Take the BFTB performance test!

[Toby42]

So what's the numbers mean? High number of minutes good or bad??

[MarkShot]

XXXX250 999 means:

XXXX - The particular game engine.

250 - The approximate number of units (both sides) active at that moment in the scenario.

999 - The maximum number of simulated minutes that the player's system can compute in one minute of wall clock time.

---

250 is the average unit count for BFTB scenarios.

600 is the maximum unit count for BFTB scenarios.

I did this for a number reasons. First, I was curious for myself whether my existing desktop would handle BFTB reasonably. Second, I thought others might find this useful in getting a sense of how BFTB runs compared to the other games, and what they might expect from BFTB if they buy it. Combined with my XLS, you should be able determine if you'll be satisfied with BFTB's performance. There is not right or wrong answer to that question, since one player's tedious wait is another's immersive experience. Of course, those with concerns can and should perhaps wait for the demo, but it is not yet clear just how sizeable the demo scenario(s) will be.

However, I suppose if the demo only has 100 units, then I can do a BFTB100 benchmark. Then, potential customers can use the ratios found in this thread for BFTB100:BFTB250:BFTB600 to extrapolate what they will see if they buy the game.

Well, I hope that answers your question.

[MarkShot]

In case I wrote too much, higher is better.

[Toby42]

I ran the HTTR scenario on my "OLD" desktop. Don't laugh at it! I'll have to try it on my laptop. It's a lot stronger machine, and newer..

174 minutes
Pentium 4
3.06 ghz
1 gig of shared Ram
XP Service pack3

[MarkShot]

That's a very respectable score. Higher is better.

[MarkShot]

I'm not a PC hardware guru, but it is beginning to look like clock speed is the key factor even more than CPU model despite L1|L2 cache sizes, pipelines, predictive execution, scalar processors, etc...

[Toby42]

ORIGINAL: MarkShot

That's a very respectable score. Higher is better.

My pc specs is the largest drawback to picking up BFTB now. That's why I was waiting for the demo and buying the earlier titles first!!

[oldspec4]

Mark..scored an 87 for CotA250.

Am running an AMD quad at 2.8Ghz, 8GB DDR3 ram, 9800gt, and Win 7(64).

[Prince of Eckmühl]

Markshot,

Are you wanting to know what it would take to open the game up to larger forces, to bust the 600-unit barrier?

[Prince of Eckmühl]

ORIGINAL: Prince of Eckmühl

BFTB250: 147
BFTB600: 40
COTA250: 116

Intel Core2Duo 3.33Ghz + 3GB RAM + Windows 7 Home Premium

Edit: Fat-fingered BFTB600; corrected.

Out of curiosity for how the numbers would scale with processor speed, I upped the system bus from 166mhz to 200mhz. That increased my system speed to 4ghz. I got these results at that setting:

BFTB250: 175
BFTB600: 48
COTA250: 140

The game's ability to churn through the game clock is definitely CPU dependent. The twenty-percent increase in processor speed generates proportionately similar improvements in game performance.

[MarkShot]

The 600 unit scenario wasn't really any engine or scenario limitation. It just happens to correspond to the largest scenario available in BFTB.

I suppose you could create a 1200 unit scenario, but I suspect that there might four fold drop in the simulation rate based on current results.

[hlhudson]

Mark,

BFTB250: 126 Simulated Minutes
BFTB600: 36 Simulated Minutes

COTA250: 121 Simulated Minutes

HTTR250: 177 Simulated Minutes

PC @ Work
Intel E8400 Dual Core @ 3GHz
4GB RAM
NVIDIA Quadro NVS 290 with 256MB RAM
Windows XP SP3

[PirateJock]

As a compare and contrast ...

CPU - Intel Core Duo 2.4GHz
RAM - 4MB
Graphic - Mobile Intel 965 Express
OS - WinVista Home

BftB250 = 83
BftB600 = 29
CotA250 = 87

Samsung NC10
CPU - Intel Atom N270 1.6GHz
RAM - 2MB
Graphic - Mobile Intel 945 Express
OS - WinXP Home

BftB250 = 32
BftB600 = 8
CotA250 = not on machine

[MarkShot]

Well, for those of you who don't have any of the games and have very low spec systems, you should see:

tm.asp?m=2479970

I have concluded that despite HTTR being the oldest game of the series, COTA is actually the best game to get for a CPU anemic PC. Not so much because of the simulation rate, but rather because of the scenario mix which it contains.

[Toby42]

ORIGINAL: MarkShot

Well, for those of you who don't have any of the games and have very low spec systems, you should see:

tm.asp?m=2479970

I have concluded that despite HTTR being the oldest game of the series, COTA is actually the best game to get for a CPU anemic PC. Not so much because of the simulation rate, but rather because of the scenario mix which it contains.

Are you supposed to run your tests at the fastest rate(>>>) or the normal rate(>>)?

[MarkShot]

Tony,

See post #1 of this thread for what you need to do.

[Toby42]

ORIGINAL: MarkShot

Tony,

See post #1 of this thread for what you need to do.

That's what I thought. I ran the HTTR test scenario on my laptop and the results were almost identical to my old desktop!


HTTR 250

Intel Core i3-330m Processor
2.13ghz, 3mb L3 cache
Intel Graphics shared memory
4 gig of ram

176 minutes!

I'm surprised. I thought that my new laptop would blow away my six year old desktop!!!

[MarkShot]

That does seem to be a pretty high score. I think you better treat your desktop with some respect!

[MarkShot]

Actually, based on more analysis a normalized figure for average BFTB unit count would appear to be closer to 150 than 250. Additionally, as a normalized figure even that number is high, since in most of the larger scenarios, it is only by the 50-66% mark that so many units are present on the map.

Oops ... actually, the 250 number is reasonable, since when I got the normalized units I had divided the figure by 2 to reflect just the human player side. Given the arrival of reinforcement lag, 250 is probably a decent approximation.

[cdbeck]

Mark, I'm quite certain you meant 4 GB of RAM in your systems specs posted in the second post. Otherwise I would say the game is very much not RAM dependent... [;)]