Performance

Eclipse vs. bake

Test Environement

Test-System	CPU: RAM: OS : HDD:	Intel Xeon W3520 (4x 2.67 GHz) 4 GB Windows XP Raid
Test-Application	~200 projects
Ruby-Version	ruby 1.9.2
Tests-Runs	30

Test Results

Test	Eclipse 3.6.1, CDT 6	bake
Build whole workspace	13:01 min	7:19 min
Rebuild without clean	3:05 min	0:50 min
Clean whole workspace	0:41 min	0:08 min
Clean the cleaned workspace	0:35 min	0:02 min

Note

For the build Eclipse needs additional 300 MB of RAM, in contrast bake takes up 35 MB of RAM at it’s peak.

CMake with Unix makefiles vs. bake

Test Environement

Test-System	CPU: RAM: OS : HDD:	Intel Xeon W3520 (4x 2.67 GHz) 4 GB Windows XP Raid
Test-Application	~100 projects
Ruby-Version	-
Tests-Runs	-

Test Results

• Build whole workspace: bake is ~10% faster

• Build again without clean: bake is ~50% faster

• Clean whole workspace: bake is ~80% faster

Performance using different rubies

Test Environement

Test-System	CPU: RAM: OS : HDD:	Intel Xeon W3520 (4x 2.67 GHz) 4 GB Windows XP Raid
Test-Application	~200 projects
Tests-Runs	30

Test Results

	ruby 1.8.6p398	ruby 1.8.7p352	ruby 1.9.2p180	ruby 1.9.3p0
Build whole workspace	13:58 min	8:07 min	7:19 min	7:28 min
Build single file	1,20 sec	1,87 sec	2,38 sec	1,29 sec

Note

• ruby 1.8.6 uses only ONE native thread for multiple ruby threads.

• ruby 1.9.2 for Windows needs very long to startup for complex applications, which is fixed in 1.9.3