LINK TO THE LATEST GRAPH

Right, I'm planning on making a fairly in depth presentation of how CC performs on various systems. For this, I need practical data, hopefully from as many people as possible. Obviously, I can't do that on my own (I'm not many people), so I'm requesting that you put your Computer through a quick test. More tests may come at a later date (stress testing etc) but for now it will just be a medium load test to get a basic perception of how the stats work out.

THE TEST:

Prerequisite settings:

-640x480 resolution
-NxWindowed set to 1
-Windowed Mode
-DeltaTime set to the vanilla 0.016667 (60 SFPS)
-Timescaling on (Press ctrl+p, if it doesnt say "ONE sim update per frame" then press ctrl+o)
-preferably with no extra mods
Procedure:

Close all applications.
Download this:
Attachment:
Tester.rte.zip [7.04 KiB]
Downloaded 268 times


and install it
Run the test scene from the campaign menu, and wait for it to stop doing things.
Close CC, and then navigate to /Cortex Command/Tester.rte/ and open Results.txt
Fill in OS, RAM, and PROCESSOR fields.
Give me the results.

How to post your results:
upload to pastebin.com or other offsite host. please do not directly post results.
Results should look like this

If you have any questions, feel free to ask. Any input on how I should process this data is also welcome, Currently I'm thinking of 4 3d graphs, 1 for each operating system, plotting processor and RAM against average FPS drop. If anyone knows any good, free stats graphing programs that can do 3d graphs that would be super. I hate trials.

Would like to note that for windows at least, DXDiag is a useful tool for grabbing your system infos if you don't know them offhand.

Cheers Duh. I get mine from the readout at the bottom of my computer, most of the important stuff is there anyway

For a heavy strain test I'd suggest getting people to drop a nuke, perhaps several.

As I said in the other thread, that adds randomness. And for testing, we want to cut out as much of that as possible. Although it would work well until someone made a test scene.

For keeping things always onscreen, you could use the concrete box map. A standardized "heavy load bomb" that spit out x particles, x glows of x,y size, and x mosrotatings, with accompanying light and medium versions, to use with or without AI, would help too.

Eh, I might script prom grounds to have a simple battle go on or something, standard vanilla death style stuff.
For now though, these stats should be fairly important. We can already see from Archaos' stats that the i7 is blowing my barely comparable laptop and Duh's nearly identical box, both using (albeit different model) core 2 duo CPUs out of the water. If we can get more people in on this to actually give a concrete spread of stats (I'll do my main box this week at some point, I haven't been using it due to the W7 RC expiring <_<) we should be able to recommend "buy this to run CC like a boss" to anyone complaining about lag, as well as pointing them to HTMCCLL

So, for anyone who says that the AMD is at a disadvantage, I haven't experienced it. This is most likely due to the fact that I have 6 gigs of RAM - from what I can tell, a lot of CC's lag and crashes are due to memory allocation.

I and for the heavy strain, use 3 fuel air bombs and drop then ontop of you base in tutorial.(that would really effect the framerate.)

I'm happy Cortex Command doesn't use my video card.
I have a really bad one, a Nvidia 6200 GeForce...

roflcopterz, I'm hesitant to accept those results just because they arent using the standardised 640x480 resolution.. If someone else with a c2quad gets different results I might count it as an outlier, just a warning.

Hey guys.

Working on a testbed scene to spit out HIGHLY SCIENTIFIC results.

Need some ideas for tests.

I think five repeats across maybe 3-5 different test types would be good, yeah?

So, say, test 1 is particle spawning.
5 count battery of spawning 100 particles.
5 count battery of spawning 1000 particles.
5 count battery of spawning 10000 particles.
Averages results of each battery, spits them out into a file.

Test two: terrain displacement. Not sure exactly how to test this; ideas accepted.

Test three: actor related, not sure exactly how. Maybe just a MOID stress test?

Further tests if anyone has better ideas.

I've already got a basic testing framework.

I'll also probably do some testing of different lua functions to determine overhead for using them.

re-do




for terrain you could spawn a few giant MOSR's and displace them a pixel then have them dissapear so they eat away the terrain
for actor maybe spawn 1 actor every 1 second (all the actors would be getshitbymo = 0) until you see some death counts going up to show moid limit

Terrain displacement: have 2 modes, one that uses 200-500 invisiblle particle digger bombs to burn out large chunks, and one that just does one giant MOSR deepcheck.
MOID stress would be good, as my laptop starts complaining once I get about 150 MOIDs of zombies on screen, whereas my main box works fairly happily above 200 (scripts witholding)
A glowsplosion test could also be necessary, using a mixture of glow and non-glow MOSPs.

Jesus this is going to take so much graphing. Still waiting on a program for it, unless excel does it and I'm missing something..

faux edit: cheers roflcopterz