3 Commando Brigade

Volunteers Needed for Important Server Benchmark Tests

I'm doing some extensive tests on our public server to identify what causes the most impact on server FPS - high AI counts (hardware stress) vs high player counts (network stress).

The former would conclude that under-performing server hardware is the issue, the latter would conclude that network configuration is the issue.

I've already done some tests with just 1 player (myself) and had some interesting results. I now require more players for further tests.

9 players are required for testing. Patience will be needed and tests could take up to an hour. Bear in mind that participation could result in greater server performance, benefiting everyone.

If you're interested please state your name below and we can sort out a time that would be best for everyone.


Requirements
None. Mods will not be run for the sake of consistency and removing margin of error.

Testers
1. Goodson

2. Bloopers

3. Serjames

4. Benzy

5. Shango

6. Apollo

7. Mikk

8. Ivano

9. J2Ninja

10. Depro


Control Variables
Server: Public (No other servers running at time of tests. Each test started from a fresh server restart.)
Console version: 1.10
BattleEye: Off
Mods: None (Bikeyed)
NoLogs: On
Mission: Blank mission on Stratis.

Independent Variables
- AI count
- Player count

Above settings are true for all tests.

yep sure

No mods needed. I'll update my post.

Happy to help.

But how will you know people aren't running mods? (#bikey). I have first hand experience of asking people to run no mods for testing (or only a certain set of mods), them agreeing, and eventually finding that they were in fact running additional mods "because I didn't think it would matter". Tests need to be carefully controlled if they are to give meaningful results. <scientist hat off>

I will be bikeying the server. Other control variables:

Control Variables
Server: Public (No other servers running at time of tests. Each test started from a fresh server restart.)
Console version: 1.10
BattleEye: Off
Mods: None (Bikeyed)
NoLogs: On
Mission: Blank mission on Stratis.

Independent Variables
- AI count
- Player count

Above settings are true for all tests.

Good stuff Prof.Goodson.

Using in-game admin commands (#monitor). This will give us all the information we need.

Update: This no longer needs to go ahead. Thanks for putting your names down.

You can thank Mikk for helping us find what was causing our server to run at much lower FPS than expected. Turns out the server was running on 'balanced' power settings, effectively underclockig the CPU. It's not been set to "high performance" and running at full speed (3.8GHz).

After some simple tests, here'e the difference:

Before
Number of AI: 256
Server FPS: 65%

After
Number of AI: 256
Server FPS: 94%

The max FPS any server can reach is 50 FPS (100%), meaning our server is now running at near top speed all the time.


Thanks again Mikk!