Tri-Monitor Gaming

rekijitsu

Distinguished
Jul 14, 2008
34
0
18,540
Hi all,

Theoretically, if there are 3x monitors together, that means 3x more pixels to work with. Does this mean if I wanted to use 3x Monitors for gaming, and I wanted similar framerates to 1x GTX 480 with 1 monitor, then I would need 3x GTX 480s to get the same result? Or would 2x GTX 480s still be improved performance regardless of how many pixels are needing displayed?

I know you need more GPUs for tri-monitor, but will the performance degrade from 1x card to 1x monitor if using 2x cards for 3x monitors?

My thinking is even with more pixels, we could get away with using 2 cards for 3 monitors, but it might degrade performance we would normally see with 1 card to 1 monitor. Anyone have experience with this or ideas as to how much the wider resolution would impact performance?

From what I hear the GTX 480 has impressive scaling in 2x SLI, but hits the CPU bottleneck at 3x. I am hoping to get this running for Final Fantasy XIV, which is a very intense game, running right now at about the highest settings but with Ambient Occulsion off. Would 1 extra card do the trick for 3x monitors or not? My thinking is even if it is all pixels, the textures must already be in VRAM, and that 3 monitors doesn't necessarily mean 3 cards to achieve the same FPS you had with 1 card on 1 monitor. I could be wrong, please let me know what you all know!
 

skolpo

Distinguished
Sep 20, 2009
695
0
19,060
It all comes down to the resolution of the monitors, but it'll hardly take three cards for three monitors to perform as well as one card with one monitor. A triple monitor setup will be generally 40-50% slower than a single monitor setup using the same GPU. So the ratio is more like 1:1 = 2:3. In some games though, you do see roughly a 60-70% performance hit.

I've been searching for some benchmarks about the performance difference and this is what I was able to find:

http://www.desktopforums.com/thread230679.html

An important thing to point out is that the GTX 480 with a single monitor setup is already overkilling most games out there, meaning that it's hitting extremely high framerates. Adding two more monitors will not show significantly noticeable visual difference. However for nVidia, you will need two cards to run a triple monitor setup. If you're more interested in a single-card setup, using an HD5870 or HD5970 with ATi's Eyefinity is a better option.
 

CRdloader

Distinguished
Nov 8, 2009
13
0
18,510
I'd disagree with the above poster from personal experience. Running furmark on 4 screens at once i found the frame rate was just about 1/4 as when i ran it only on one screen. I normally use a 2 monitor set up and the second monitor just displays temps and such since mot games are still only single screen compatible. when there is no gaming on the second screen and just a temp monitor software, FPS is barley effected compared to single. If you plan on gaming on 3 monitors and want no performance hit you probably want 3 times the power under the hood (gfx wise). The card is rendering 3 times as many pixles. Now some say you only need 50-70% power on screen one and 3. This may be true in FPS because most of the gunfire and action is in front. If tons of action gets on the side monitors you will need the extra power. If you quickly more to keep most action and such on the center screen you should be ok with a little less then 3x the power. If action is coming form all directions you will need more power. If you truly want each screen working the same as your single does now, you need 2x the power. To bad more games are not dual/tri screen compatible. i would love to play gta4 in tri screen.
-cody
 

skolpo

Distinguished
Sep 20, 2009
695
0
19,060



Furmark is not a great depiction of real-world use. I don't see how you can use that as a measurement tool for everything else out there. As shown from benchmarks before, games like Torchlight, Dirt 2, and Batman: AA, which all utilize tri-monitor setups very well, have shown only a 50% performance hit. And even for first-person shooters, each screen is still utilized to a great extent and not simply focused primarily on the center screen (other than the gun).

And GTA 4 is Eyefinity/Tri-monitor compatible...
 

dalta centauri

Distinguished
Apr 1, 2010
885
0
19,010

They released a patch? I remember seeing GTA4 having serious issues with eyefinity.
 

dalta centauri

Distinguished
Apr 1, 2010
885
0
19,010


I tried with my 22" and 2x19" once, was a terrible experiance for me due to how the textures looked, stuttering, and camera angles.
Wouldn't say it's compatible.


Mind you I dimmed down the settings considerably for my 5770.
 

dalta centauri

Distinguished
Apr 1, 2010
885
0
19,010

Same can be said with a 5850...
I would go and replace 5850 with a 5970 but then...there's another problem with gta4.
It was badly ported for the pc, and although it can handle the resolution there are a series of problems.
 

skolpo

Distinguished
Sep 20, 2009
695
0
19,060
Doesn't change the fact that it's compatible, and plenty of people have ran the game satisfactorily with Eyefinity. Just because your certain setup doesn't run the game as smooth doesn't mean it's incompatible. If I had an 8600GT and tried to run Physx in Mirror's Edge, the game will be insanely choppy and unplayable. Does that mean the game is incompatible with Physx? Of course not.
 

dalta centauri

Distinguished
Apr 1, 2010
885
0
19,010

You just compared a game stamped with physx on it when gta4 doesn't have eyefinity or multi-monitor ready on it.
So the point is?
 

skolpo

Distinguished
Sep 20, 2009
695
0
19,060

"Just because your certain setup doesn't run the game as smooth doesn't mean it's incompatible."
 

skolpo

Distinguished
Sep 20, 2009
695
0
19,060
So you're telling me all the people that run GTA4 happily in Eyefinity are just delusional since, according to you, it should not be compatible? ATi clearly do not support Physx since it's an nVidia-proprietary implementation. I used that Physx comparison simply because it's a parallel example to your argument, and to explain to you that defining "compatibility" doesn't work like that. Your belief of "if it can't run on my system, it shouldn't be considered compatible for everyone else's" is broken. Am I missing something here?

Anyway, this has swayed way off-topic.