Performance testing image - Yuuzhan Vong at War mod for Star Wars: Empire at War: Forces of Corruption
Behold the Yuuzhan Vong Army! This mod adds the Yuuzhan Vong Race as a playable faction and all units of this era. We are a community of modders working for the greater goal, so anyone who wants to help or support us is welcome!
There has been quite a discussion regarding our new concussion missile model.
So I did some performance testing.
Setup:
8x Vanilla Broadside Cruiser with an increased fire rate (10 times faster than vanilla, which makes a 30 times increased fire rate when it uses its "Barrage" ability)
10x Y-Wing Squadron
10x Y-Wing Advanced Squadron
10x X-Wing Squadron
3x KDY Republic Cruiser
6x B-Wing Squadron
3x TIE Defender Squadron
2x Victory 2 Frigate
The performance test starts 15 seconds into the battle, once the Broadside Cruiser has target locked with its "Barrage" ability and is about to start firing.
From that moment on the test runs for 50 seconds. No changes in camera position. All units on the battlefield are in view for a maximum of rendered assets.
The test has been repeated several times (10x each) alternating between the two projectile models.
The showcased graph represents the worst test run (with the smallest average fps difference).
The presented graph shows the frames per second (fps) comparison of the two models where the x-axis represents frames, and y-axis represents seconds.
The red graph represents the old vanilla model, whilst the blue one represents our new model.
Additional data:
2016-05-24 17:59:26 - swfoc (old missile)
Frames: 2221 - Time: 102219ms - Avg: 21.728 - Min: 14 - Max: 47
2016-05-24 18:03:19 - swfoc (new missile)
Frames: 2259 - Time: 98110ms - Avg: 23.025 - Min: 15 - Max: 43
Benchmark software: FRAPS
PC specs:
OS: Windows 10 Pro x64
RAM: 16 GB
CPU: i7-4790K @4.00GHz
GPU: GTX980ti
Conclusion:
The performed test runs suggest that our new model does not affect performance in a noticable negative way during normal gameplay, as the performance during this stress test is equal to the vanilla model on average, with a tendency to performing even slightly better than the vailla model.
Kad has been kind enough to write a description detailing how we go about testing new elements and ideas for the mod. We are trying to go about our production in a very structured manner, so that everyone may benefit from our findings.
We welcome comments and questions, and if we have missed something in our attempts to analyze effectiveness, please let us know what we might have overlooked.
Sigh if only alamo had been multithreaded and not 32 bit. Cause it's sad that it took this long for devs to finally start both.
That would be so awesome.
This engine can do a lot. Could even do way more as it is, if we had proper access to some core files.