GTX 690 vs SLI 680 vs Xfire 7970

[Shmister]

I'm trying to upgrade my graphics card and have been stuck with a hard decision and can't decide on which card/cards to pick up. I've read many reviews and looked at charts for all them and still can't decide. I'm stuck between a GTX 690 (which is essentially two 680's in 1 card isn't it?) or Dual GTX 680. Even though i prefer Nvidia , i have been considering getting 2 7970's in crossfire. I would be mostly using them for gaming hours on end I have dual monitors if that is any factor as well. My case is huge so i can surely fit a 690 card. I'm just stuck and don't know which card to choose, so any input or user reviews on these cards is greatly appreciated. Or should i even just wait for 7xx series? Thanks in advance . Cheers

 

[hero1]

I'm trying to upgrade my graphics card and have been stuck with a hard decision and can't decide on which card/cards to pick up. I've read many reviews and looked at charts for all them and still can't decide. I'm stuck between a GTX 690 (which is essentially two 680's in 1 card isn't it?) or Dual GTX 680. Even though i prefer Nvidia , i have been considering getting 2 7970's in crossfire. I would be mostly using them for gaming hours on end I have dual monitors if that is any factor as well. My case is huge so i can surely fit a 690 card. I'm just stuck and don't know which card to choose, so any input or user reviews on these cards is greatly appreciated. Or should i even just wait for 7xx series? Thanks in advance . Cheers

The best option would be to go for either GTX 680 SLI or 7970 CFX. The 7970 is better option for price and performance and they are cheap atm. But if you're into NVidia then GTX 680 is better than a single GTX 690, heck you can even go for GTX 670 and still beat GTX 690. So up to you. Also, you might want to add a 3rd monitor if you are going to game on CFX or SLI since only CF supports 2 monitors but it sucks with the bezel pretty much covering the middle. So 3rd monitor a must if you want to do surround/eyefinity and CF 7970 is the best you have to go with since it will be same price as GTX 670 SLI but cheaper thn GTX 680 or GTx 690. Cheers!

 

[devilofdeaths]

if you can get a gtx titan thay are the fastest single gpu card.a little slower than a 690 but if the game doesn't support SLI then it beats everything else http://www.newegg.com/Product/Product.aspx?Item=N82E16814130897

 

[DarkSable]

Don't get a 680 - they're only 5% faster than a 670, but they're 25-30% more expensive.

Also, you're only going to be gaming on a single monitor, right? A single card will be MORE than enough - don't blow you money all at once; buy a single 670, and upgrade a few years down the line.

 

[17seconds]

Several review sites are starting to use frame latencies as a way of testing video card performance. There have always been subjective reviews that have noted that SLI runs smoother than Crossfire, but with the new frame latency testing they are able to measure it. SLI is the way to go with a roomy case, and if you can get a pair of 680's or 670's in SLI, that will be more stable and deliver more fluid gameplay than a Crossfire setup. Oh, and you'll also get to play GPU-accelerated PhysX games as a bonus.
http://www.pcper.com/reviews/Graphics-Cards/NVIDIA-GeForce-GTX-TITAN-Performance-Review-and-Frame-Rating-Update/Frame-Rat

 

[computeguy]

Don't get a 680 - they're only 5% faster than a 670, but they're 25-30% more expensive.

Also, you're only going to be gaming on a single monitor, right? A single card will be MORE than enough - don't blow you money all at once; buy a single 670, and upgrade a few years down the line.

That's not true. A single GTX 680 cannot max out several current games.

 

[devilofdeaths]

That's not true. A single GTX 680 cannot max out several current games.

ya even a 690 can't even max out crysis 3 with 45+fps

 

[CarolKarine]

crysis has finally redeemed itself after the "2" flop that was a console port.

oh, right. I'd say 7970's cause they're cheaper outright, and generally win in performance. (and cause red is a way better color than green)

 

[iam2thecrowe]

Its not always about benchmarks and FPS
from a techreport article:

"Interestingly, in this page of Ryan's Titan review, he reproduces images that suggest a potentially serious problem with AMD's CrossFire multi-GPU scheme. Presumably due to sync issues between the two GPUs, only tiny slices of some frames, a few pixels tall, are displayed on screen. The value of ever having rendered these frames that aren't really shown to the user is extremely questionable, yet they show up in benchmark results, inflating FPS averages and the like.

That's, you know, not good."
source: http://techreport.com/blog/24415/a [...] e-and-more

http://www.pcper.com/reviews/Graph [...] /Frame-Rat < this is the page refered to as "this page of Ryan's Titan review? make up your own mind but i would never get crossfire again, having had a poor experience with xfire 6850's, stuttering, poor game support or late game support from amd, i would never go crossfire again until i know its problems are fixed. there are plenty of others that share this experience and many other articles detailing crossfire problems with frame latency and the like. Having said that sli is not without its problems but they are minimal in comparison.

 

[17seconds]

Its not always about benchmarks and FPS
from a techreport article:

"Interestingly, in this page of Ryan's Titan review, he reproduces images that suggest a potentially serious problem with AMD's CrossFire multi-GPU scheme. Presumably due to sync issues between the two GPUs, only tiny slices of some frames, a few pixels tall, are displayed on screen. The value of ever having rendered these frames that aren't really shown to the user is extremely questionable, yet they show up in benchmark results, inflating FPS averages and the like.

That's, you know, not good."
source: http://techreport.com/blog/24415/a [...] e-and-more

http://www.pcper.com/reviews/Graph [...] /Frame-Rat < this is the page refered to as "this page of Ryan's Titan review? make up your own mind but i would never get crossfire again, having had a poor experience with xfire 6850's, stuttering, poor game support or late game support from amd, i would never go crossfire again until i know its problems are fixed. there are plenty of others that share this experience and many other articles detailing crossfire problems with frame latency and the like. Having said that sli is not without its problems but they are minimal in comparison.

It's a good thing to see more review sites use frame latency as a way of judging GPU performance. Tom's even implemented it in it's latest Titan review. There is more to the story than just pure FPS, and according to that PC Perspective review, even FPS might not be all we make it out to be. "Runt" frames, as they call them, are only fractions of a complete frame, but are counted by FRAPS as a whole frame in their FPS counts. According to their tests, AMD setups may be inflating their FPS numbers artificially by essentially counting a 0.01 frame as whole frame. This is nothing new to anyone who has paid attention to past discussions of multi-GPU smoothness, except that now there is an emerging framework for measurement.

 

[Shmister]

Thanks for all the replies! Since most of you and from reviews i've read say crossfire has some issues, I'm going to stick with Nvidia. Out of all the cards available, would Dual 680s be the best. I heard the Titan isn't worth the money and the 690 is basically just two 680s. I have only two monitors at the moment , but i might add a third of need be. I dont no if i will play games on eyefinity mainly just on single monitor with others to use for web browsing while playing. So all in all would the SLI 680's give me the best performance out of the available cards nvidia has to offer?

 

[bystander]

source: http://techreport.com/blog/24415/a [...] e-and-more

http://www.pcper.com/reviews/Graphics-Cards/NVIDIA-GeForce-GTX-TITAN-Performance-Review-and-Frame-Rating-Update/Frame-Rat

Wow! I had no idea. That makes me feel better about some of these new benchmarks showing much higher CF performance than SLI.

 

[hero1]

It's a good thing to see more review sites use frame latency as a way of judging GPU performance. Tom's even implemented it in it's latest Titan review. There is more to the story than just pure FPS, and according to that PC Perspective review, even FPS might not be all we make it out to be. "Runt" frames, as they call them, are only fractions of a complete frame, but are counted by FRAPS as a whole frame in their FPS counts. According to their tests, AMD setups may be inflating their FPS numbers artificially by essentially counting a 0.01 frame as whole frame. This is nothing new to anyone who has paid attention to past discussions of multi-GPU smoothness, except that now there is an emerging framework for measurement.

I call BS on this. So many reviewers do the latency test and no one has made such claim. It's nearly impossible for a GPU to inflate FPS, let alone raise a number that is 99.99% away from 1 to 1. Get it, it not possible to move 0.01 to 1 and thus make it a whole frame. That's bs. I have had Nvidia cards through my life, from high school to February 2013 and recently switched to AMD. I have had zero issues with it apart from the beta driver 13.2 shutting down my second GPU. I have never had a pleasant experience with SLI even though it scales better. I had red screens, black screen, jagged lines from using SLI in game and browsing around. The only card that came close to doing that from AMD was an XFX which I think it was just a bad card. I would give both camps a good vote but to make a claim that FPS are inflated that's an outright lie, especially when it comes to making a 0.01 into 1.

 

[hero1]

Thanks for all the replies! Since most of you and from reviews i've read say crossfire has some issues, I'm going to stick with Nvidia. Out of all the cards available, would Dual 680s be the best. I heard the Titan isn't worth the money and the 690 is basically just two 680s. I have only two monitors at the moment , but i might add a third of need be. I dont no if i will play games on eyefinity mainly just on single monitor with others to use for web browsing while playing. So all in all would the SLI 680's give me the best performance out of the available cards nvidia has to offer?

You make a good choice based on what you need and want. I would go with GTX 680 since they are great cards, no doubt. Just make sure you grab one from either eVGA or ASUS since they have a great customer support and I know from dealing with both of them. I would personally go with eVGA due to their lifetime warranty. 2x GTX 680 is a beast in the making and you will never look back, unless you want to get Titan and SLI that.

On the other hand, if you don't plan on gaming on all 3 monitors you can stick to just a single GTX 680 and save the money. It iwll not be tasked that heavily when only one monitor is doing graphic intense work and the others handle browsing. So, in your case I would stick to a single GTX 680 and if you need another one you can always add it down the road.

 

[bystander]

I call BS on this. So many reviewers do the latency test and no one has made such claim. It's nearly impossible for a GPU to inflate FPS, let alone raise a number that is 99.99% away from 1 to 1. Get it, it not possible to move 0.01 to 1 and thus make it a whole frame. That's bs. I have had Nvidia cards through my life, from high school to February 2013 and recently switched to AMD. I have had zero issues with it apart from the beta driver 13.2 shutting down my second GPU. I have never had a pleasant experience with SLI even though it scales better. I had red screens, black screen, jagged lines from using SLI in game and browsing around. The only card that came close to doing that from AMD was an XFX which I think it was just a bad card. I would give both camps a good vote but to make a claim that FPS are inflated that's an outright lie, especially when it comes to making a 0.01 into 1.

If you read through THG's review that did these tests, they were testing in the middle of process, and not looking at the end result. That site has a new piece of hardware that takes what it receives as output to the monitor and analysis the end process. This is the first such set of tests on the net, so it won't be the same. However, even when using the old process, AMD's cards have been showing signs of a problem, we just didn't know how big of a problem it is.

As far as the inflation of numbers, it may not be a large amount. It may be very small, depending on how much adjusting would be required to create evenly spaced frames. It is possible that if you get the cards displaying of frames evenly spaced, they may not need a lot of delays after that, which would result in a small impact on frame rate. It really depends on how much variance there is between frame rendering.

 

[17seconds]

Thanks for all the replies! Since most of you and from reviews i've read say crossfire has some issues, I'm going to stick with Nvidia. Out of all the cards available, would Dual 680s be the best. I heard the Titan isn't worth the money and the 690 is basically just two 680s. I have only two monitors at the moment , but i might add a third of need be. I dont no if i will play games on eyefinity mainly just on single monitor with others to use for web browsing while playing. So all in all would the SLI 680's give me the best performance out of the available cards nvidia has to offer?

Well, sticking with the theme of frame latencies, that would be the exact reason to go for the Titan versus a dual-GPU setup, despite the higher FPS numbers. Personally, I would go for the least expensive option. If you were going for a Titan, however, I would recommend waiting until prices come down and/or custom models are released.

http://www.hardwarecanucks.com/forum/hardware-canucks-reviews/59785-nvidia-geforce-gtx-titan-6gb-performance-review-19.html

If the GTX TITAN’s price is exclusively associated with raw frame rates, it seems like NVIDIA may have completely misjudged its positioning somewhere along the line. $1000 for a product that simply strides in the vast open space between the GTX 680 and GTX 690 yet is priced identically to the latter seems absolutely preposterous at first glance. But it isn’t. NVIDIA’s strategy here is spot-on and there are two reason for that: frame times and consistency.

When it comes to a consistent gameplay experience, the TITAN is head, shoulders and knees above its compatriots. The low standard deviation shown between frame times is a fraction of what other solutions offer, which in turn leads to significantly smoother onscreen performance. For example, the GTX 690 exhibits many of the issues normally associated with dual GPU cards like minor stuttering and outright frame hesitation, a situation the TITAN deftly avoids by virtue of being a single core solution. Gamers won’t have to wait for the latest SLI profile for optimal new game support either. So while the GTX 690 may outclass the TITAN in terms of raw framerate production, the GK110 clearly holds the edge this key area.

Even though it is a single GPU solution, AMD’s HD 7970 GHz just isn’t in the same league as the GeForce TITAN. Its comparative frame times border on abysmal and no amount of overclocking will improve that fact. Had AMD addressed the stuttering in key games like Far Cry 3 and Hitman Absolution (both of which are Gaming Evolved titles I may add) we could be having a very different conversation right now, especially considering the GHz Edition’s strong framerates.

 

[hero1]

If you read through THG's review that did these tests, they were testing in the middle of process, and not looking at the end result. That site has a new piece of hardware that takes what it receives as output to the monitor and analysis the end process. This is the first such set of tests on the net, so it won't be the same. However, even when using the old process, AMD's cards have been showing signs of a problem, we just didn't know how big of a problem it is.

As far as the inflation of numbers, it may not be a large amount. It may be very small, depending on how much adjusting would be required to create evenly spaced frames. It is possible that if you get the cards displaying of frames evenly spaced, they may not need a lot of delays after that, which would result in a small impact on frame rate. It really depends on how much variance there is between frame rendering.

That's why I said it is a lie to report that a 0.01 has been recorded as 1 and pin it on the GPU. I think it's great that they have such tools now, I read both articles, and I hope this pushes AMD to give us a better driver and reduce the latency experienced throughout the tests (even here at Toms).

Cheers to your reply mate!

 

[Shmister]

Thanks Everyone. I'm just going to get 2 GTX 680's, since the titan is $1000 atm. 1200wat PSU is plenty to support 2 680's right?

 

[17seconds]

Thanks Everyone. I'm just going to get 2 GTX 680's, since the titan is $1000 atm. 1200wat PSU is plenty to support 2 680's right?

Believe it or not, 750 watts is the recommended minimum wattage for SLI 680's.
http://www.guru3d.com/articles_pages/geforce_gtx_680_sli_review,4.html

 

[Shmister]

Oh wow thats less than i thought. Has any of you had the chance to use the Titan card? Just curious on how much better it actually is.

 

[Mousemonkey]

I call BS on this. So many reviewers do the latency test and no one has made such claim. It's nearly impossible for a GPU to inflate FPS, let alone raise a number that is 99.99% away from 1 to 1. Get it, it not possible to move 0.01 to 1 and thus make it a whole frame. That's bs. I have had Nvidia cards through my life, from high school to February 2013 and recently switched to AMD. I have had zero issues with it apart from the beta driver 13.2 shutting down my second GPU. I have never had a pleasant experience with SLI even though it scales better. I had red screens, black screen, jagged lines from using SLI in game and browsing around. The only card that came close to doing that from AMD was an XFX which I think it was just a bad card. I would give both camps a good vote but to make a claim that FPS are inflated that's an outright lie, especially when it comes to making a 0.01 into 1.

Considering the rest of your post that is just funny! You've had zero issues apart from one, so you've had one issue then, not zero but one. Perhaps AMD count things the same cack handed way.

 

[iam2thecrowe]

I call BS on this. So many reviewers do the latency test and no one has made such claim. It's nearly impossible for a GPU to inflate FPS, let alone raise a number that is 99.99% away from 1 to 1. Get it, it not possible to move 0.01 to 1 and thus make it a whole frame. That's bs. I have had Nvidia cards through my life, from high school to February 2013 and recently switched to AMD. I have had zero issues with it apart from the beta driver 13.2 shutting down my second GPU. I have never had a pleasant experience with SLI even though it scales better. I had red screens, black screen, jagged lines from using SLI in game and browsing around. The only card that came close to doing that from AMD was an XFX which I think it was just a bad card. I would give both camps a good vote but to make a claim that FPS are inflated that's an outright lie, especially when it comes to making a 0.01 into 1.

this fps inflation makes sense in my experience. i haven't had any ati/amd issues until i got my 6850. The first ATI card i had was an 8500le, then a 9800 pro - both great cards with no problems, then nvidia 7600gt and then 8800gts, then i got my amd 6850. I was never really happy with it, it never seemed to put out it's claimed performance and always seemed stuttery. when coming to the end of its useful life a couple months ago i added a second for crossfire, biggest mistake ever. Crossfire worked well in 50% of games, and it only worked well when fps was at 60 fps. When things got tough thats when you notice the microstuttering. I wasnt happy with the fps cap solution using radeonpro, tried all kinds of drivers, windows reinstall etc. so to anyone that has a 7970 or two running 1080p and gets 60fps all the time, that is why you dont notice the stuttering. you would more likely notice it when your fps drops to 40 or less fps.

 

[Shmister]

Best answer selected by Shmister.