Triplescreening

[Wayne Douglas]

hi Guys,

I have a new triplescreen setup, with 3 24" monitors runs with a triplehead2go unit. The graphics card i have is geforce gtx460 and im not get great fps when playing Race on online. If i change graphics card will it make a big difference to fps as ony 40-50 online and a bit flickery when around cars or is triplehead2go affecting it. Can anyone recomment a better card?

Cheers

 

[glight]

I recently built a triple screen gaming rig and I ended up building a $1500 PC to power the thing.
Here's the thread:http://www.racedepartment.com/sim-racing-hardware/52132-gaming-on-3-monitors.html
I run an i5 2500k at 4.4GHz with 2x6950 2GB

What I found out was that I needed 2 higher end graphics cards to get the fps I wanted but at the same time 3x1080p requires a high amount of VRAM. 1.5GB is the minimum required to play modern games at that resolution so you have two options: The GTX580 in SLI (1.5GB) or the 6950 in Crossfire (2GB). Two 6870s or two 460s might have been enough GPU power but neither of these cards come with more than 1GB memory.

If you are playing older games/sims you might be good with a 460 sli setup since they will not need as much VRAM. Just be aware that current games will not be a possibility without turning the resolution WAY WAY down.

5760x1080=6.22megapixels
4800x900=4.32megapixels
3840x768=2.94megapixels
1920x1080=2.07megapixels

 

[Aaron Freind]

Im getting 80fps solid on rfactor, 60+/- on iracing etc, running 5040x1050, using 2x GTX285 1gb in SLI, with everything MAXED out. No overclocking. I went for the I7 2600K, on an Asus Sabertooth P67, and 8gb ram. I had alot of issues with the Nvidia 266 drivers, but the 270's are rockets again.

 

[Paul Hamwa]

If you are playing older games/sims you might be good with a 460 sli setup since they will not need as much VRAM. Just be aware that current games will not be a possibility without turning the resolution WAY WAY down.

By "current games" do you mean games like F1 2010 and Dirt2?

 

[glight]

Yes. Before I upgraded I was trying to play F1 2010 with a single 6950 2GB and I could get 24-30 fps as long as I kept the resolution at 5020x900 and settings on low. I believe this to be unacceptable because the image quality is nowhere near what it should be and the framerates were barely playable.

 

[Paul Hamwa]

Yes. Before I upgraded I was trying to play F1 2010 with a single 6950 2GB and I could get 24-30 fps as long as I kept the resolution at 5020x900 and settings on low. I believe this to be unacceptable because the image quality is nowhere near what it should be and the framerates were barely playable.

That's what I thought as those games are notoriously greedy resource hogs. Your 6950 would tear through games like rFactor, Netkar and RaceOn I'm imagining...

 

[Jim Cole]

As stated before, regardless of the game if you are pushing 3 monitors, you are going to need some strong graphics system to be able to handle all the pixels that need to be drawn. Using a single monitor would be a different story.

 

[glight]

To be truly sure I think someone needs to do some benchmarking with a 6950 1GB vs. 6950 2GB vs. 6870 vs. 6950 2GB CF. That would be an interesting article.

Without any numbers, I just have my own experience to go by which leads me to agree with Jim on this one. You're going to need more than one card to pull off eyefinity properly (at least 6870 CF). The only card that has a chance of doing single card eyefinity is the 6950 2GB, the 6870 might be able to pull off older games but I wouldn't trust its' 1GB memory to be able to go above Low detail settings.

 

[otta56]

Hello,

I have been able to benchmark rFactor using two GTX580 3GB cards in sli and 2 AMD 6970 cards in crossfire. Currently using three 24 inch IPS 60hz monitors @ 5988X1200.

The most demanding track so far has been Targa Florio which uses about 1.7 GB of frame buffer with a 30 car field. Frame rates have generally been higher with the NVidia setup varying from 45 FPS to 93 FPS depending on the section of the course. The AMD setup has been running from 27 FPS to about 60 FPS on the same track with similar AA and AF settings.

Some observations:

Both setups produce stuttering in corners regardless of AA and AF settings. This seems to be a coding issue in the sim, particularly when dynamic shadows and shadow blurring are enabled in the rFactor menus although it never disappears entirely no matter what settings are used. The stuttering is more apparent in the AMD setup

rFactor appears to perform best with vsync enabled. Both setups show significant tearing on 60hz IPS monitors. The .plr "Max Framerate=-xx" does not eliminate the tearing on its own. This can probably only be eliminated with 120hz monitors. In the meantime, vsync with either setup eliminates the tearing completely but at some cost in the fluid sensation of speed unlocked framerates would give.

With the NVidia setup three identical DVI cables are a must. I had constant blanking of one monitor due to a dvi cable mismatch.

Careful tweaking of settings will give very similar quality images in both setups. The settings are quite different between AMD and NVidia drivers however. For instance, FSAA in the AMD setup gives a better quality image for rfactor than the NVidia combination of 16xCSAA and Transparency 4xSGSSAA which caused a large performance hit in FPS.

Forcing AFR2 in the NVidia setup gives slightly better framerates (1-2 FPS) and slightly less stuttering than AFR1.

Overclocking the video cards did not increase the frame rates or smoothness of rfactor in either setup. It does significantly increase power consumption, heat and fan noise however. The lesson here was to buy the highest clocked version available of either card and run it at the manufacturer's set speeds.

Pushing three 1920 X 1200 monitors will require more that 1.5 GB memory per card on whichever setup is used. This means two 2 GB 6970s or two 3 GB Nvidia cards. The NVidias are more likely to run into memory constraints if a 1.5 GB card is used.

Temperatures for both cards were about the same 73C on the hottest core in a dual card setup. One card will always have two screens connected so the GPU will be about 5C hotter on that card regardless.

Peripheral lag was not a significant factor on the cpus I tested, Intel's 2500k and 2600k. The 2500k is just as fast as the 2600k for rFactor. NetKar Pro did not show any significant speed differences either. This may be because rFactor 1 is not multi-threaded. I don't know about NetKar's code.

 

[Jim Cole]

A little note about the vsync setting. What this does is nice, but you might also try and force triple buffer at the same time. Change graphics settings to where you get a constant 60fps or more before changing to vsync and triple buffer, once there, make the changes and be amazed at the difference in smoothness. The third buffer really makes a difference.

Vsync does have it's limitations though, as it will look at performance and chop the frame rates in half if your fps drops below your monitors refresh rates. This will cause the effects you noticed with things not looking quite smooth. As mentioned above, keep your frame rates over your refresh rate to be sure this works smoothly.