Feedback - Extreme FPS Loss through binocs #3782

[grey13]

I had the same fps problem. I went into the single digits looking at a forest of the new trees with sight zoom /binocs. I have screen shots but I am unable to attach them.

Edited March 25, 2010 by grey13

[AHWULF]

When you go into a zoomed view, the game renders more 3d trees, thus putting stress on whatever is wrong in how the (mostly) Nvidia drivers/opengl interacts with the shader that draws the tree bits. It doesn't happen for me on an ati 4870 so it clearly something fixable, but figuring out what it is isn't easy. It also possible its something we can't fix other than badgering NVidia to issue a fix in their drivers (if you read the driver update notes, you'll notice all sorts of fixes for games). Proving that is hard.

Note that in 131 we have all new shaders and higher resolution plus new grass and clutter, so comparing to 130 is relatively moot.

Does anyone see this issue in an ATI card with 131?

[hick]

When you go into a zoomed view, the game renders more 3d trees, thus putting stress on whatever is wrong in how the (mostly) Nvidia drivers/opengl interacts with the shader that draws the tree bits. It doesn't happen for me on an ati 4870 so it clearly something fixable, but figuring out what it is isn't easy. It also possible its something we can't fix other than badgering NVidia to issue a fix in their drivers (if you read the driver update notes, you'll notice all sorts of fixes for games). Proving that is hard.

Note that in 131 we have all new shaders and higher resolution plus new grass and clutter, so comparing to 130 is relatively moot.

Does anyone see this issue in an ATI card with 131?

What's interesting is I do not have this problem at all with my Quadro.

[mcketten]

I am just really not seeing this on my rig. (9800GT)

Sure I get some reduced frames from looking at high res trees with all the bells and whistles turned on. It does seem more pronounced through binos than if I walked right up to the tree but nothing bringing me to teens fps numbers.

Ticketed anyway for a look. #3782

question gophur: are you running with triple-buffering enabled and vsync on?

i have noticed that with a majority of people reporting this once they turn vsync/trip buff on it either goes away entirely or reduces in severity.

[AHWULF]

IF you normally get more than 60fps you should always have vsync on, otherwise (oddly) your FPS may suffer a lot. The only time to have vsync off is when you have a slow video card or CPU and never see 60fps. Even then it will be smoother with it on.

[hick]

Even then it will be smoother with it on.

I think that's the real advantage to vsync when average Framerates are below your monitor's refresh rate.

For instance, in 1.30 I get a consistent 60 FPS with Vsync on (tbuffer not required).

in 1.31 I get a a consistent 30 FPS with VSync on (tbuffer off). With Vysnc off I get anywhere from 18-60 FPS. I'll take a steady 30 over a 40 frame spread any day. (incedentally, it's 30fps because my frames are rendering just out of phase with my monitor's refresh cycle, thus the monitor only loads a new frame ever 2 cycles).

[Bacon55]

Does the Quaddro use the same drivers as the consumer cards?

[hick]

Does the Quaddro use the same drivers as the consumer cards?

no, the use drivers optimized for OpenGL used in MCAD and DCC applications (among others).

A nice paper from nVidia explains it well (paper is from 03, so some of the features have moved to the GeForce line, but the actual way images are rendered is still the same)

http://www.nvidia.com/object/quadro_geforce.html

[ghostrider]

has anyone downloaded the new nvidia beta drivers and tried those?

[Bacon55]

has anyone downloaded the new nvidia beta drivers and tried those?

Yes, no difference.

Clean install from safe mode, driver cleaner, as well.

[Bacon55]

So the problem isn't the hardware, the stats speak for themselves.

On virtually every other game a 260 runs exactly as expected, about 80% of a 285, or about the same as a 5830.

I've heard the GTS 250 for example on this game outperforms it.

That doesn't make sense, aside from a software error. In terms of hardware the 260 is pretty much superior in every way.

The benchmarks prove this.

http://www.tomshardware.co.uk/geforce-gtx-radeon,review-31608-18.html

http://www.tomshardware.co.uk/geforce-gtx-260,review-31573-14.html

http://www.tomshardware.co.uk/geforce-gtx-260,review-31573-16.html

http://www.tomshardware.co.uk/geforce-gtx-260,review-31573-15.html

So before you blame the hardware, accept that this card and others like it perform well in other environments...this pretty much has to be a software issue.

[Bacon55]

Ok, tried forcing triple buffering and Vsync, stays solid at 60 fps no matter what I'm looking at.

Everyone try it out.

[OLDZEKE]

Try spawning Nijmegan/Lent depot, go to top of depot and look due north and check fps with and w/o binos

[OLDZEKE]

OK, some what I think are odd happenings with this.

Geforce 260 vid card. Driver ver 196.34

Offline A51 standing at the post looking up the road test:

Anistropic filtering, setting this to none or to 16x in nvidia control panel makes no fps difference in the test. Either the game is forceing full anistropic filtering or it's not accepting any setting from the control panel? Maybe a nvidia drivers bug? Other nvidia users seeing this?

Can ati users check to see if anistropic filter settings in the cat center affect fps in this test?

[hick]

OK, some what I think are odd happenings with this.

Geforce 260 vid card. Driver ver 196.34

Offline A51 standing at the post looking up the road test:

Anistropic filtering, setting this to none or to 16x in nvidia control panel makes no fps difference in the test. Either the game is forceing full anistropic filtering or it's not accepting any setting from the control panel? Maybe a nvidia drivers bug? Other nvidia users seeing this?

Can ati users check to see if anistropic filter settings in the cat center affect fps in this test?

don't think it's a driver issue unless it applies to all nVidia products (I use a Quadro card, so completely different firmware and drivers). I noticed the same thing when setting AF in the Beta. Either the game is forcing it or not allowing it.

[AHWULF]

When you turn it on / off does the screen appearance change?

The game does not do anything with that feature on purpose.

[OLDZEKE]

When you turn it on / off does the screen appearance change?

The game does not do anything with that feature on purpose.

Not that I can tell, another oddity is that in 1.30 forceing 16x anistropic actulay ups my fps some over no anistropic... I'm gonna uninstall/reinstall my vid drivers, if I still have the installer as otherwise it's a 10 hour d/l

OK uninstalled/reinstalled the drivers, no change.

Best as I can tell from the ground textures the game is doing 16x regaurdless of vid setting. Ground tile grass texture looks sharp and crisp.

Edited March 26, 2010 by OldZeke

[Bacon55]

Not that I can tell, another oddity is that in 1.30 forceing 16x anistropic actulay ups my fps some over no anistropic... I'm gonna uninstall/reinstall my vid drivers, if I still have the installer as otherwise it's a 10 hour d/l

OK uninstalled/reinstalled the drivers, no change.

Best as I can tell from the ground textures the game is doing 16x regaurdless of vid setting. Ground tile grass texture looks sharp and crisp.

I noticed that even forcing Vsync off with Nvidia control panel did NOT force Vsync off in game.

Something is awry.

[peegee77]

Ok, tried forcing triple buffering and Vsync, stays solid at 60 fps no matter what I'm looking at.

Everyone try it out.

No luck for me. Tried a standardised view offline (the one looking along the road from the default inf spawn) and here is what I got (without binos / with binos)

Vsync OFF, Triple Buffering OFF

31 / 12

Vsync ON, Triple Buffering ON

30 / 10

Vsync ON, Triple Buffering OFF

30 / 10

Vsync OFF, Triple Buffering ON

38 / 11

So, a slight boost to the normal view running triple buffering without vsync, but no effect on the view through binos.

------------------------------

Windows XP Home Edition (5.1, Build 2600) Service Pack 3

Processor: AMD Athlon 64 X2 Dual Core Processor 6000+, MMX, 3DNow (2 CPUs), ~3.0GHz

Memory: 3072MB RAM

Card name: NVIDIA GeForce 9500 GT

Display Memory: 1024.0 MB

Current Mode: 1280 x 1024 (32 bit) (60Hz)

Driver Name: nv4_disp.dll

Driver Version: 6.14.0011.9621 (English)

[Bacon55]

There is definately an issue with the zoom/buffering. However with a 9500 I think that might be a case of not enough power.

I don't mean to be a d*ck but it'd be nice to get a little more feedback on this from the Rats.

[peegee77]

There is definately an issue with the zoom/buffering. However with a 9500 I think that might be a case of not enough power.

I don't mean to be a d*ck but it'd be nice to get a little more feedback on this from the Rats.

The 9500 GT is certainly not the latest or greatest graphics card, I know. But it should be perfectly adequate for the task. I am less and less convinced that the GPU performance (as long as it is above a certain minimum requirement) has much to do with it. I can crank all my settings up to max, or chop them all back down to fugly, it doesn't have much affect on FPS at all.

But again, if it is CPU load, what is the issue. I've got an AMD Athlon 64 X2 6000+ ~3.0GHz stock. Not the best or the newest processor you might say, but it still has enough oomph for the job, and handles everything else I use it for without a problem.

I can't think of any other game or application where changing your FOV slashes your fps down to 1/3rd of normal. That has got to be an issue in the coding.

I agree that you could probably avoid the issue by building the biggest and baddest PC ever seen just to play WWIIOL. But that is not actually addressing the basic issue - it is simply chucking so much power at the problem that you no longer see it.

There is a basic problem with zoomed views that has always been there, and it has never been fixed. In 1.30.x.x it wasn't such a big deal, because you "average rig" might get around 60fps, and if it dropped to 20fps while zoomed it was still playable. But if the "average rig" is only getting 30fps in 1.31.x.x and it falls to 10fps zoomed, that is gonna be a gamebreaker for a lot of people.

[ghostrider]

I've tried playing about with settings both in game and in the nvidia control panel; full performance settings vs full quality settings and there is very little or no change to overall FPS. Zoomed in FPS on trees and bushes is always about 10-12.

I'm running an 8800 GTS 640mb

[AHWULF]

Have you all tried running the game at some low resolution like 800 x 600? Does it change the behavior at all? If your FPS doesn't improve with fewer pixels then it must be a CPU side issue (could be *in* the driver or the code).

This would be a handy clue for us.

BTW we are still in Beta so don't assume nothing will be fixed :-)

[ghostrider]

currently my resolution is 1920x1080; my 24in screen's native resolution is 1920x1200 but for video editing purposes i put it to 1080.

if i drop my resolution in game to 800x600 i do get an FPS improvement as to be expected but it still drops my FPS down to less than 30 looking at forests.

at 1280x720 my FPS looking at forests is 17

and at 1920x1080 my FPS is 12.

[specterx]

The 9500 GT is certainly not the latest or greatest graphics card, I know. But it should be perfectly adequate for the task. I am less and less convinced that the GPU performance (as long as it is above a certain minimum requirement) has much to do with it. I can crank all my settings up to max, or chop them all back down to fugly, it doesn't have much affect on FPS at all.

But again, if it is CPU load, what is the issue. I've got an AMD Athlon 64 X2 6000+ ~3.0GHz stock. Not the best or the newest processor you might say, but it still has enough oomph for the job, and handles everything else I use it for without a problem.

I can't think of any other game or application where changing your FOV slashes your fps down to 1/3rd of normal. That has got to be an issue in the coding.

I agree that you could probably avoid the issue by building the biggest and baddest PC ever seen just to play WWIIOL. But that is not actually addressing the basic issue - it is simply chucking so much power at the problem that you no longer see it.

There is a basic problem with zoomed views that has always been there, and it has never been fixed. In 1.30.x.x it wasn't such a big deal, because you "average rig" might get around 60fps, and if it dropped to 20fps while zoomed it was still playable. But if the "average rig" is only getting 30fps in 1.31.x.x and it falls to 10fps zoomed, that is gonna be a gamebreaker for a lot of people.

+1 to this post. WW2OL has a number of long-running performance bugs that have never been fixed. Computer power has increased by many orders of magnitude over 2001 and there's no reason that anyone running an "average" system for 2010 should be having any performance issues at all.

The problem is in the code. 1.31 should by no means be too much for an average, or even low-end 2010-era system to handle, but clearly something is breaking down under the hood.