Catleap 2560x1440 and GTX 670 performance

[cmdrdredd]

My catleap arrived today. So far it's been a great upgrade. Better black levels, better colors, brighter picture, crisper text. On and on... There was a thread asking about 2560x1440 gaming with a GTX 670 and other questions occationally about what is playable at that resolution. So I did some benchmarks with a single card and SLI. Overclocked to the levels in my sig, 1242 max boost clock and 6772 memory clock. As you can see below many games simply don't do well with high AA levels or any AA at all. Others are very playable with 4x on a single card. I am sure FXAA would be fine if you decided to use that instead. The fps hit with FXAA is very small. Now on to the numbers.

Note: Metro 2033 and Batman AA have unreliable minimum FPS results. All detail settings maxed out in DX11 except as noted.

Uniengine Heaven Benchmark.

Single Card
8xAA - 7.1fps min/29fps avg/ 68.9fps max
4xAA - 20.8fps min/36.9fps avg/88.6fps max
0xAA - 27.7fps min/43.6fps avg/104.9fps max
SLI
8xAA - 16.7fps min/64.5fps avg/163fps max
4xAA - 34.1fps min/83.4fps avg/210.6fps max
0xAA - 32.3fps min/101fps avg/269.3fps max

Metro 2033 built in benchmark. Average of 3 runs. Advanced Physics and depth of field is off

Single Card
MSAA - 7.6fps min/27.67fps avg/78.65fps max
AAA - 9.82fps min/36.67fps avg/95.91fps max
SLI
MSAA - 13.27fps min/56fps avg/247.89fps max
AAA - 14.23fps min/75.33fps avg/277.24fps max

Batman: Arkham City Physx turned off

Single Card
8xAA - 19fps min/48fps avg/68fps max
4xAA - 31fps min/66fps avg/191fps max
0xAA - 40fps min/90fps avg/113fps max
4xAA + Physx High - 22fps min/58fps avg/75fps max

SLI
8xAA - 21fps min/100fps avg/144fps max
4xAA - 11fps min/132fps avg/186fps max
0xAA - 16fps min/156fps avg/237fps max
4xAA + Physx High - 23fps min/79fps avg/138fps max

Crysis average of 3 runs. DX10

Single Card
8xAA - 2.35fps min/28.32fps avg/42.71fps max
4xAA - 19.78fps min/28.81fps avg/44/26fps max
0xAA - 23.04fps min/35/90fps avg/ 51.33fps max

SLI
8xAA - 23.08fps min/62.84fps avg/88.68fps max
4xAA - 22.75fps min/63.93fps avg/92.25fps max
0xAA - 57.48fps min/83.75fps avg/113/93fps max

Crysis 2 Adrenaline Bench Central park benchmark. Average of 3 runs. FXAA only.

Single Card
15.1fps min/41fps avg

SLI
18fps min/84.8fps avg

Battlefield 3 quick test. No timedemo but only observed FPS in a short game. FPS can change with different maps.

Single Card
MSAA 4x - 30-40fps
MSAA 0x - 40-55fps

SLI
MSAA 4x - 60-100fps
MSAA 0x - 100-150fps

 

[hokies83]

So you like it huh? it is good to see someone taking advice on these monitors there just insane no to have for that price.

 

[Fire&Blood]

Wow, such huge hits over 1200p. I'm all for 16:10 but they get hit even harder at 1600p.

Do you think AA is as much of a must have at 1440p as it is at 1080p?

Which monitor did you use prior to getting the Catleap?

Did you observe any ghosting and how do you feel about 60Hz on it overall?

I know the 7950 I hope to get soon won't do better, probably worse in some of the games you tested but even stepping it up to top single GPU's and clocking them exceptionally high wouldn't change much. Looks like we'll be well into 2013 before a a<$500 GPU shows that can handle 1440 even remotely close as single GPU's handle 1200p today.

From single gpu perspective, do you think you would game less because of the performance hits? Ultimately, was the upgrade worth it?

 

[cmdrdredd]

Thanks for sharing.

Do you think AA is as much of a must have at 1440p as it is at 1080p?

Which monitor did you use prior to getting the Catleap?

Did you observe any ghosting and how do you feel about 60Hz on it overall?

On some games no. Let me give you an example. Playing Skyrim where it's slow paced and has ornate arcitecture and design, the aliasing is noticeable because you're always looking at a table, items laying around, characters up close, wood buildings, stone etc.. When I play Battlefield 3 I am too busy to look at the edges of objects and observe shifting pixels. I haven't done a whole lot with it yet, but I fired up a few titles and looked around, most of them are not so bad without AA. Being SLI equipped I can usually throw AA on with not much problem. Batman doesn't show much aliasing to my eyes at this resolution, but you can use FXAA and have very little performance hit and minor blurring.

My previous monitor was a BenQ 2400WD and I don't notice any severe ghosting. I do see a very slight bit in fast FPS titles (BF3, Crysis). I wouldn't call it distracting though. It's very subjective so it's hard to say whether or not someone else would be sensitive to it.

 

[toyota]

thats some magical SLI scaling you got there as you more than doubled your framerate and in some cases tripled it...

 

[SheHateMe]

Did you buy a perfect pixel or...? How much did you pay for it? Also, which seller?

 

[Larnz]

thats some magical SLI scaling you got there as you more than doubled your framerate and in some cases tripled it...

Yeah in nearly all of them the sli scaling is greater than 100% which seems a bit...

 

[cmdrdredd]

Yeah in nearly all of them the sli scaling is greater than 100% which seems a bit...

This is not 1080p guys...

Sli scales better with higher resolution. There are times a single card is unplayable with AA and you throw a second card in and it jumps dramatically in fps.

 

[BFG10K]

This is not 1080p guys...

Sli scales better with higher resolution. There are times a single card is unplayable with AA and you throw a second card in and it jumps dramatically in fps.

How can two cards be more than twice as fast as one card?

For example, if one GTX670 is powerful enough to generate 50FPS, how can a second add another 60FPS or 70FPS, more than doubling the score?

Unless you're running more than two GPUs, >100% scaling is not possible outside of occasional benchmarking noise. To get results consistently more than double like you are means there's something wrong with your numbers.

 

[cmdrdredd]

What do I gain by lieing? I am typing exactly what I saw. No more, no less.

Guru3d also lies? 2560x1600 34fps single card to 64fps sli. Pretty much in line with my results. Except they are using an older x58 setup with 3.8ghz cpu. Varying from title to title and AA method used.

 

[PBat51]

Thanks for the info.... I'm seriously debating about getting a catleap, but I realize I don't have the gpu to really pull it off. My single 560 is just not going to cut it.

Back to looking at 120Hz monitors

 

[cmdrdredd]

Honestly, there is a lot to love about the IPS display aside from gaming. You can still run 1920x1080 and let your video card scale it. Text looks funny but games are fine.

I think they key thing I got from this is a single GTX 670 is enough for 1440p provided you can compromise on settings. Maxing out every game won't happen and you probably can't run AA in every title either. This is what I suspected and I checked for myself. Some titles are very good even with a single card, but overall no.

 

[toyota]

What do I gain by lieing? I am typing exactly what I saw. No more, no less.

Guru3d also lies? 2560x1600 34fps single card to 64fps sli. Pretty much in line with my results. Except they are using an older x58 setup with 3.8ghz cpu. Varying from title to title and AA method used.

http://www.guru3d.com/imageview.php?image=39348

http://www.guru3d.com/imageview.php?image=39356

they are not more than doubling their framerate like you are. heck you have some cases where you are tripling your max framerate which makes no sense.

 

[cmdrdredd]

It makes sense when you look at the game tanking with a single card with maxed settings at 2560x1440. It simply can't do it all the time.

Since you think I'm a liar...

Single Card

SLI

SLI

Single Card

 

[PBat51]

Honestly, there is a lot to love about the IPS display aside from gaming. You can still run 1920x1080 and let your video card scale it. Text looks funny but games are fine.

I think they key thing I got from this is a single GTX 670 is enough for 1440p provided you can compromise on settings. Maxing out every game won't happen and you probably can't run AA in every title either. This is what I suspected and I checked for myself. Some titles are very good even with a single card, but overall no.

Yea but whats the fun of that. It has me thinking of what would be an affordable crossfire solution that could get the job done.

 

[cmdrdredd]

Yea but whats the fun of that. It has me thinking of what would be an affordable crossfire solution that could get the job done.

Affordable? I wouldn't know...I would say a 7950 would likely be a better solution than a 670 if you are considering a single card provided you overclock the 7950. I'd wait for some 660ti SLI results to compare to 7870/7850 and then decide.

 

[imaheadcase]

What i find funny about this whole catleap situation is that people could buy monitors way better than ones you can get in the states for years and years but people recently started doing it because same panels as apple uses. hehe


Thanks cmdrdredd for the benchmarks. I have to wait till next year to upgrade stuff, if no other good monitors come out will get one of these. Given that monitors evolve so slowly for PC might still be a good bargin.

 

[Arkaign]

What i find funny about this whole catleap situation is that people could buy monitors way better than ones you can get in the states for years and years but people recently started doing it because same panels as apple uses. hehe


Thanks cmdrdredd for the benchmarks. I have to wait till next year to upgrade stuff, if no other good monitors come out will get one of these. Given that monitors evolve so slowly for PC might still be a good bargin.

I think people got excited not because it had anything to do with Apple at all, but rather that for once you could get a quality 2560 IPS for under $300, which definitely was NOT the case for years and years. Best one could do were the 1200P IPS panels for around the same price in 24". And yes, I have a relative who spends 2 months out of the year mostly in Seoul, and is rabid about whatever good deals/cool gadgets come down the pipe.

 

[BFG10K]

What do I gain by lieing? I am typing exactly what I saw. No more, no less.

Nobody said you were lying, just that your results cannot be explained by normal means. No website consistently shows >100% scaling from two cards like you’re showing.

That’s not possible unless there’s something wrong with your single card scores, or you’re running more than 2 GPUs (are you?).

Guru3d also lies? 2560x1600 34fps single card to 64fps sli. Pretty much in line with my results. Except they are using an older x58 setup with 3.8ghz cpu. Varying from title to title and AA method used.

It's not in line with your results at all, and neither are other reviews. That’s the point.

Not one of Guru3D’s average FPS results managed to double the performance, yet you’re frequently getting more than double.

It makes sense when you look at the game tanking with a single card with maxed settings at 2560x1440. It simply can't do it all the time.

It doesn't make sense at all. Again, if one card can get 50 FPS, how can the same card suddenly get more when it's paired with another? The second card doesn’t make the first one go faster, the aggregate performance just goes up.

It’s like getting more than double the performance by going from one CPU core to two. It’s not possible unless something else has changed.

Assuming completely ideal scaling, you could get 100% performance in very rare situations, perhaps 105% because of benchmarking noise. But to consistently get more than that (like 2.26x in one case), yeah, there’s something wrong with the results.

I notice your system is running overclocked, so it’s quite possible it’s actually reporting incorrect benchmarking numbers.

 

[cmdrdredd]

So now you can't overclock or all results are false. Yeah.. Bs man... BS.

 

[BFG10K]

So now you can't overclock or all results are false. Yeah.. Bs man... BS.

No, I&#8217;m saying if I got results like you, I&#8217;d be checking to see what went wrong. And if I was running my hardware outside spec, that&#8217;s the first place I&#8217;d be checking.

Again, if a card scores 20 MPix/sec, two of them can&#8217;t consistently score more than 40 MPix/sec unless something else has changed. This is elementary.

As for unstable CPU overclocks, they can cause benchmarks to run incorrectly between runs, in which case the numbers being reported back don&#8217;t always reflect reality. You won&#8217;t see that as visible artifacting or hitching, but in the form of much subtler differences.

 

[3DVagabond]

You would get Scaling greater than 100% if the card you did the single tests with had issues. Then the 2nd, higher performing card would increase the scores more than 2x. Might be worth checking the single card results against the other GPU.

 

[imaheadcase]

I think people got excited not because it had anything to do with Apple at all, but rather that for once you could get a quality 2560 IPS for under $300, which definitely was NOT the case for years and years. Best one could do were the 1200P IPS panels for around the same price in 24". And yes, I have a relative who spends 2 months out of the year mostly in Seoul, and is rabid about whatever good deals/cool gadgets come down the pipe.

Which is in direct result of Apple wanting so much and flooded the market with cheaper ones. Which have been out for years.

 

[dmoney1980]

OP- thanks for the info! There are a lot of people here looking at this monitor and debating about the performance hit. This should provide a good baseline for people out there with a 670 or comparable card.

Question on your BF3 benchmark - was that on a 64 player server?

 

[cmdrdredd]

No, I just hopped into the first game that came up. Turned on the fps monitor and kept an eye on it for like 5 min. Hardly scientific and probably not a good test. It is what it is though.

Again I never claimed my results would reflect everyone else's. I only report what I observe and nothing more. I thing a single card is ok if you want to reduce some settings. Sli is really good at this resolution. If I had AMD cards I would run the same tests again, but you can find many review sites to gather info from on that end.