Poor performance

[Amraam]

Hey all,

 

I installed DCS again recently (I had the russian version, now have the UK version also :)).

 

I have tried loads of different settings, but I'm only getting ~15-30FPS tops :( OK, I'm running at 1920x1200 res, but surely my system should cope?

 

E6750 @ 3.2Ghz

6GB PC2-6400 RAM

GTX260 216SP x2 in SLi

 

I'm running standard DCS, no mods, patched to latest version.#

 

Any ideas?

 

Thanks!!

[EtherealN]

First of all: discount your graphics cards. They don't really matter. In the sense of detailed simulation you should consider this a military grade simluator, not a regular gaming product that competes with graphical resources.

 

What are your water settings and view distance settings? (These can have massive impact.)

Also, at which circumstances do you get those 15-30 FPS? This matters greatly to identifying the cause-

[Amraam]

First of all: discount your graphics cards. They don't really matter. In the sense of detailed simulation you should consider this a military grade simluator, not a regular gaming product that competes with graphical resources.

 

What are your water settings and view distance settings? (These can have massive impact.)

Also, at which circumstances do you get those 15-30 FPS? This matters greatly to identifying the cause-

 

Well tbh, I've been too busy at work to thouroughly test the problem, but it happened when I was sitting on the ramp during the training missions ( iwas just watching them to refresh my memory).

 

Yeah I realise it's more of a CPU dependant game. I'll tinker on. Would it be worth getting a newer CPU or OC'ng a bit more?

 

Oh, this was also with Windows 7 which obviously is supported yet as it's only RC1 atm.

[EtherealN]

On OCing, this is where I really would need to know the circumstances. There are some circumstances where my system (see sig) will get 15-30 FPS, but most of the time I'll have 100+.

 

If you have a license for FRAPS one awesome way to check it out would be to fly a mission, then save the track file. Then run the track file while using FRAPS to save FPS data. Since the Tracks aren't actual "video" recordings but rather just recordings of raw input data for the simulation to compute this would allow some good comparisons to be done. If you don't have a FRAPS license this is a bit convoluted to chieve, though.

 

If you want to self-test and is comfortable in customizing your CPU OC, this is what I'd recommend:

 

1) Design a mission. At startup, have nothing at all. Then have a trigger zone that spawns a LOT of stuff. Fly to that trigger zone.

2) Play the mission. Record FPS as best as possible.

3) CHANGE your OC settings.

4) Repeat. If possible, repeat through playing track files, to ensure that you get as close as possible to the changed CPU freq being the only difference.

[Amraam]

On OCing, this is where I really would need to know the circumstances. There are some circumstances where my system (see sig) will get 15-30 FPS, but most of the time I'll have 100+.

 

If you have a license for FRAPS one awesome way to check it out would be to fly a mission, then save the track file. Then run the track file while using FRAPS to save FPS data. Since the Tracks aren't actual "video" recordings but rather just recordings of raw input data for the simulation to compute this would allow some good comparisons to be done. If you don't have a FRAPS license this is a bit convoluted to chieve, though.

 

If you want to self-test and is comfortable in customizing your CPU OC, this is what I'd recommend:

 

1) Design a mission. At startup, have nothing at all. Then have a trigger zone that spawns a LOT of stuff. Fly to that trigger zone.

2) Play the mission. Record FPS as best as possible.

3) CHANGE your OC settings.

4) Repeat. If possible, repeat through playing track files, to ensure that you get as close as possible to the changed CPU freq being the only difference.

 

Cool I'll try that mate, I've got a full version of FRAPS from my movie making days :)

 

TBH, I think Windows 7 is the problem...

 

Mioght be time to dual boot XP back on.

[159th_Viper]

Navigate to your Eagle Dynamics\Ka-50\BlackShark\data\scripts folder and open the options.lua with Notepad++ and find the following entries:

 

 

["heatBlr"] = 1,

["scenes"] = "high",

["water"] = 1,

["fullScreen"] = true,

["visibRange"] = "High",

 

 

 

["effects"] = 3,

["lights"] = 2,

["haze"] = 0,

["terrPrld"] = "40",

 

 

Edit and Save to the following:

 

 

["heatBlr"] = 1,

["scenes"] = "high",

["water"] = 0,

["fullScreen"] = true,

["visibRange"] = "High",

 

 

 

["effects"] = 2,

["lights"] = 1,

["haze"] = 0,

["terrPrld"] = "40",

 

See if that makes a difference.

[Feuerfalke]

First of all: discount your graphics cards. They don't really matter.

 

I have to disagree. If you are running the game on 1920x1200x32bit or even higher resolutions and maybe even have FSAA or AF enabled, you're CPU won't help you much if you have a low-end-graphics-card.

 

I know on these forums nobody wants to hear that, but when I upgraded from my overclocked 8800gts to my gtx285 my FPS more than trippled. (running with 8xAF and 8SxFSAA)

[Amraam]

I have to disagree. If you are running the game on 1920x1200x32bit or even higher resolutions and maybe even have FSAA or AF enabled, you're CPU won't help you much if you have a low-end-graphics-card.

 

I know on these forums nobody wants to hear that, but when I upgraded from my overclocked 8800gts to my gtx285 my FPS more than trippled. (running with 8xAF and 8SxFSAA)

 

I was thinking that also. Multi GPU systems really shine when using high resolutions.

 

I'll have a proper play about tonight when I get home :)

[159th_Viper]

 

TBH, I think Windows 7 is the problem...

 

Might be time to dual boot XP back on.

 

I won't if I were you :)

 

Performance FAR superior on Win7 as opposed to XP ;)

[Amraam]

I won't if I were you :)

 

Performance FAR superior on Win7 as opposed to XP ;)

 

Hehe scratch that idea then :D

 

I do like Windows 7, I may even pay for it! haha

 

Much smoother than Vista IMO :)

[Feuerfalke]

I was thinking that also. Multi GPU systems really shine when using high resolutions.

 

I'll have a proper play about tonight when I get home :)

 

True. Especially if you use driver-quality settings. :)

 

If you are using FSAA and HeatBlur, you will especially notice this, but also with large forests and buildings fading in and out under higher quality settings.

[EtherealN]

Feuerfalke, this would be true for obscene settings in the case of his system. Even in the case of a single, non-SLI, card he has much higher GPU performance than I have. I usually use 8x AA and 4x AF (I get higher FPS with 8x AF but at the cost of some artifacts). The fact that he gets lower performance in spite of having a better GPU (even when allowing for non-SLI operation) allows me to discount the GPUs from the problem.

 

But yes, reverting to XP is NOT a solution. Vista or W7 is the way to go, since those OSes have far superior dual-core handling.

Edited October 1, 2009 by EtherealN

[EtherealN]

Actually, upon review... The 260-216 actually has a lower texture fillrate than my 9800GTX+. Pixel fillrate is way better though. However, memory bandwith is far superior, raw computation power is way better and so on.

 

(Should be noted that I do sometimes overclock my 9800GTX+ quite drastically - by the gods it's a dream for that shit when it's got a dedicated 240mm fan cooling it - but not for the purpose of DCS, since I've never found any performance impact except when using supersampling and stuff like that.)

[Feuerfalke]

Feuerfalke, this would be true for obscene settings in the case of his system. Even in the case of a single, non-SLI, card he has much higher GPU performance than I have. I usually use 8x AA and 4x AF (I get higher FPS with 8x AF but at the cost of some artifacts). The fact that he gets lower performance in spite of having a better GPU (even when allowing for non-SLI operation) allows me to discount the GPUs from the problem.

 

But yes, reverting to XP is NOT a solution. Vista or W7 is the way to go, since those OSes have far superior dual-core handling.

 

Not exactly, the GTX260 has a newer chip than your graphics card, but it is the pendant to a 9600 GTX+. Even the 9800GT has twice as much shaders (128 as opposed to 64 of the GTX260) and your shaders are 400MHz faster (GTX260 1242MHz, 9800GT: 1688 MHz), just to name a few specifications.

Your card also uses more renderpipelines, thus scales a lot better with a faster CPU/RAM.

[Amraam]

Not exactly, the GTX260 has a newer chip than your graphics card, but it is the pendant to a 9600 GTX+. Even the 9800GT has twice as much shaders (128 as opposed to 64 of the GTX260) and your shaders are 400MHz faster (GTX260 1242MHz, 9800GT: 1688 MHz), just to name a few specifications.

Your card also uses more renderpipelines, thus scales a lot better with a faster CPU/RAM.

 

The GTX260 has 216 Shader processors?

[Feuerfalke]

The GTX260 has 216 Shader processors?

 

216 Steam Processors, sorry.

 

Are you using AF and FSAA anyway?

Edited October 1, 2009 by Feuerfalke

[EtherealN]

Hm. Very interesting points FF. I would like to object but, before that, I must acknowledge that your points are very good and I cannot refute them with absolute confidence.

 

I would say that clock isn't everything. This is the same phenomenon as in the old mid-90's PC vs Mac battle, where the PC consistently had a higher clock speed than the MACs but was just as consistently beaten in performance for many applications, simply through the x86 architecture and the Motorola/IBM PowerPC architectures being better at other things. A more recent example would be Xbox 360 vs PS3: there's no way to compare them clock-for-clock because their hardware (in spite of being VERY closely related) has drastically differing architecture suitable for different types of applications.

 

But that point made, I must say that I do not know nVidia shader architecture closely enough to judge if there's anything that offsets the clock speed diffs. My judgement would have to be based on generalized benchmarks where the 260-216 beats mine - though not on the texture fillrates.

 

For the OP though this should still be eminently testable if they have confidence in their hardware handling abilities: get RivaTuner or use the nVidia utilities to UNDERCLOCK your graphics card(s). If this has a near-linear FPS impact, the source of the problem will reveal itself.

[Amraam]

216 Steam Processors, sorry.

 

Are you using AF and FSAA anyway?

 

I can't remember to be honest, but knowing me, probably haha. However, it would also have been the first thing I would have turned off to see if it made a difference.

[EtherealN]

Amraam, do you use the nHancer utility?

[Amraam]

Amraam, do you use the nHancer utility?

 

I have in the past, usually I use Rivatuner, but I haven't looked for a Windows7 compatible version recently