Gaming performance with a 4K Display - Get a mortgage :)

[GoodBytes]

Anandtech brings us gaming performance with a 4K display

 

Computer specs they used:

Intel Core i7-4770K @ 4.2 GHz, High Performance Mode
Corsair Vengeance Pro 2x8GB DDR3-2800 11-14-14
GIGABYTE Z87X-OC Force (PLX 8747 enabled)
2x GIGABYTE 1200W PSU
Windows 7 64-bit SP1
Drivers: GeForce 320.18 WHQL / Catalyst 13.6 Beta

 

 

 

 

 

Source: http://www.anandtech.com/show/7120/some-quick-gaming-numbers-at-4k-max-settings

[bartekxx12]

3 7950s are looking like an amazing value if you're going for pure performance. They're outperforming 2 Titans in SLI and are cheaper than 1... Extremely impressive

Once AMD releases that new driver which fixes the crossfire issue I am definitely getting a 2nd one for myself and a 1440 or 1600p monitor :P

[Shyxlol]

Looks like they didn't have enough 780s :/ would have liked to see the difference of those 3 gigs VRAM

[felixbearpig]

im suprised the single titan got 20fps on metro 2033 maxed out. it gets 25 on 1440p

[Vitalius]

What I'm surprised by is the way the Titans scaled. It's about an 85% (on average) increase per Titan on Dirt.

Note that this is comparing how much adding a Titan to the system increases the FPS compared to having 1 Titan in the system on it's own. So comparing 57.93 to each increase in FPS, for Dirt.

57.93 -> 100.62 -> 159.25 -> 201.87 for Dirt

XX% 74% 101% 74% Increase in FPS per Graphics Card added

And

20.62 -> 36.62 -> 55.29 -> 61.27 for Metro

XX% 78% 90% 29% Increase in FPS per Graphics Card added

(Metro is a graphic intensive game after all)

And

13.63 -> 27.4 -> 41.07 -> 57.78 for Sleeping Dogs

XX% 101% 100% 123% Increase in FPS per Graphics Card added

...

I thought SLI wasn't very efficient at scaling? This looks pretty efficient to me. :|

Of course, it's assuming you have no other bottleneck (like CPU).

[Woodypc]

The scaling is game dependent, on my 2 xfire rig I had about 99% on some games.

[Shyxlol]

The scaling is game dependent, on my 2 xfire rig I had about 99% on some games.

But I was also wondering that especially the fourth card still scaled so well, at least for 2 of those 3 games (but I'm not really sure about dirt, seeing how high the fps numbers are I don't like it in comparison). I always thought there was a bigger diminishing return on the number of cards.

[Vitalius]

But I was also wondering that especially the fourth card still scaled so well, at least for 2 of those 3 games (but I'm not really sure about dirt, seeing how high the fps numbers are I don't like it in comparison). I always thought there was a bigger diminishing return on the number of cards.

This basically.

I expected it to go like this:

90-100% -> 50-75% -> 25-50%.

But it was not only fairly consistent (with the exception of Metro, but that game likely is an exception), but the numbers were high at the same time (70-90's).

Essentially going more like this:

75-100% -> 70-100% -> 70-90%

That is awesome. While I understand it is very game dependent, it's still pretty cool that some games actually utilize all of the GPU's even at Quad SLI.

[kuddlesworth9419]

Well Dirt 3 can be played with pretty much anything on the chart at 4K. Everything else though you will need 4 way SLi Titans or 780's.

[Vitalius]

Well Dirt 3 can be played with pretty much anything on the chart at 4K. Everything else though you will need 4 way SLi Titans or 780's.

 

Well, honestly, what is "playable FPS" is also game dependent. For example, 20-25 FPS on Far Cry 3 can be easily considered playable to many people. Even ones who are sticklers for high FPS in games. I think this has to do with the engine the game runs on, but I'm not sure.

I consider 30+FPS playable for most games, but 60+FPS for certain ones.

Far Cry 3 is one of the exceptions to that where I consider 17+ FPS playable.

[Lukiose]

Wow, a 7950 outperforms the 680 due to the 3GB vram!

Excellent... and adding another 7950 is so cheap right now.

[danomicar]

This is just so unnecessary... Probably looks stunning but 1366x768 is plenty for me

[Nicktrance]

I'm surprised how well the Titans scale considering 4x sli isn't even officially supported.

 

This is just so unnecessary... Probably looks stunning but 1366x768 is plenty for me

 

No offence but have you tried at least a 1080p monitor? Many people say that until they try a better display.

[kuddlesworth9419]

I can't play may games properly unless it is 55+. Anything lower and my aim is off completely but also it just feels wrong to play at low frame rates which is one of the reasons why I don't play console games anymore.

[w_km]

Remember that these benchmarks are at MAX settings  (Not included in the title or description).  Turn down the AA and some other settings and you'll be getting 100% better numbers.

[Vitalius]

Remember that these benchmarks are at MAX settings  (Not included in the title or description).  Turn down the AA and some other settings and you'll be getting 100% better numbers.

I don't care what anyone says.

AA is completely unnecessary at 4K. Especially when it comes to PPI. If we can eventually get a 27-33" 4K monitor, I completely doubt that AA will make any kind of visible difference. :P

[Humbug]

No offence but have you tried at least a 1080p monitor? Many people say that until they try a better display.

I'm sure you're right; even for resolutions much higher than 1080p.

 

But I'm going to stick with 1080p for my next upgrade as well because I simply can't afford the constant upgrades and GPU muscle required for gaming at ultra high resolutions. I'd rather aim for 60 fps at 1080p and be happy being able to max out everything.

 

It's a lot about budget...

[Nicktrance]

I can't play may games properly unless it is 55+. Anything lower and my aim is off completely but also it just feels wrong to play at low frame rates which is one of the reasons why I don't play console games anymore.

 

Also I'm nowhere near as accurate with a controller as I am with a mouse.

[Nicktrance]

I don't care what anyone says.

AA is completely unnecessary at 4K. Especially when it comes to PPI. If we can eventually get a 27-33" 4K monitor, I completely doubt that AA will make any kind of visible difference. :P

 

I agree, for now. Then again people were saying the same thing about 1080p...

[Vitalius]

I agree, for now. Then again people were saying the same thing about 1080p...

Well considering how GPU crippling AA is, I think that the higher the resolution goes, the less AA we have to use. Overall, I think that's better picture quality regardless. :P

[kuddlesworth9419]

Yea that to I can't play any game even racing games properly with a controller. It has to be a steering wheel for racing games. With flying type mechanics I am fine with a mouse and keyboard but once again am terrible with a controller. I used to play on the PS3 quite a bit but could never actually get used to using any controller so I gave up and sold my PS3 to get better PC hardware.

[JawsOfLife]

I have no idea why they used SSAA for this test. That's absurd. That' rendering these games at 7680x4320 for 2x SSAA. Double these numbers and that's what you should expect for 4K performance.

[Inkz]

Anyone know why they did not test the 7970 or 7970 GE?

[That Norwegian Guy]

I have no idea why they used SSAA for this test. That's absurd. That' rendering these games at 7680x4320 for 2x SSAA. Double these numbers and that's what you should expect for 4K performance.

 

If they really used super sampling on a 4K monitor they should seriously consider seppuku. That's the most idiotic thing I've ever heard to ever get benchmarked, and all these results are 110% worthless.

[zeros]

Why 2x1200W PSU's? that seems overkill.