PDA

View Full Version : CPU vs. GPU limitations



TheFamilyMan
03-05-2008, 09:27 AM
I put together a quick mission of 8 spitfires, 8 hurricanes against 8 He111 and 8 Ju88 over crimea, noon on a clear day. Just as the combat starts, my display stutters and drop to about 15 fps at times but eventually goes back to normal after a minute or two. If I record a track from this mission and play it back, it's smooth as glass thoughout with fps pretty much pegged at 60 (vsync enabled), even when hopping from plane to plane. I've also noticed that playing online with lots of aircraft causes no performance problems. Anytime I run offline, i.e. all AI, my system eventually chokes past a certain number of aircraft. Which leads me to this conclusion: the computation for the flight and damage models can really tax your CPU, much more so than the actual graphics load provided your graphics card(s) are up to it. Too bad IL-2 can't take advantage of multiple cores http://forums.ubi.com/images/smilies/16x16_smiley-sad.gif

Anyone else notice this?

My specs: 939 Athalon64 X2 4200+ (oc'ed), 2G DDR 450, 7800gt SLI (oc'ed).

TheFamilyMan
03-05-2008, 09:27 AM
I put together a quick mission of 8 spitfires, 8 hurricanes against 8 He111 and 8 Ju88 over crimea, noon on a clear day. Just as the combat starts, my display stutters and drop to about 15 fps at times but eventually goes back to normal after a minute or two. If I record a track from this mission and play it back, it's smooth as glass thoughout with fps pretty much pegged at 60 (vsync enabled), even when hopping from plane to plane. I've also noticed that playing online with lots of aircraft causes no performance problems. Anytime I run offline, i.e. all AI, my system eventually chokes past a certain number of aircraft. Which leads me to this conclusion: the computation for the flight and damage models can really tax your CPU, much more so than the actual graphics load provided your graphics card(s) are up to it. Too bad IL-2 can't take advantage of multiple cores http://forums.ubi.com/images/smilies/16x16_smiley-sad.gif

Anyone else notice this?

My specs: 939 Athalon64 X2 4200+ (oc'ed), 2G DDR 450, 7800gt SLI (oc'ed).

Urufu_Shinjiro
03-05-2008, 09:42 AM
Yes, this is my problem right now, I have an X1950XT and it is more than capable of doing anything I need done in this sim but when I get online and the cpu has to start calculating all the ground objects that permiate most good server maps these days it chokes and I'm barely pulling out frams above 30fps. I have an AMD 939 X2 3800+ oc'd from 2.0 to 2.6ghz but I beleive there is a problem somewhere performance wise in that oc as all indications are a cpu limitation not gpu. My 3dmark06 is a little low for my system as well. I plan on upgrading befor the year is over so I'm not super concerned with it but it is a pain.

roybaty
03-05-2008, 10:23 AM
I forgot who, but someone here has a quad core Intel and they said they were having excellent results with large missions.

Xiolablu3
03-05-2008, 10:48 AM
What settings are you using?

My 4200+ X2 skt 939 and 7900GS has no problems in large maps on Perfect.

I havent tried the latest patch with the enhancements yet however, is this what you mean?

Have you tried changing your drivers? Do you have plenty of RAM?

striker-85
03-05-2008, 10:55 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by TheFamilyMan:
Which leads me to this conclusion: the computation for the flight and damage models can really tax your CPU, much more so than the actual graphics load (provided your graphics card(s) are up to it).

Anyone else notice this?
</div></BLOCKQUOTE>

Yes, I have noticed this exact same thing. When I upgraded my 7950GX2 to an 8800GTX I didn't notice much improvemnt. After spending some time using various tools to monitor my system I determined it was CPU bound. I upgraded to a Core 2 Duo E68000 Extreme and overclocked it and made many more tweaks to my system and it greatly improved my frame rate and eliminated any stutters.

K_Freddie
03-05-2008, 11:16 AM
In your GPU card utilities, setup more VRam for a balance of textures and computations.
Keep in mind the CPU is more a 'data pusher' and most computations are done on the GPU side
http://forums.ubi.com/groupee_common/emoticons/icon_cool.gif

Urufu_Shinjiro
03-05-2008, 11:18 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Xiolablu3:
What settings are you using?

My 4200+ X2 skt 939 and 7900GS has no problems in large maps on Perfect.

I havent tried the latest patch with the enhancements yet however, is this what you mean?

Have you tried changing your drivers? Do you have plenty of RAM? </div></BLOCKQUOTE>

There's some details in this thread here, please feel free to add anything http://forums.ubi.com/eve/forums/a/tpc/f/49310655/m/733...331015326#7331015326 (http://forums.ubi.com/eve/forums/a/tpc/f/49310655/m/7331015326?r=7331015326#7331015326)

I've since gotten a 24" lcd and running 1920x1200 hasn't changed much, still getting about 32 avg on BD track but you would think that online you would get a little more since BD is supposed to be really tough on the system but once I'm up and flying I get around 28-35 fps flying in skies of valore and others and this is with no mods enabled.

Stew278
03-05-2008, 11:41 AM
I appear to be CPU limited too. I have an X2 4200+ set at 2.8GHz and 2GB ram but I still get chugging behavior on some maps (offline). The Berlin map in particular seems to put a strain on it. I've seen more or less the same average frame rates with 3 different vid cards: 7800GT, 7900GS, and X1950XT. I know the GPU is getting quite a workout because the vid card temps are the highest I've seen in any of the games I play, but I still think the CPU is the bottleneck. The X1950XT is a decent bit more powerful than those other two GPU's, but gave negligible improvement in frame rates for me; I got frames in the mid 30's on TBD in perfect mode at 1280x1024 with all 3 cards. I get 40 avg now using the beta dll's, but those took out the seafoam for ATI cards, which probably explains the improvement.

Guess I could check by rolling the CPU back to stock clocks and seeing how bad the FPS declines.

Another funny thing I noticed, running my vid card using the 2D clocks (500MHz core, 600MHz ram) gave only 3FPS drop (37 vs 34) in the avg FPS on TBD with all other settings the same. The 3D clocks for my card are 621MHz core and 900MHz ram. The GPU also ran 10 degrees C cooler with the lower clocks (73 vs 63C). So a 20% increase in core and 50% increase in ram frequencies and an increase in voltage from 1.2-1.45volts only gave &lt;10% increase in FPS. The max and min FPS values were 81 and 7 in both cases too.

Patriot_Act
03-05-2008, 11:57 AM
I just ran the Black Death track.

Got a score of 119-84-158-14

broken down it means............
119 current fps
84 average
158 peak
14 lowest.

Started fps counter 5 seconds into the track.
Ended counter about 5 seconds before the end.

Quad core 64 FX-74
4 gigs DDR2 RAM
XP Pro X64 Edition
2X 7800GT SLI
1920x1200 res "perfect" mode.

How does that score stack up?


P.A.

coolcody77
03-05-2008, 12:47 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Urufu_Shinjiro:
I've since gotten a 24" lcd and running 1920x1200 hasn't changed much, still getting about 32 avg on BD track but you would think that online you would get a little more since BD is supposed to be really tough on the system but once I'm up and flying I get around 28-35 fps flying in skies of valore and others and this is with no mods enabled. </div></BLOCKQUOTE>

Same here. I believe I'm pretty much CPU limited as everyone else here states. One way I know is that if I run in 1920x1200 versus 1680x1050 or 1440x900 I see almost no difference in framerate (My framerate tends to be low during battle in SP and in MP). Also if you look at IL-2's CPU usage, it's always using 100% of a core at any given time... plus my performance with an 8800GTX is nowhere near what it should be. I remember thinking about that ever since I ran benchmarks on perfect mode - my results are below. HOWEVER, I do find that turning some other details down besides landscape can help improve FPS by lowering CPU usage. I leave view distance at High, but crank down some others to medium or so.

----------------------
2008-02-26 18:34:29 - il2fb
Frames: 4448 - Time: 154000ms - Avg: 28.883 - Min: 11 - Max: 50 - The Black Death perfect (OpenGL, Water=3)
----------------------

Here's my system:
AMD Athlon 64 X2 5000+ dual-core (2.6GHz)
nVidia GeForce 8800GTX 768MB w/ 169.13 beta drivers
2GB of G.Skill DDRII PC6400 RAM (800MHz)
Windows XP Pro SP2

I created a thread in Help & support for everyone to check out and compare their performance results with everyone: IL-2 Performance & Benchmarks - Let's compare numbers (http://forums.ubi.com/eve/forums/a/tpc/f/49310655/m/8731029736)

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by roybaty:
I forgot who, but someone here has a quad core Intel and they said they were having excellent results with large missions. </div></BLOCKQUOTE>

Well a quad-core is going to do you no good over a dual core because IL-2 is not multithreaded - meaning it will only use one core. Actually you're likley to see a 15%-20% decrease in framerate when using a quad clocked at the same speed as a dual.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">
broken down it means............
119 current fps
84 average
158 peak
14 lowest.

Started fps counter 5 seconds into the track.
Ended counter about 5 seconds before the end.

Quad core 64 FX-74
4 gigs DDR2 RAM
XP Pro X64 Edition
2X 7800GTX SLI
1920x1200 res "perfect" mode. </div></BLOCKQUOTE>

84FPS average? No way, not on perfect mode. What did you crank everything else besides landscape down to low? There's no way a CPU only 400MHz faster than one like mine can give you that (your minimum is spot own however). Besides I thought an FX-74 was only a dual core? AMD Athlon 64 FX-74 & Quad FX Platform Review on H-Enthusiast (http://enthusiast.hardocp.com/article.html?art=MTIzMywyLCxoZW50aHVzaWFzdA==)
It says "As you can see from the CPU-Z specs, each FX-74 processor has two cores clocked at 3GHz, each core with 1MB of L2 cache."

willyvic
03-05-2008, 12:59 PM
Looks pretty good. Where in the track did you get the 14fps?


Mine as follows (Vsync locked to 60 fps):

2008-03-05 11:52:28 - il2fb
Frames: 9183 - Time: 155750ms - Avg: 58.959 - Min: 33 - Max: 60
Got the 33 momentarily on the initial convoy run.

Dual core E6700
XP pro
2 gigs ram
2x 512 7950gt's
1280x960

WV

TheFamilyMan
03-05-2008, 03:35 PM
My point for this topic is this:

Playing tracks does not really push your CPU hard since all the physics and AI info is NOT being computed; that number crunching was done when the track was recorded (and this is what crushes the CPU). When I run the BD track, or any of the other track I've tried, it looks great though I've never benched them. Running online always looks great too on any map that I've every tried. It's only the offline many on many missions (not tracks) that crush my CPU, simply because the FM, DM and AI computations of all those aircraft are working its butt off. I've tried lowering the res and IQ settings and found it only helps modestly, if at all.

IMHO, all I care about it how playable is the sim and how good it looks. I cannot say I really notice any fps much above the low 40's, nor do I worry about getting max fps. I'd much rather have max eye-candy with playable (40-60) fps than visa versa. For now I'll suffer with my 'underpowered' CPU. I'll upgrade when the 45nm cpus start kicking some serious ***.

Sergio_101
03-05-2008, 04:20 PM
Incorrect Ghalen.
Partly incorrect.
A FX-74 is adual core, but no one runs one FX-74 chip.

I have a similar rig with 8800Ultra SLI and get similar results
with a 96 average and 210 peak.

Again, quad core FX-74 is correct, I run dual core dual processors
as all socket 1207 motherboards are capable of dual CPU.

One difference is I use 8 gigs of RAM.
And he makes no mention of water cooling.
My rig is water cooled.

Those scores are not out of line at all for a AMD "quadfather" FX-74.

Intel is not up to the standard of two FX-74 cpu's.

Sergio

Xiolablu3
03-05-2008, 05:40 PM
Have you guys tried a reformat?

As soon as things start running sluggish for me, then I do a complete windows reinstall and fomat the HD.

It can be done on a Sunday while watching the football or a good film, just moving things on when it asks you too.

We could spend forever trying to find the poroblem, whereas with a reformat we know that you have a clean system/base to start with.

Xiolablu3
03-05-2008, 05:44 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by TheFamilyMan:
My point for this topic is this:

Playing tracks does not really push your CPU hard since all the physics and AI info is NOT being computed; that number crunching was done when the track was recorded (and this is what crushes the CPU). . </div></BLOCKQUOTE> http://forums.ubi.com/images/smilies/agreepost.gif

willyvic
03-05-2008, 08:35 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by coolcody77:

84FPS average? No way, not on perfect mode. What did you crank everything else besides landscape down to low? There's no way a CPU only 400MHz faster than one like mine can give you that (your minimum is spot own however). Besides I thought an FX-74 was only a dual core? </div></BLOCKQUOTE>



I don't find it hard to believe.

With Vsync off I get:
2008-03-05 19:39:29 - il2fb
Frames: 14223 - Time: 155328ms - Avg: 91.567 - Min: 39 - Max: 118

That's OpenGL/Perfect/Water 3

E6700
WinXP
2 gig ram
2x 512 7650gt
1280x960

WV.

Choctaw111
03-05-2008, 09:06 PM
Geez willyvic, your average and Min are higher than mine! Maybe I need to do a reformat http://forums.ubi.com/images/smilies/bigtears.gif

coolcody77
03-05-2008, 09:21 PM
Well im kind of confused then. Looking at this this ExtremeTech C2D vs AM2 article (http://www.extremetech.com/article2/0,1697,2014654,00.asp) an e6700 would perform 10-25% better than an AMD 5000+ processor (in the 3DMark 2006 CPU test 1 graph) that leaves a big gap from the 325% average FPS you're getting in your benchmark there, willy, and what mine show (Granted, I'm running 1920x1200 but if it's CPU limited that should not effect it). Same with Patriot_Act's average FPS -- sure an FX-74 sure outperforms an Athlon 5000+, but not up to 2x, according to benchmarks ???

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by TheFamilyMan:
My point for this topic is this:

Playing tracks does not really push your CPU hard since all the physics and AI info is NOT being computed; that number crunching was done when the track was recorded (and this is what crushes the CPU). When I run the BD track, or any of the other track I've tried, it looks great though I've never benched them. Running online always looks great too on any map that I've every tried. It's only the offline many on many missions (not tracks) that crush my CPU, simply because the FM, DM and AI computations of all those aircraft are working its butt off. I've tried lowering the res and IQ settings and found it only helps modestly, if at all. </div></BLOCKQUOTE>

I agree to that as well. While I can handle a playable FPS in TheBlackDeath, if I were to fly similarly stressful missions real-time in SP or MP the FPS hit would be much worse. Good theory, I didn't think about that before http://forums.ubi.com/groupee_common/emoticons/icon_smile.gif

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Sergio_101:
Incorrect Ghalen.
Partly incorrect.
A FX-74 is adual core, but no one runs one FX-74 chip.

I have a similar rig with 8800Ultra SLI and get similar results
with a 96 average and 210 peak.

Again, quad core FX-74 is correct, I run dual core dual processors
as all socket 1207 motherboards are capable of dual CPU. </div></BLOCKQUOTE>

First, quit calling me Ghalen. I didn't know who he was until LoneRanger explained him to me, and I still don't know why you think I'm him. From what I hear, that's quite an insult.

Second, it doesn't matter if you had 10 FX-74 chips, it won't improve performance unless the game is multithreaded. Same with dual or quad-core CPUs to my knowledge -- IL-2 will only use 100% of ONE core of ONE CPU (Or 50% of a dual-core CPU) because the game was designed back when CPUs only had one core, meaning it can't be slpit into "multiple parts". Some of you guys should read up on dual core / multi CPUs some time.

Patriot_Act
03-06-2008, 12:42 AM
There are a few variables in any game.
But i can say from experiance that Windows Vista
is between 15% and 20% slower than XP.
XP Pro X64 edition is that much faster than XP Pro 32bit.
Oh, Vista 64 is no faster than Vista 32 at this time.
Maybe in the future? Vista 64 does see all 4 gigs of RAM, vista 32 does not.

As to the game not using multiple CPUs, maybe.
But my Logitech leyboard has a CPU and RAM
usage LCD, it has four bars indicating the
usage of each CPU.
It rarely uses more than one CPU, but it does in fact
show the usage of two or more
in cluttered areas like the simulated cities.
Generaly CPU usage rarely exceeds 20% according to
the Logitech readout.
RAM usage maxes at about 22% on any map.

I got no reason to BS you.
Maybe the FX-74 CPU is simply configured better for this game?
I have seen in benchmarking articles where Brand "X" CPU is faster
in one game than brand "Y" and slower in another.

I had a FX-62 AM2 rig for a while and it was only a little slower
with the exact same video cards.

I have a new rig I just built with a Phenom quad core based on my old AM2 rig with
a 8800ULTRA card. So far it's slower on IL-2 http://forums.ubi.com/groupee_common/emoticons/icon_confused.gif
But it kicks butt on the newer games like UT-3 and Crysis.

P.A.

Chris0382
03-06-2008, 04:24 AM
another possible solution for faster play is to use Ultimate Defrag and move the IL-2 folder to the outer edges of the HD Platter were it spins faster allowing faster access to those files.

p1ngu666
03-06-2008, 05:04 AM
how does crossfire fair in il2?

Airmail109
03-06-2008, 05:47 AM
Ive got a quad core Oc'd to 3.0ghz, runs Il2 smooth as butter. With Vsync enabled fps are locked at a constant 60fps even over Berlin. Ive made some huge missions as well with hundreds of ground targets, AAA, ships and near to a hundred aircraft and the fps stays at 60 with no drops.

mbfRoy
03-06-2008, 06:01 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by coolcody77:
Well im kind of confused then. Looking at this this ExtremeTech C2D vs AM2 article (http://www.extremetech.com/article2/0,1697,2014654,00.asp) an e6700 would perform 10-25% better than an AMD 5000+ processor (in the 3DMark 2006 CPU test 1 graph) that leaves a big gap from the 325% average FPS you're getting in your benchmark there, willy, and what mine show (Granted, I'm running 1920x1200 but if it's CPU limited that should not effect it). Same with Patriot_Act's average FPS -- sure an FX-74 sure outperforms an Athlon 5000+, but not up to 2x, according to benchmarks ???
</div></BLOCKQUOTE>
Uhh... he's got 2x7950 gt video cards (I don't think there is a 7650?) which is kind of a monster setup for IL2, plus screen res obviously has an impact on fps if you are benchmarking the game using a track.

EDIT: By the way, this is my benchmark result using FRAPS in the black death track with vsync enabled:

Frames: 6555 - Time: 143000ms - Avg: 45.839 - Min: 13 - Max: 63

Core2Duo 6400 @2.13ghz
1x 8800GT 512mb @ 1680x1050
2gb RAM
XP 64bit

[OpenGL]
TexQual=3
TexMipFilter=2 &lt;-- 4x AF set in nvidia control panel
TexCompress=0
TexFlags.UseDither=1
TexFlags.UseAlpha=0
TexFlags.UseIndex=0
TexFlags.PolygonStipple=0
TexFlags.UseClampedSprites=0
TexFlags.DrawLandByTriangles=1
TexFlags.UseVertexArrays=1
TexFlags.DisableAPIExtensions=0
TexFlags.ARBMultitextureExt=1
TexFlags.TexEnvCombineExt=1
TexFlags.SecondaryColorExt=1
TexFlags.VertexArrayExt=1
TexFlags.ClipHintExt=1
TexFlags.UsePaletteExt=1
TexFlags.TexAnisotropicExt=1
TexFlags.TexCompressARBExt=0

TexFlags.TexEnvCombine4NV=1
TexFlags.TexEnvCombineDot3=1
TexFlags.DepthClampNV=1
TexFlags.SeparateSpecular=1
TexFlags.TextureShaderNV=1

HardwareShaders=1

Shadows=2
Specular=2
SpecularLight=2
DiffuseLight=2
DynamicalLights=1
MeshDetail=2
VisibilityDistance=3

Sky=1
Forest=3
LandShading=3
LandDetails=2

LandGeom=3
TexLarge=1
TexLandQual=3
TexLandLarge=1

VideoSetupId=17
Water=4
Effects=2
ForceShaders1x=0

Urufu_Shinjiro
03-06-2008, 08:49 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Chris0382:
another possible solution for faster play is to use Ultimate Defrag and move the IL-2 folder to the outer edges of the HD Platter were it spins faster allowing faster access to those files. </div></BLOCKQUOTE>

That might help with load times but would have 0 effect of FPS.

Patriot_act, I think your good performance might have to do with the onboard memory contoller. If I recall corerectly this game really responds well to tighter memory performance more than cpu mhz. The Phenom would have an advantage over the C2D in that area.

willyvic
03-06-2008, 10:24 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by mbfRoy:

Uhh... he's got 2x7950 gt video cards (I don't think there is a 7650?) which is kind of a monster setup for IL2, plus screen res obviously has an impact on fps if you are benchmarking the game using a track. </div></BLOCKQUOTE>


You are correct, that was typo on my part. I do have twin 7950gt's. And my res, 1280x960, is a bit lower. Filtering is set to 4x AA and 16x AF.

If it helps, here is my setup and conf.ini. There is nothing special about the system. Stock speed (2.66) and stock ram (corsair DDR2 x 800). I have never overclocked.


[window]
width=1280
height=960
ColourBits=32
DepthBits=24
StencilBits=8
ChangeScreenRes=1
FullScreen=1
DrawIfNotFocused=0
EnableResize=0
EnableClose=1
SaveAspect=1
Use3Renders=0

[game]
Arcade=0
HighGore=0
mapPadX=0.66875
mapPadY=-0.046875
viewSet=59
Intro=0
NoSubTitles=1
NoChatter=1
NoHudLog=0
NoLensFlare=1
iconTypes=2
eventlog=eventlog.lst
eventlogkeep=1
TypeClouds=0
3dgunners=0

[Render_OpenGL]
TexQual=3
TexMipFilter=3
TexCompress=0
TexFlags.UseDither=0
TexFlags.UseAlpha=0
TexFlags.UseIndex=0
TexFlags.PolygonStipple=0
TexFlags.UseClampedSprites=0
TexFlags.DrawLandByTriangles=1
TexFlags.UseVertexArrays=1
TexFlags.DisableAPIExtensions=0
TexFlags.ARBMultitextureExt=1
TexFlags.TexEnvCombineExt=1
TexFlags.SecondaryColorExt=1
TexFlags.VertexArrayExt=1
TexFlags.ClipHintExt=0
TexFlags.UsePaletteExt=0
TexFlags.TexAnisotropicExt=1
TexFlags.TexCompressARBExt=0

TexFlags.TexEnvCombine4NV=1
TexFlags.TexEnvCombineDot3=1
TexFlags.DepthClampNV=1
TexFlags.SeparateSpecular=1
TexFlags.TextureShaderNV=1

HardwareShaders=1

Shadows=2
Specular=2
SpecularLight=2
DiffuseLight=2
DynamicalLights=1
MeshDetail=2
VisibilityDistance=3

Sky=2
Forest=3
LandShading=3
LandDetails=2

LandGeom=2
TexLarge=1
TexLandQual=3
TexLandLarge=1

VideoSetupId=17
Water=3
Effects=1
ForceShaders1x=0

PolygonOffsetFactor=-0.15
PolygonOffsetUnits=-3.0

http://www.310thvcs.com/images/pics/WVSetupIL2_0208.jpg

WV.

coolcody77
03-06-2008, 11:37 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Urufu_Shinjiro:
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Chris0382:
another possible solution for faster play is to use Ultimate Defrag and move the IL-2 folder to the outer edges of the HD Platter were it spins faster allowing faster access to those files. </div></BLOCKQUOTE>

That might help with load times but would have 0 effect of FPS.

Patriot_act, I think your good performance might have to do with the onboard memory contoller. If I recall corerectly this game really responds well to tighter memory performance more than cpu mhz. The Phenom would have an advantage over the C2D in that area. </div></BLOCKQUOTE>

Ah that would explain the margin if CPU MHz couldn't. However I think this thread is going to be full of unproven rumors. For example: "the outer edges of the HD Platter where it spins faster allowing faster access to those files".

Not only would that have little effect on FPS in-game as you mentioned, but look at some results using HD Tach (http://www.simplisoftware.com/Public/index.php?request=HdTach) (click to enlarge):

http://img227.imageshack.us/img227/9449/hdtachox4.th.png (http://img227.imageshack.us/my.php?image=hdtachox4.png)

See how the read speed decreases towards the end of the disc? http://forums.ubi.com/images/smilies/16x16_smiley-surprised.gif So far 1) CPU MHz, 2) memory controllers, and 3) motherboards & chipsets could effect results. Do we have any benchmarks of different memory controllers showing improvements in games? Because I thought memory controllers didn't matter as much these days as long as they support the speed of ram you'd like to use. You know, now that I think about it, most of us here having mediocre performance are AM2 owners. Maybe IL-2 responds that much better to the Core 2 Duo implementation of memory controllers to make a huge performance difference? Urufu is showing more like what I'd expect; around 30 for an average... not 80 like some of you people. Willy I'll take a look at your conf.ini there and see if I have anything set different that would effect my results.

P.S. Benchmarks have shown there was negligible (Almost none) improvement from XP Pro 32-bit and XP Pro 64-bit; I'd expect to see the same with Vista. (Unless the app was created as a native 64-bit app, which IL-2 was not).

P.P.S. What's 'Dither' and 'Polygon Stripple'?

Patriot_Act
03-06-2008, 12:00 PM
I do not use a Phenom or AM2 in the rig that I posted.
It's a dual FX-74 ASUS L1N64-SLI WS work station mother board.
It is capable of quad SLI if I desire.

So far my AM2 rig is about half as fast and I am not sure why. http://forums.ubi.com/images/smilies/16x16_smiley-surprised.gif

Got to go.

P.A.

willyvic
03-06-2008, 12:02 PM
"Use Dither helps eliminate the ugly color "banding" when running the game in 16-bit color mode. Has no effect in 32-bit color mode. Reccomended on in 16-bit color mode and can be turned off if running with 32-bit color."


"Polygon Stipple is a way of emulating or faking translucent shadows when the stencil buffer is turned off. It does this by creating a "dotted" effect that looks translucent (not unlike the stubble of incoming facial hair). Turn this off for a small performance boost. If stencil buffer and polygon stipple are both turned off, you'll get completely black shadows which isn't much of a drawback."

Taken from here: http://forums.ubi.com/eve/forums/a/tpc/f/23110283/m/6651075613

WV.

mbfRoy
03-06-2008, 02:25 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by coolcody77:
[...]
You know, now that I think about it, most of us here having mediocre performance are AM2 owners. Maybe IL-2 responds that much better to the Core 2 Duo implementation of memory controllers to make a huge performance difference? Urufu is showing more like what I'd expect; around 30 for an average... not 80 like some of you people. Willy I'll take a look at your conf.ini there and see if I have anything set different that would effect my results.

P.S. Benchmarks have shown there was negligible (Almost none) improvement from XP Pro 32-bit and XP Pro 64-bit; I'd expect to see the same with Vista. (Unless the app was created as a native 64-bit app, which IL-2 was not).

P.P.S. What's 'Dither' and 'Polygon Stripple'? </div></BLOCKQUOTE>
The top three things that have an impact on FPS are: cpu, video card and memory speed (hard disk too if it needs to be accessed). Then it's a matter of having good drivers and a "clean" OS. Out of curiosity, did you change anything in the nvidia control panel? There's a setting, "Threaded optimization", that needs to be turned off for two-core CPUs to run properly.

coolcody77
03-06-2008, 03:27 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by mbfRoy:
The top three things that have an impact on FPS are: cpu, video card and memory speed (hard disk too if it needs to be accessed). Then it's a matter of having good drivers and a "clean" OS. Out of curiosity, did you change anything in the nvidia control panel? There's a setting, "Threaded optimization", that needs to be turned off for two-core CPUs to run properly. </div></BLOCKQUOTE>

That's what I'd think -- that's what leads me to believe IL-2 responding in such performance boosts to memory controllers is more untrue speculation. It may help, but it shouldn't be a major contributing factor. And the only things I have changed from default in the nvidia control panel are the Vertical Synch set to OFF, and Threaded Optimization also to OFF.

I can't help but think my 5000+ should pull more than a 28-30FPS average on TheBlackDeath from look at others' results; I've got a fairly fresh Windows install, all chipset and other drivers installed, hard disks defragmented, etc. Even with an e6700 which does perform significantly better than a 5000+, it doesn't explain an over 200% boost. I'm not saying anyones results weren't true; I'm saying that even though CPU affects the game greatly, there's got to be another factor or reason behind such different results.

mbfRoy
03-06-2008, 03:40 PM
Maybe it's the videocard. I had better min and average framerates with a 7600GS in The Black Death... but not better FPS overall (when playing). Most likely the game is better suited for that series of cards, plus two 7950 means a lot of power, definitely more than a single 8800GTX for this game IMO

Inadaze
03-06-2008, 04:53 PM
CoolCody, something is definitely up with your frame rates. I'm running an Opeteron 185 which is a dual core at 2600mhz, 2 gig corair 3200 ddr, and a 768 8800GTX OC. My BD frames with 16xAA 16XAF, at 1680 x 1050 are -

Min - 19, Max - 85, Avg - 47.400


That's with clouds-2, water-2, effects-1 and land geometry-3.

coolcody77
03-06-2008, 05:39 PM
I posted my system specs on page 1:

AMD Athlon 64 X2 5000+ dual-core (2.6GHz)
nVidia GeForce 8800GTX 768MB w/ 169.13 beta drivers
2GB of G.Skill DDRII PC6400 RAM (800MHz)
Gigabyte GA-M55SLI-S4 motherboard
Windows XP Pro SP2

Inadaze, are your results running Perfect landscape mode? My FPS really increases if I disable perfect and reset it to excellent, but man does water look horrible at that detail level.

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by mbfRoy:
Maybe it's the videocard. I had better min and average framerates with a 7600GS in The Black Death... but not better FPS overall (when playing). Most likely the game is better suited for that series of cards, plus two 7950 means a lot of power, definitely more than a single 8800GTX for this game IMO </div></BLOCKQUOTE>

Maybe better suited for an older game like IL-2, but i just checked some benchmarks and the 8800GTX blows away even two 7950's in SLi. That's one of the main motivations when I bought this card, it literally outperformed anything else even in SLi at the time. Only ATi's newest card or two GTS's in SLi can compete.

mbfRoy
03-06-2008, 06:02 PM
Yeah I'm talking about this game in particular

Inadaze
03-06-2008, 06:10 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by coolcody77:


Inadaze, are your results running Perfect landscape mode? My FPS really increases if I disable perfect and reset it to excellent, but man does water look horrible at that detail level. </div></BLOCKQUOTE>

Yup, perfect mode, water 2, 4.09 patch with the increased view distance enabled.

furyan65
03-06-2008, 08:28 PM
Maybe its driver related?

willyvic
03-07-2008, 01:32 AM
About the 7950s
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">

Maybe better suited for an older game like IL-2, but i just checked some benchmarks and the 8800GTX blows away even two 7950's in SLi. That's one of the main motivations when I bought this card, it literally outperformed anything else even in SLi at the time. Only ATi's newest card or two GTS's in SLi can compete. </div></BLOCKQUOTE>

I don't know about "blows away" but I do have to agree with you on one point. While my twins run IL2 great, they do have some difficulty keeping up with a few of the more modern games at higher resolutions. IE; COD4 and Crysis, to name a couple. And while Crysis is a rig killer extraordinare, I thought COD4 would be a piece of cake. I was wrong.


WV.

Patriot_Act
03-10-2008, 03:21 AM
I just ran the Black Death track.

Got a score of 119-77-190-29

broken down it means............
119 current fps
77 average
190 peak
29 lowest.

Started fps counter 5 seconds into the track.
Ended counter about 5 seconds before the end.

Quad core 64 FX-74
4 gigs DDR2 RAM
XP Pro X64 Edition
2X 8800GT SLI
1920x1200 res "perfect" mode.

How does that score stack up?

Threw you guys a curve ball.
I tested two Evega 8800GT cards.
Lost 10 points on the BD test!

Go figure. Tweeked to the gills except for over clocking.
Still slower than the two 7800GTX's

Peak and lowest scores went up a tad.

3D Mark-06 scores rose from a best of 9,493 with the 7800s
to 14,221 with the 8800s.... Some late first person shooter games
improved a lot.
Now I know why my AM2 machine with A 8800 ultra
IS SLOWER. the 7800 series cards are simply as good
or better than later cards for this old game.

P.A.

TgD Thunderbolt56
03-10-2008, 07:32 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by mbfRoy:
...plus two 7950 means a lot of power, definitely more than a single 8800GTX for this game IMO </div></BLOCKQUOTE>


Emphatically NOT TRUE! Even if the pair of 7950's both have 512 frame buffer, they won't have as much raw horsepower as a single 8800GTX. It may take a bit of tweaking in the NV control Panel and/or in the IL2 conf.ini, but a single 8800GTX will be faster under ALL circumstances provided all other factors are the same.

Remember, just because each card may have 512, that doesn't mean the bus will think you have 1gb of VRAM. Additionally, there is no substitute for stream processors which the GTX has in spades over the 7950's.

I've run a pair of 7900GT 512's and a single 8800GTS 512 in the same rig I have now and the 8800GTS is easily faster, both in benchmarks and in frame rates in games (I've compared IL2, UT3 and COD4).

The 79xx series of cards (especially the ones with 512) are great cards for IL2, but they aren't faster than a single 8800 (GT, GTS, GTX) no matter how badly we want them to be...even in sli http://forums.ubi.com/groupee_common/emoticons/icon_wink.gif

Having said that, I will go on record as saying that once you're in the realm of 79xx sli or higher (i.e. current series 8800's), you won't see much difference in image quality or FPS by just upgrading your GPU. I went from a P4 Northwood 3.4 oc'd to 3.6 to a C2D E6700 and almost doubled my FPS across the board....at STOCK SPEEDS! Once I took my CPU from 2.66 up to 3.2 my bottleneck became my GPU so I upgraded to a single 8800GTS 512 which again has increased my FPS while allowing really high levels of AA and AF at high resolutions.

To answer the original poster's question...it depends. The CPU/GPU spiral mirrors the economic wage/price spiral and it depends on what you're doing with your rig and where you are with it right now respective to CPU/GPU.

Stiletto-
03-10-2008, 11:02 PM
I have had my ATI 3870X2 for about a month now, and love it, basically same as sli, two cards on one board (though when they enable 2 of these to work, basically 4 graphics card, it might be awesome, if not ridiculouos). I can pretty much run maxed out graphics with large amounts of FSAA, but if I have an overcast mission with lots of clouds, I believe the aliasing of the cloud layers bogs my system down. Still, I would agree with the people who said the CPU makes more of a difference, when I can tell an occasional slow down (that isn't the clouds), it definatley is from the processing department trying to draw abundant 3d models on the screen, or from 20 ships blasting their AA guns away. Got a quadcore here, most of the time it's like butter though with max graphics, I should really run the blackdeath track with frames on shouldn't I?

Choctaw111
03-11-2008, 11:04 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Aimail101:
Ive got a quad core Oc'd to 3.0ghz, runs Il2 smooth as butter. With Vsync enabled fps are locked at a constant 60fps even over Berlin. Ive made some huge missions as well with hundreds of ground targets, AAA, ships and near to a hundred aircraft and the fps stays at 60 with no drops. </div></BLOCKQUOTE>

I would like to see the bottom portion of your conf.ini and what screen res you are running at. What options are you running on your NVidia control panel (assuming your running NVidia cards). I should be getting performance like that. What driver are you using?