PDA

View Full Version : Far Cry 2 - SLI seems to be broken


wazooda
11-01-08, 12:45 AM
Whelp.... after running the built-in benchmark, I can say that I am getting ZERO improvement with SLI over single card on my 280GTX setup.

Here are my results -

SLI - Enabled:

Settings: Demo(Ranch Small), 1600x1200 (60Hz), D3D10, Fixed Time Step(No), Disable Artificial Intelligence(No), Full Screen, Anti-Aliasing(None), VSync(No), Overall Quality(Ultra High), Vegetation(Very High), Shading(Ultra High), Terrain(Ultra High), Geometry(Ultra High), Post FX(High), Texture(Ultra High), Shadow(Ultra High), Ambient(High), Hdr(Yes), Bloom(Yes), Fire(Very High), Physics(Very High), RealTrees(Very High)

Average Framerate: 45.57
Max. Framerate: 68.46
Min. Framerate: 32.26


SLI - Disabled:

Same Settings

Average Framerate: 44.92
Max. Framerate: 67.43
Min. Framerate: 33.23

I'm running nHancer 2.4.2 and the 180.43 drivers on Vista 32 Ultimate with 4gigs of ram and a e6850 Duo Core OC'd to 3.6mhz.

WTF?!!?!

JasonPC
11-01-08, 01:27 AM
I don't have SLI, but the benchmark I saw with the 9800GX2 seemed to indicate SLI was not working. I asked in the GTX 2XX board whether or not SLI was working at all but I don't think anyone answered.

rap1d^
11-01-08, 03:09 AM
For me anyway I get little to no improvment using SLI in FarCry2. I have tried several older drivers with no sucess....:(
I have seen the issue mentioned on a few forms including Nvidia's SLI zone.

AthlonXP1800
11-01-08, 04:27 AM
I think nHancer is to blame for broken SLI, here was Far Cry 2 benchmarks with single and SLI GTX 280 with Core i7 showed SLI worked with the same 180.43 driver but they didnt used nHancer.

http://www.vr-zone.com/articles/-updated-x58-sli-test-on-crysis-warhead--far-cry-2/6149.html

g-a-m-e
11-01-08, 07:40 AM
http://img223.imageshack.us/img223/5462/slifarcryyh4.jpg (http://imageshack.us)
http://img223.imageshack.us/img223/slifarcryyh4.jpg/1/w600.png (http://g.imageshack.us/img223/slifarcryyh4.jpg/1/)

178.24 driver. I'm on XP though. Maybe that's what's different ?

wazooda
11-01-08, 04:25 PM
Yeah.... I saw the results over at vr-zone.... but I just don't get their results. I guess it could be because I'm running an E6850 @3.6ghz instead of an i7 @3.2.... but that doesn't make sense to me.

When I turn on the load balance graph, it shows sometimes completely filled.... but most of the time it is hardly filled at all. Thus, SLI is active.... just not doing much most of the time.

I also tried it with and without nHancer.... no change.

I'd like to see some forum members run the test with SLI on and off. *YOU* guys I trust... :D

Ninja Prime
11-01-08, 04:33 PM
Yeah.... I saw the results over at vr-zone.... but I just don't get their results. I guess it could be because I'm running an E6850 @3.6ghz instead of an i7 @3.2.... but that doesn't make sense to me.

When I turn on the load balance graph, it shows sometimes completely filled.... but most of the time it is hardly filled at all. Thus, SLI is active.... just not doing much most of the time.

I also tried it with and without nHancer.... no change.

I'd like to see some forum members run the test with SLI on and off. *YOU* guys I trust... :D

Probably you should ask ChrisRay over on the nvidia card forums, all he does is troubleshoot SLI problems all day. ;)

Bman212121
11-01-08, 10:15 PM
Whelp.... after running the built-in benchmark, I can say that I am getting ZERO improvement with SLI over single card on my 280GTX setup.

Here are my results -

SLI - Enabled:

Settings: Demo(Ranch Small), 1600x1200 (60Hz), D3D10, Fixed Time Step(No), Disable Artificial Intelligence(No), Full Screen, Anti-Aliasing(None), VSync(No), Overall Quality(Ultra High), Vegetation(Very High), Shading(Ultra High), Terrain(Ultra High), Geometry(Ultra High), Post FX(High), Texture(Ultra High), Shadow(Ultra High), Ambient(High), Hdr(Yes), Bloom(Yes), Fire(Very High), Physics(Very High), RealTrees(Very High)

Average Framerate: 45.57
Max. Framerate: 68.46
Min. Framerate: 32.26


SLI - Disabled:

Same Settings

Average Framerate: 44.92
Max. Framerate: 67.43
Min. Framerate: 33.23

I'm running nHancer 2.4.2 and the 180.43 drivers on Vista 32 Ultimate with 4gigs of ram and a e6850 Duo Core OC'd to 3.6mhz.

WTF?!!?!

I would bet you're CPU limited. You have more than enough graphics power at your disposal, but you're not using AA. I would bet if you turned on 4x AA you'd see the loading bars go up, and still have the same framerates. The only other issue is that the 512MB of vram on those cards will probably run out with too much AA, so your fps is going to tank. Another thing you can do is turn down the fire, physics, and real trees all down to low, and your framerates should go up if it's a cpu issue.

This game can use a Quadcore, and it uses a lot of processing power. I can fully load my SLI bar top to bottom with the rig in my sig, so I know SLI does work properly in this game. I'm using the 178.24 drivers, as the 180.43 caused me to BSOD. Also, I don't use nHancer either.

I'm going to make a couple of passes with SLI enabled and disabled, and I'll post my results. Using 1920 x 1200 DX10 4x AA / 16x AF Very High preset.

This is one 8800GTX stock, best run:

Total Frames: 1520, Total Time: 51.03s
Average Framerate: 29.79
Max. Framerate: 49.89 (Frame:0, 0.02s)
Min. Framerate: 23.65 (Frame:1204, 41.05s)

It uses about 50% of my CPU at times. I'm going to enable SLI and see what that shows.

Hmm CPU usage looks to be a little bit more, but it's still around 55 - 60% As for the results, here they are.
Peak value for CPU is around 70%.

Total Frames: 2339, Total Time: 51.01s
Average Framerate: 45.85
Max. Framerate: 60.07 (Frame:329, 6.28s)
Min. Framerate: 35.22 (Frame:1582, 34.53s)

About a 50% increase in average fps, the max was the only thing that didn't go up much. Let me do a run with lower settings now and see what that yields.

1600 x 1200 no AA. Without a doubt I'm seeing CPU limitations when I look at the fire. The framerates drop and so do the loading bars. You can see it a touch when I use my normal settings, but it is very noticeable when I turn them down. I was going to post a pic, but you can't see the SLI bars for some reason. Either way they are about half as filled when using higher res + AA.

Total Frames: 2585, Total Time: 51.01s
Average Framerate: 50.68
Max. Framerate: 78.41 (Frame:403, 6.57s)
Min. Framerate: 34.28 (Frame:1769, 34.49s)

This little article done by PCGH I think might also kind of show the cpu limits. http://www.pcgameshardware.com/aid,663817/Reviews/Far_Cry_2_GPU_and_CPU_benchmarks/?page=2

I'm guessing the i7 has some decent tweaks because I had to set everything on low to try to even get the minimums they were getting.

K007
11-01-08, 11:17 PM
I would bet you're CPU limited. You have more than enough graphics power at your disposal, but you're not using AA. I would bet if you turned on 4x AA you'd see the loading bars go up, and still have the same framerates. The only other issue is that the 512MB of vram on those cards will probably run out with too much AA, so your fps is going to tank. Another thing you can do is turn down the fire, physics, and real trees all down to low, and your framerates should go up if it's a cpu issue.

This game can use a Quadcore, and it uses a lot of processing power. I can fully load my SLI bar top to bottom with the rig in my sig, so I know SLI does work properly in this game. I'm using the 177.92 and 178.24 drivers, as the 180.43 caused me to BSOD. Also, I don't use nHancer either.

I'm going to make a couple of passes with SLI enabled and disabled, and I'll post my results. Using 1920 x 1200 DX10 4x AA / 16x AF Very High preset.

This is one 8800GTX stock, best run:

Total Frames: 1520, Total Time: 51.03s
Average Framerate: 29.79
Max. Framerate: 49.89 (Frame:0, 0.02s)
Min. Framerate: 23.65 (Frame:1204, 41.05s)

It uses about 50% of my CPU at times. I'm going to enable SLI and see what that shows.

Hmm CPU usage looks to be a little bit more, but it's still around 55 - 60% As for the results, here they are.

Total Frames: 2339, Total Time: 51.01s
Average Framerate: 45.85
Max. Framerate: 60.07 (Frame:329, 6.28s)
Min. Framerate: 35.22 (Frame:1582, 34.53s)

About a 50% increase in average fps, the max was the only thing that didn't go up much. Let me do a run with lower settings now and see what that yields.

1600 x 1200 no AA. Without a doubt I'm seeing CPU limitations when I look at the fire. The framerates drop and so do the loading bars. You can see it a touch when I use my normal settings, but it is very noticeable when I turn them down. I was going to post a pic, but you can't see the SLI bars for some reason. Either way they are about half as filled when using higher res + AA.

Total Frames: 2585, Total Time: 51.01s
Average Framerate: 50.68
Max. Framerate: 78.41 (Frame:403, 6.57s)
Min. Framerate: 34.28 (Frame:1769, 34.49s)

Setting real trees to low gives about the biggest gain which is about 3 avg - 5 fps max.

erm..he is on gtx280...so he is not limited by 512vram he has 1gb ><...

But yea try putting the AA up and see if that changes the results..it might go down yes..but the question is by how much..or if any at all.

Bman212121
11-01-08, 11:26 PM
erm..he is on gtx280...so he is not limited by 512vram he has 1gb ><...

But yea try putting the AA up and see if that changes the results..it might go down yes..but the question is by how much..or if any at all.

I was going by his sig. It says that he is using 9800GTX SLI there. I assumed he meant testing 9800GTX SLI versus one GTX 280. It would make more sense if he just bought 2 GTX 280's though.




I'm actually seeing weird things happen, when the fps drops, so does the cpu usage. That almost indicates a bug or that something can't thread right.

The geometry seems to be one of the killers for min fps. If I turn that from high to low in DX9 mode it can change the fps by 10 frames. Say 38 to 48 min fps which makes around a 25% difference.


DX9 mode again. My full settings Very high preset:

Total Frames: 2352, Total Time: 51.01s
Average Framerate: 46.10
Max. Framerate: 72.34 (Frame:365, 6.58s)
Min. Framerate: 31.32 (Frame:1586, 33.78s)

Slightly lower fps, but CPU usage seems to be going up in DX9 mode, which might indicate why it's slightly slower. I'm seeing about 10% more usage at peaks so around 80% total processor usage. 66% usage when the framerate dips to it's min. That is probably also 10% higher than DX10 mode.

Going to try with geometry on low and see what happens.

Total Frames: 2800, Total Time: 51.00s
Average Framerate: 54.90
Max. Framerate: 80.68 (Frame:400, 6.12s)
Min. Framerate: 39.39 (Frame:1915, 34.71s)

I'd say that is where the culprit is. 76% CPU usage when the framrate dips. The peak value is about the same.

So something with the geometry is causing SLI to slowdown for me, as I don't see as much scaling, and the cpu usage goes down. Try setting that to low in DX9 mode and see what you get.
EDIT: I wonder if that's what they tweaked in the newer driver...

Burnt_Ram
11-02-08, 11:26 AM
i have played around with the geometry setting, and it indeed does take load off the cpu. i gained around 5-7 fps. not sure what the geometry setting does exactly, but the game looks petty much the same to me on medium as it does on very high?

g-a-m-e
11-02-08, 11:54 AM
I think it has something to do with draw distance. The way objects pop up ?

Burnt_Ram
11-02-08, 11:56 AM
I think it has something to do with draw distance. The way objects pop up ?that makes sense. i've been waiting for tweakguides.com to explain the various settings. soon i think.