We need help narrowing down nVidia problems

[peleprodigy]

I think the real problem is weeding out the poorly configured systems (potentially user error). Excluding the outliers would be my first step.

I think the frame rate question goes back to something I was asking several weeks ago, a question that was never answered: what system did the developers use when they maxed out the game during development? That is, what system combination did they use to get the ideal performance at best visual quality? If I had labored over a game for several years I would certainly want to know how it looked and played in all its glory. But they never answered the question, all there is is the 'suggested' specs on the box. and all of us that blow the suggested specs out of the water are left to wonder.

[ August 26, 2007, 02:56 PM: Message edited by: peleprodigy ]

[SlapHappy]

Originally posted by mike_in_texas:

Thanks for the advice, SlapHappy. I've already tried the MS hotfix with no noticeable difference in performance, unfortunately.

The difference in framerates can be drastic. For example, in the Al Hawl scenario with everything set to 'best', I can get anywhere from >100 down to 4-5 FPS depending on what's in my camera's FoV.

Mike

Mike, yeah, so do I...Which is kind of what I was wondering about. You guys are talking about your minimum FPS, then, right?

I wonder if this has more to do with the map size issue which I think all of us are suffering from. Perhaps that has been the culprit all along.

If you are, however, getting that kind of performance on much smaller maps then it becomes much harder to say if it's an issue specific to the 8800 series cards.

Perhaps only baseline examples would really be useful in this case (testing with the same exact content, with same exact settings across different cards). My numbers are kind of all over the place, too. I think this is a case where an integrated benchmarking feature would come in handy.

[kevinkins]

Instead of getting a PC system into Battlefront why not arrange a visit to the person(s) having the problem and "inspect" their system? Many many industries use this approach to solve problems just like we are reading about. It's called Observational Research and its very cost effective. It should not surprise anyone that the single system BF purchased has no issues. Fleeting problems are just that .. they make no sense ... and these guys are good at what they program.

Kevin

[ August 26, 2007, 04:21 PM: Message edited by: Kevin Kinscherff ]

[Battlefront.com]

Peleprodigy,

I think the real problem is weeding out the poorly configured systems (potentially user error). Excluding the outliers would be my first step.

We did that the first weekend the game was out and we're still left with a muddle. User error can never be ruled out, but we certainly haven't figured out a common cause and have some pretty "with it people" that are having a problem. As for what systems we used to test... a couple dozen, no problems. Remember, there are literally thousands of possilble configurations out there so the fact that there is a fairly small (but significant) number of diverse systems with a similar problem is not all that surprising to us. Better than the days of DOS and Windows 3.1, but unfortunately "Plug and Play" is still not as good as it should be.

Kevin,

Instead of getting a PC system into Battlefront why not arrange a visit to the person(s) having the problem and "inspect" their system?

If there was someone local, perhaps, but there is a big difference between the sort of software developers your talking about and a small group like us with exactly one programmer. So to us it is not cost effective. Charles needed a new system anyway so we simply tailored the purchase to something that appeared to be pretty open and shut manifestation (nVidia 8800 + multiprocessor). But alas, not so easy. It's frustrating.

Steve

[mike_in_texas]

SlapHappy,

Yes, the 4-5 FPS is about the minimum rate I get when running the Al Hawl scenario, when I have most of the town in the FoV.

I haven't evaluated too many other scenarios, but if people find it helpful I'll be more than glad to. I've tweaked the various settings -both directly through my graphics cards' GUI as well as the options in CM:SF- to no avail inasmuch as finding a setting that gets consistently acceptable (in my opinion: 20-30 FPS) rates when running with 'better' or greater model quality.

Mike

[The DesertFox]

Originally posted by Battlefront.com:

Hi all,

After waiting for nearly 4 weeks Charles' brand new system arrived yesterday with Vista, AMD Athlon, and nVidia 8800 installed. He fired it up today and guess what? Great framerates.

Hi!

it would be great to know what you guys consider great framerates. Any chance to post the results of Charles combo here with say the Allah´s Fist scen as a benchmark? Only start up the scen from the blue side and don´t move.

My 7950GT rig is giving me 15 FPS

My 8800GTS rig scores 8 FPS

Game resolution, AA, AF, or in-game settings only marginally affect this by 1-2FPS.

cheers

Helge

[The DesertFox]

Originally posted by rune:

Allah's first scenario, default screen US which everyone is complaining about, 14-16 fps. With FSAA turned off, I got 25 fps.

Seems like you are a lucky bastard, since these are the fastest results so far. It can´t be the quad because a core2duo@3400 is significant faster. Dou you have an ULTRA card or a OCed GTX in your rig ???

Did you post the initial framerate after firing up Allah´s fist from the blue side without moving your mouse ???

cheers

[peleprodigy]

It will be terribly upsetting if 25 fps is the best frame rate that can be achieved with this game. For me, 30 fps is the floor for an acceptable frame rate, with rare stutters below that during heavy on screen action.

[Cougar_DK]

Steve, could it be possible to make a debug version of the exe file that creates a text file or the like with useful info to you guys? I.e. you count the frames, versus the number of polygons etc.

I do this as a professional developer when the debug is pretty nasty, but again, I'm not developing games.

[rune]

Fox,

eVga card, so it is a 8800GTX overclocked. Yes, I posted the initial startup screen, blue forces without moving the screen or mouse, so default on everything. Actually with my motherboard, i am running at 3.00 ghz overclocked, using the Asus Booster and overclock program. I could do better if I had better memory, but the Striker Extreme motherboards are very picky with memory.

Rune

[Cpl Steiner]

Can I just mention something that might be useful here. I just played the Al R'Lyeh Mosque Scenario and was getting VERY acceptable frame-rates in the 20s and 30s. There was a whole Stryker company on the attack in that scenario, and lots of built up areas, so I think the polygon count must be comparable to many scenarios. However, the map is quite a lot smaller than the ones in most scenarios.

I am beginning to think the graphics issue is a red herring and it is a LOS calculation issue. Smaller maps equal substantially better frame-rates. If you are getting very low frame-rates, try this scenario with FRAPS running and see what you get.

[The DesertFox]

Originally posted by rune:

Fox,

eVga card, so it is a 8800GTX overclocked.

626 MHz GPU and Memory at 2000 MHz certainly make a difference However isn´t it funny that my good 'old 7950GT shows nearly the same performance ?

Originally posted by rune:

I could do better if I had better memory, but the Striker Extreme motherboards are very picky with memory.

You know that you can unlink memory ocing from FSB ocing on your 680i SLI board? As for your quad...I wouldn´t want to go above 3Ghz with it without having it water-cooled. These things turn damn hot compared to the core-2-duos.

[rune]

Fox,

yep, I know...but with testing, I don't want to throw another unknown into the mix, and actually turn off oc'ing when doing the fps stuff. The 680i chipset is picky on memory, but at least the bios problems seem to be ironed out.

I also have a XP system with 6800 and a Vista system with 6800 card, identical systems, and no surprise, the XP system runs faster. I temporarily unhooked my Vista X64 system with a 7800 card in it, so can't do a comparision there. I should try different desktop sizes to see how much a difference it makes.

Rune

[nghtmre15]

The following system encounters severe framerate issues when looking at any map bigger than the smallest size:

Windows Vista 64-bit Edition

Intel Core 2 Quad Q6600 2.4 GHz

4 GB DDR2800 RAM

Nvidia GeForce 8800 GTX

Nothing is overclocked, and no problems like this are encountered with any other programs that I own.

[Battlefront.com]

Map size has more impact than units since units can be LOD'd a lot cleaner than terrain can. So yeah, I would expect a small map with a lot of units to run better than a large map regardless of numbers of units. There are also particular types of terrain that make a map run slower than another one, such as trenches, steep elevation changes, trenches, flavor objects, trenches (heh... have I mentioned trenches enough?).

20-30 fps is what we consider ideal framerate for CM:SF. Remember, this isn't an FPS game where every dropped frame below 30-40 can mean "life and death". As with CMx1 games when they came out, we find that 10-20 fps is quite playable. Less than 10 fps things become problematic for sure.

Cougar_DK

I do this as a professional developer when the debug is pretty nasty, but again, I'm not developing games.

This is actually how we managed to get the ATI clicking problem fixed. However, in this case Charels knew roughly where to start looking for a problem (i.e. clicking). With this framerate thing... it's way too systemic and therefore a special debug build probably isn't the way to go.

Steve

[Cougar_DK]

Originally posted by Battlefront.com:

Cougar_DK

</font><blockquote>quote:</font><hr />I do this as a professional developer when the debug is pretty nasty, but again, I'm not developing games.

This is actually how we managed to get the ATI clicking problem fixed. However, in this case Charels knew roughly where to start looking for a problem (i.e. clicking). With this framerate thing... it's way too systemic and therefore a special debug build probably isn't the way to go.

Steve </font>

[Dirtweasle]

Just installed and licensed CMSF on my new machine and it flashes the trees on and off, then stabilizes for a bit, set the units to move and press go and the thing locks up;

CMSF v1.01

Vertical = ON

Display = DESKTOP

3d qual = BETTER

3d textrue = BETTER

AA/Multi Sample = ON

Language = English

~~~~~~~~~

Windows XP Professional (5.1, Build 2600) Service Pack 2

Intel® Core2 Duo CPU E6750 @ 2.66GHz

Gigabyte P35C DS3R Mobo

2046MB RAM

Nvidia GeForce by MSI 8800GTS 640

running at 1280 x 1024 (32 bit) (75Hz)

DirectX 9.0c (4.09.0000.0904)

Will try loading the 1.02 patch and report back...

PS - Medieval War II TW runs fine, no other stability issues or glitches with anything else.

This is a brand new just built PC, let me know what to try and fiddle with beside using the most current vcerion of CMSF.

EDIT to add, no better luck with v1.02 ...

EDIT to add, changed to newest nvidia drivers, no change.

Modified CMSF settings

Vertical = FF

Display = DESKTOP

3d qual = BALANCED

3d textrue = FAST

AA/Multi Sample = OFF

Language = English

...no change. Trees bushes and whatnot flash, and then when you give a unit an order and press go, game freezes up.

:confused:

[ August 28, 2007, 04:10 PM: Message edited by: Dirtweasle ]

[rune]

Try affinity of 1 processor, and the microsoft hotfix.

Rune

[Trooper5]

I'm getting pretty good performance 20 to 50 fps, but if I engage antialiasing the Battlefront intro screen disappears.The high end frames comes with textures and model set at improved or balanced I also have the shadows,and V sync off. My system.

nForce 680i MOBO

Intel core 2 duo E6700 @ 3.25GHZ

2 GB DDR2 RAM @ 800MHz

GeForce 8800 GTX 768MB

Western Digital Raptor 2 x 150GB 10000 rpm

[peleprodigy]

20-30 fps is something to be settled on and not an ideal, no matter the game in question. there is no reasonable explanation for this game to not run maxed out with at least 30 fps for 80% of the machines listed on this thread. are you really telling me that combat mission:shock force is the most graphically intensive game on the market today? please. I wish I could see fewer excuses and rationalizations and more optimizations and solutions.

[Dirtweasle]

...never mind rune, found it in the other thread. Back later.

EDIT TO add

The affinity setting got me running, thanks. No way to have the game either do that or run with it both selected? That's kind of the point of a dual core processor, no?

I D/L'd the hot fix, but have not installed it. Not running any sort of power management scheme, beside always on that is.

How can I check the frame rate, it looks pretty good playing al amarah, not complaining really mind you, but might be worthwhile measuring it when things get going in the battle.

...by the way, thanks again for the quick reply!

[ August 28, 2007, 07:36 PM: Message edited by: Dirtweasle ]

[Azzu]

I had a really strange drop from game to desktop after I installed the game. I cannot give you the exact error message because a) it would be in Finnish, had to translate myself and Never been able to get the same error again.

I have:

Windows Vista Home Premium

Intel C2D 6550, 2.33 Ghz

2MB Ram

Nvidia 8600 GTS

When I first time ran the game it worked perfectly. Good FPS even when relatively big map with lots of buildings (I played quick battle). I played the game for about 15 mins. After that the game crashed to desktop and the Windows message was in english something like: "Your graphic card had to be shut down. Windows will now restart your graphics card.". After that it ran very lousy with appalling framerates. Even after reboot. Absolutely no problems with any other games. Tried to re-install CMSF and Nvidia drivers but no change. Really strange...

Edit: typos

[Sekra]

Dirtweasle: I had that same problem but installing the MS hotfix did the job for me. I'm not running any power management either so it will help you still I think. I've never also had any problems with other games until now.

[rune]

Dirtweasle,

No problem, glad you are in the game and playing. For a true test, send me your email, will send you a scenario to make your system cry.

Rune

[YankeeDog]

Originally posted by peleprodigy:

... are you really telling me that combat mission:shock force is the most graphically intensive game on the market today? please. I wish I could see fewer excuses and rationalizations and more optimizations and solutions.

I don't know about *most* graphically intensive, but if the history of CMx1 is any indicator, the CM games require considerably more graphics resources than contemporary FPSes, at least on large maps w/ large # of units. So I wouldn't be at all surprised if CMx2 is more "graphically intensive" than, say, BioShock.

AIUI, reason has to do with the sheer number of units and terrain features combined with the size of the field of view you can have in CMx2 -- you rarely see ranges of more than a few hundred meters in an fps, and a couple of dozen moving units in that fov, tops. CM games can beat these figures by a factor of magnitude.

There are obviously some problems with the framrates in CMx2 and the 8800 series cards -- players with this kind of hardware *should* be seeing better framerates. But do not assume that, just because your 8800 runs BioShock at 60+ fps on max detail settings, that it should also be able to run a 4kmx4km armored battalion vs. armored battalion CMx2 scenario at 60+ fps. Apples to oranges.

Cheers,

YD