No labels