Performance comparison between X3-R and X3-TC

Observe · Post by **Observe** » Mon, 15. Dec 08, 00:17

Introduction

Following is an investigation into game performance differences between X3-Reunion and X3-Terran Conflict. We all know X3-TC performs worse than X3-R, but I wanted to obtain quantifiable comparison data.

To do this, I created (as much as I could) an “apples to apples” test environment with similar startup conditions for both games. I chose Unknown Sector 18,10 for my testing because it is an isolated out-of-the-way sector with minimal (none) traffic getting in the way of measurements. I modified astronomical and other objects in 18,10 to suit my testing purposes.

Each comparison was performed following Windows re-boot and waiting several minutes for the computer to settle down before beginning – with minimal processes and no other applications running besides FRAPS. In-game measurements were taken after waiting 2 minutes for Jobs etc to stabilize.

Reference threads

[MOD IDEA] X3 Terran Conflict - improved performance

[MOD coming soon] X3-Reunion performance callback

Test Hardware and Software

First I installed Windows XP on re-formatted Hard Drive, updated Windows to Service Pack 3, and installed all the latest hardware drivers. Then I performed fresh install of X3 Reunion and X3 Terran Conflict. Neither game has any added scripts or mods beyond those described for the test sector, and simple game init script for startup with Buster in sector 18,10.

Computer

AMD Athlon single-core 64 Processor 3200+ (unclocked)
RAM 3.00 GB
Nvidia GeForce 9600GT 512 MB RAM

Operating System

Windows XP SP 3 (fresh install)

Measurement software

FRAPS version 2.8.0

X3 Reunion

Version 2.0.01 (fresh install, no mods or added scripts)

X3 Terran Conflict

Version 1.3 (fresh install, no mods or added scripts)

Test sector description

[ external image ]

X3-R sector xml header

<o t="1" x="18" y="10" r="14" size="22500000" m="0" p="-1"> (sector)

X3-TC sector xml header

<o id="396" t="1" x="18" y="10" r="14" size="22500000" m="0" p="-2" qtrade="100" qfight="100" qbuild="100" qthink="100"> (sector)

X3-R and X3-TC Sun and Planets xml

<o t="2" s="31" neb="0" stars="0"/> (background)
<o t="3" s="0" x="-30704518" y="-1488443930" z="615762029" color="8421504"/> (Sun)
<o t="4" s="23" x="0" y="0" z="-450000000" color="8421504" fn="0"/> (Planet)
<o f="1" t="4" s="30" x="28213826" y="282138263" z="282138263" color="8421504" fn="0"/> (Planet)

Other objects (X3-R and TC)

10 asteroids
1 Federal Argon Shipyard (SS_FAC_A_SHIP)
1 Solar Power Plant (SS_FAC_S_POWER2)
1 North Gate

Graphics Settings

X3-R and X3-TC

[ external image ]

Definitions

Anisotropic filtering

This is a very computationally demanding process of enhancing the image quality of textures when viewed in different directions at different angles.

Automatic Quality Control

The Automatic Quality Control (AQC) system is a built-in feature to try to prevent frame rates dropping too low and should be left switched on for all but the highest spec systems (I turn off)

Anti-aliasing

Smooth’s the jagged appearance of diagonal lines and curves

Results

X3-R

0. Load time: 19 seconds
1. Startup face North Gate: 100 – 150 FPS
2. Shipyard: 90 – 120 FPS
3. Planet 90 – 110 FPS
4. Power Plant: 100 – 110

No Jobs startup face North Gate: 120 – 170 FPS

X3-TC

0. Load time: 40 seconds
1. Startup face North Gate: 60 FPS
2. Shipyard: 56 FPS
3. Planet 55 FPS
4. Power Plant: 52 FPS

TC “Classic” mode (shift-n): 73 FPS
No Jobs startup face North Gate: 60 FPS
No Mission Director files – startup face North Gate: 60 FPS
No Director and no Jobs files - startup face North Gate: 66 FPS

X3 Solar Power Plant replaced by TC Terran SPP (SS_FAC_TR_POWER): 20 FPS

Images (left X3-R, right X3-TC)

1. (146, 60)
[ external image ]

2. (112, 57)
[ external image ]

3. (89, 52)
[ external image ]

4. (100, 52)
[ external image ]

TC “Classic” mode (shift-n): 73
[ external image ]

TC Terran SPP: 20 FPS
[ external image ]

X3-TC scripts removed: 100 FPS
[ external image ]

Conclusions

Granted the conditions were somewhat limited in this test series. Actual game experience would involve combat, SETA, and other situations which tend to further degrade performance.

One of the biggest noticeable differences between X3-R and X3-TC seems to involve the new 3D models. For example FPS difference in TC between “looking” at an Argon Solar Power Plant compared with a Terran Solar Power Plant:

Argon SPP = 52 FPS
Terran SPP = 20 FPS

Probably (guessing) many of the other new TC ships and models contribute to similar performance lag.

Another possible contribution to X3-TC poor performance may be the various game scripts.

All TC scripts = 60 FPS
No TC scripts = 100 FPS

Clearly the game needs these (or at least some) scripts. The question is can they be further optimized?

I will document additional data as I collect it.

Akula87 · Post by **Akula87** » Mon, 15. Dec 08, 01:19

Eh... you're amazing

Neven87 · Post by **Neven87** » Mon, 15. Dec 08, 03:10

wow, very detailed analysis. I was quite sad at the FPS im getting in TC but alot of time im spending in terran sectors, which explains alot.

Another thing about terran sectors, they are HUGE compared to the others, i.e. one of the saturns being over 400km

SSwamp_Trooper · Post by **SSwamp_Trooper** » Mon, 15. Dec 08, 13:43

Very clear and detailed analysis, i didn't realize how much performance difference there was

Observe · Post by **Observe** » Tue, 16. Dec 08, 03:17

FYI: In case anyone wants to know, I tested both Steam and DVD versions of the game with no significant performance difference between them. Therefore, the listed measurement data applies to either.

enael · Post by **enael** » Tue, 16. Dec 08, 23:03

you're right in absolute but i'm not agree with your conclusion.

For me the point is :

- 2006 i tried X3 and could not play cause i had a pentium 4 (3 ghz) and a common graphics cards.
- 3 years later i play X3TC with a dual core E8600 and a nividea 9800GT and the game is awesome. I have no more problems even there is more scripts and better graphics.

Developpers made a pretty good job with this game. Don't forget there is thousand and thousand objetcs the game has to deal in real time. For me the real challenge will be to creat a complex universe with a real life.

Observe · Post by **Observe** » Wed, 17. Dec 08, 06:14

enael wrote: you're right in absolute but i'm not agree with your conclusion.

Actually, I didn't really offer any conclusion other than what the data demonstrated.

enael wrote:Developpers made a pretty good job with this game. Don't forget there is thousand and thousand objetcs the game has to deal in real time.

True enough. However, it was interesting that performance remained about the same regardless of presence or absence of Jobs and/or Mission Director cues. In other words, the "thousands and thousands of objects the game has to deal in real time" seems to NOT be the main cause of performance difference.

One "conclusion" which can be drawn is that TC in a more or less empty "lifeless" sector, with no Jobs in the galaxy, is still performing about half as well as X3-R under the same conditions. This suggests to me there is some fundamental issue with TC - which I imagine ES is in process of evaluating and resolving.

wyvern11 · Post by **wyvern11** » Wed, 17. Dec 08, 08:35

nice analysis on this one

@enael : your machine is pretty impressive and i doubt that most players have stuff like that (although fortunately I do now

)

@observe : in general some machines have special shortcomings
with my machine running X3R I noticed different lagginess with different
situations

there are clear and observable distintions between
"not enough memory to run all scripts appropriately"
"lag due to graphics stuff"

on my former machine i started with a mid range graphics setting and later in the game i turned it down and turned it down. later i found out, that fps remained pretty much constant and memory was the main issue

in x3tc the KI-freighter scripts have noticeably improved. multiplied by say 1500 ships running this, I think this has to be taken into account (of course i noticed you stopping all jobs)

My advice is that you maybe have to find out first where your special machine hits the wall first (ram, drive-IO, memory,processor IO)

questions :
does it make a difference if you turn your graphic settings insanely high or low ?
does it make a difference if you run a time consuming universe-analyzying script (eg. log out complete factorylist with all stock info)

aquemnun · Post by **aquemnun** » Wed, 17. Dec 08, 11:26

If this helps i would like to contribute my specs and an analysis of TC running on my computer.

system specs:
pentium dual core 2.20GHz processor (not overclocked)
Nvidia 8600GT graphics card, 256mb
4GB kingston ram (4x 1GB)
Windows Xp SP3

TC specs:
all high, AAx8 AFx8
marine repair script
cycrows cheat script
homebrew fusion of turret fix mod and external docking mod (every ship fixed, yokohama has an M6 dock)

Anaylsis
smooth gameplay most of the time, skips occasionally on SETA. sectors like argon prime are smooth but may slow down if it gets very crowded or if youa ctivate seta. my guess:
normal ~60-70FPS
SETA ~40-50FPS

also I couldnt agree more about the terran stations. They are BEAUTIFUL stations but they take up TONS of memory ( and space

) which is a shame as I use mostly terran stations for my factories now.

2x L food production plants near the east gate in montalaar ~55-60FPS

2x L food production plants and an orbital defense station ~35-40FPS

F***!!!!

hope this may be of assistance to you. if we can get it back to reunion speeds I will be a very happy person. Best of luck observe!

Observe · Post by **Observe** » Thu, 18. Dec 08, 03:20

wyvern11 wrote:...in general some machines have special shortcomings. with my machine running X3R I noticed different lagginess with different
situations...there are clear and observable distintions between "not enough memory to run all scripts appropriately" and "lag due to graphics stuff"

Yes, very true. My particular system is by no means high-end. It's about mid-range Egosoft recommended for TC.

Keep in mind my personal motives are somewhat selfish. As an X3 modder, I wanted a standard by which to ensure my mods do not contribute to further performance degradation.

I encourage all mod creators make similar effort to ensure their mods/scipts/etc do not have negative impact on performance.

wyvern11 wrote:My advice is that you maybe have to find out first where your special machine hits the wall first (ram, drive-IO, memory,processor IO)

Again very true.

aquemnun wrote:If this helps i would like to contribute my specs and an analysis of TC running on my computer.

aquemnun wrote:also I couldnt agree more about the terran stations. They are BEAUTIFUL stations but they take up TONS of memory ( and space ) which is a shame as I use mostly terran stations for my factories now.

Thanks aquemnun. I suspect it's useful for Egosoft having reports from as many different systems as possible.

aquemnun wrote:if we can get it back to reunion speeds I will be a very happy person.

Yes, I believe that is the immediate goal. If Egosoft can bring the basic TC core game in-line with Reunion performance, then TC will have a good chance of being a good evolutionary step. Without that, all the added "stuff" in TC just serves to make matters worse than they need be.

Of course it's all up to Egosoft. Only they have the ability to understand what needs to be worked on to remedy this very obvious issue of poor TC performance. I for one, am confident they will succeed.

Kasabian · Post by **Kasabian** » Thu, 18. Dec 08, 07:31

So in conclusion TC basically has poorly written non-optimized scripts and stations compared to X3 glad I have not bought it yet. I am not very impressed by the game, I have not seen any real reason to get it and between bugs and poor performance I might not ever bother.

wyvern11 · Post by **wyvern11** » Thu, 18. Dec 08, 08:30

no, this is not the conclusion

for about 5000 KI-objects running fairly complex scripts putting rather heavy graphics stuff on top the performance is not so bad.

concernings bugs : it is difficult in general to bugcheck such a complex game **with a given amount of testers in a given time**. to make dead sure everything works you have to fly every ship, fire every weapon, build every factory in any possible complex, trade every ware and do each quest -- this is not payable and thus not possible

having only 5 different player ships, 10 factories and three different weapons would be better bugwise, but...

@ observe
for your "fairly selfish motives" you set up a fairly unselfish anylysis . keep up the good work !!

Observe · Post by **Observe** » Thu, 25. Dec 08, 05:29

FYI: I repeated the described test procedure following patch 1.4, and there is no difference in performance between 1.3 and 1.4.

My apology for double post in the other thread.

deathman20 · Post by **deathman20** » Thu, 25. Dec 08, 06:01

While the test is good, what about model details? Are the models in TC greater than that in R? If so that can have a huge performance impact.

I could test the difference on my system, but it wouldn't give the best results compared to most typical systems here on the forums.

apricotslice · Post by **apricotslice** » Thu, 25. Dec 08, 14:16

Just for interests sake, whats the difference on the lowest settings in each game ?

I'm running TC on the same computer that ran X3R, set to lowest settings. While the graphics are nothing to write home about on this level, the game is playable.

And I must add, my computer is well below minimum specs for TC.

I'm getting a much better game with TC, than I got with X3R with only half a meg of memory. Not as good as with 1.5meg of memory in X3R, but mainly playable. I get a slideshow in Aldrin, and fighting in the khaak sector next to a roid, but mostly, the game is very playable on this computer most people said could not possibly run the game (in the thread where I asked before I got it).

Which is where I'm curious about minimum settings. Because TC seems to be able to run a damn sight better than expected on a computer below minimum spec, than X3R could.

Fus · Post by **Fus** » Sun, 18. Jan 09, 14:02

I'm running at 1280x1025, 0AA 8AF.

I get an average of 35fps in X3 sectors but If I enter Mars and look at those ridiculous Terran stations it drops to 1.5 fps or worse.

E6550 @ 3.25Ghz
9800GT stock speeds
2Gb Corsair 800Mhz @ 928Mhz stock timings.

It's not a new start, I have 4 small complexes and about 8 single stations.
Possibly 25 TS and 2 TL (idle)
Done 1 plot and have 3 others about 1/2 way through.

Scripts - CAG, EST, Bulk Freight

Terran supersize stations do it for me, I don't even have to be close up to them.

Ps Why does a terran shipyard / dock have argon markings in the docking bay ?

dpadula · Post by **dpadula** » Tue, 20. Jan 09, 14:27

That is a great summary of the two games. I guess that I am glad that I am still playing X3R and have not purchased and moved over to X3TC yet. My little 2ghz (Single core) laptop with a 256 meg Nvidia GeForce 7800 would not be able to handle it at all. It has some troubles with X3 as it is already.

I guess I will happily wait for a few more patches first and for when I get a new computer as well.

Marodeur · Post by **Marodeur** » Tue, 20. Jan 09, 14:58

How can i kill all running scripts? I want to test it with a terran station in front. Then i can read the GPU-usage in Rivatuner... Could be interesting...

SpaceSquirrel · Post by **SpaceSquirrel** » Thu, 22. Jan 09, 09:54

I'm looking forward to your mod, who knows when Egosoft will improve the game's performance.

The game runs very fast on my computer except it constantly pauses during play. Sometimes during combat it pause for 5 or 6 seconds. Usually intermittant pausing, pause, play, pause etc.

I have a fast computer with only about 8 services running in the background. So I don't understand all of the pausing.

shinyclef · Post by **shinyclef** » Thu, 22. Jan 09, 10:09

SpaceSquirrel wrote:I'm looking forward to your mod, who knows when Egosoft will improve the game's performance.

The game runs very fast on my computer except it constantly pauses during play. Sometimes during combat it pause for 5 or 6 seconds. Usually intermittant pausing, pause, play, pause etc.

I have a fast computer with only about 8 services running in the background. So I don't understand all of the pausing.

I have a feeling that's a graphics card issue related to your drivers. I used to have a similar problem, but the pauses were long and accompanied with sound loops. Anyway, the problem was fixed with different drivers. Some drivers still had the pauses, but I eventually found one that cleared up the problem completely. My pauses were not just restricted to X3 though, they were in all games.

I don't know if it's the same problem, but it couldn't hurt to try different drivers, especially since it doesn't seem to be a common problem so probably not caused by X3. Good luck with that though.

egosoft.com

egosoft.com

Performance comparison between X3-R and X3-TC

Performance comparison between X3-R and X3-TC