GTX 690 or GTX Titan?

[cmdrdredd]

The short answer is to stick with a 690 for now.

 

[Ferzerp]

In short:

avg is irrelevant

only standard deviation is relvevant

Well, you have to look at both. 50ms frames, exact, steady isn't preferable to something that is wobbling back and forth between 12 and 16 ms for example (which is why Tom's methodology is near useless)

 

[Zodiark1593]

Despite the superior single-gpu game performance, Titan should be thought of as a compute card first and foremost, basically a poor man's Tesla, with gaming as a nice bonus. If you only use it for gaming, you lose a lot of the value Titan represents.

 

[Jacky60]

I believe you, but I almost sense sarcasm. I am starting to think that due to all of this talk about microstutter that I must suffer from it even if I don't notice it. So, that said, how can I solve my newfound microstter problem? I'm sure if I could detect it that it would really bother me, so help please.

The only solution is to buy two (yes two) 690's or if you're feeling flash buy 2 Titan's. Then :\you'll see (or rather won't see) what you have (or haven't) been missing!

 

[Jacky60]

The definition of Microstutter is what you have if you don't have an Nvidia GPU.

 

[Ferzerp]

The definition of Microstutter is what you have if you don't have an Nvidia GPU.

Are you interesting in discussing the topic, or are you just wanting to contribute to the fight that has caused this board to decline?

You think you're being witty, but it's more of the same old rhetoric spewed by both sides.

 

[Jacky60]

I'm not sure, what would you suggest sage?

 

[BTRY B 529th FA BN]

I would buy the Titan, see how it compares to the 690, and then decide which one you like better.

The entire bickering in this thread reminds me of this: Yeah

 

[cmdrdredd]

I would buy the Titan, see how it compares to the 690, and then decide which one you like better.

The entire bickering in this thread reminds me of this: Yeah

With restocking fees, this might not be a smart plan.

 

[blackened23]

With restocking fees, this might not be a smart plan.

Yep, newegg has replacement only on it, no refunds. Also, renting PC hardware from stores which do allow returns with no intention of purchase is unethical, I don't see why people do it. I've read stories from people who buy 8-9 video cards and keep the one that overclocks the best, I personally find this practice to be completely disgusting. All for .001% more performance and a higher 3dmark11 score to brag about on OCN.....They're literally raising prices industry wide for everyone, RMAs are killing the industry which is already struggling due to PC decline.

Perhaps one day the US will be like many EU/asian countries where if you buy it...it's yours - have fun with it. I don't want this but seeing some of the scumbag practices mentioned on other boards, it may be for the better.... The way people abuse return policies, it really seems they're pushing the industry towards this.

 

[BTRY B 529th FA BN]

With restocking fees, this might not be a smart plan.

Then why not sell the one "you" don't want?

If I were to lose a little one the resale, if it's sold, I could live with the knowledge gained.

 

[cmdrdredd]

Then why not sell the one "you" don't want?

If I were to lose a little one the resale, if it's sold, I could live with the knowledge gained.

Hard to sell a niche part? I've seen GTX 690s on the FS/FT forum that don't get many offers near MSRP. If someone were to want one for $1000 they could buy it new.

 

[BTRY B 529th FA BN]

Hard to sell a niche part? I've seen GTX 690s on the FS/FT forum that don't get many offers near MSRP. If someone were to want one for $1000 they could buy it new.

That's the risk you take

 

[cmdrdredd]

That's the risk you take

Don't need to. 690 is faster and already in the OP's system.

 

[BTRY B 529th FA BN]

Don't need to. 690 is faster and already in the OP's system.

If I were to see which one I liked it's the route I would take.

 

[BrightCandle]

The 690 is faster in terms of fps but its often worse in terms of frame time. It takes the 690 longer to produce the frame, but it can do two at once. Clearly the 690 must have additional latency in its pipeline compared to the titan because each individual render takes longer.

Titan is smoother in its delivery of frames, its lower latency but it can produce less fps. Its not an easy decision, especially with a price that high. I don't think I will be replacing the 680s with a titan, but I really so want to see one in action in comparison. I know from my own testing that 30 fps on a single 680 is smooth in Arma 2 but anything below 45 fps is choppy on a pair of 680s. That seems to carry across to a lot of games as a general rule so with that in mind the extra 50% performance on a 690(2x680) might not matter, in practice the faster but less frames may well be better experience wise despite the lower fps. Very tough choice.

 

[An00bis]

In all seriousness - I don't get microstutter either. Others are telling us that we should be getting MS from SLI and a single card solution is "superior". (their words, not ours!)

It has been smooth sailing, literally and figuratively using SLI on my end. However, that being said i've read pages upon pages of people suggesting Titan over the GTX 690 and 680 sli because Titan has less microstutter. Apparently, that isn't the case according to frametime graphs?::shrug:: So this just makes one question any justification for Titan over 690 or 680 sli - there's no justification in terms of microstutter, price, or framerates; 680 sli and 690 are better in all three areas.

I don't know about SLI or crossfire microstutter since I only have a single gpu but people on the internet like to exaggerate and lie about things. Almost EVERYONE says that TVs are bad for gaming and that they have huuuuge input lag, I've just tested this and they're all wrong. There's no input lag, at least it's the same as it was when I was playing on my PC monitor, all you have to do is disable some settings in your TV's menu and the games run perfectly.

 

[poohbear]

Nice attempt at a super vague explanation as to why Toms charts are "garbage" without even explaining a valid reason. But since you mention PCPer, we'll go with it:

Since frame time are the ultimate metric for card comparisons, here we see without exception that SLI is better than single card and provides a superior experience with less microstutter. Titan has more microstutter than both 680 sli and GTX 690.

Thanks for proving my point BC. Even PCPerspective agrees that SLI has less microstutter than single card, single card sucks. Ironically enough, their results agree with the results at tomshardware: good luck trying to explain that one. Are PCperspective garbage now?

I'll let you stumble over trying to explain how SLI has less microstutter than single card. Go for it.

wait, what? i thought microstutter only occurs with SLI or Xfire? its a problem that manifests itself when using 2 cards, not a single card, that's not correct ur saying?

 

[looniam]

wait, what? i thought microstutter only occurs with SLI or Xfire? its a problem that manifests itself when using 2 cards, not a single card, that's not correct ur saying?

microstutter happens in ALL cards. SLI/Xfire set ups have it worse and just posting a percentile graph doesn't tell the whole story. lets take a look using FC3:

from what you see posted the TITAN looks like a stuttering fool but looking at frame times, avg FPS, percentile AND time spent beyond the target fps tell much more (btw, i completely agree the THG testing is pure garbage, too much info is thrown out to publish a "review")

as you see the 690 has a lower frame time = more FPS but there are obvious spikes and the line looks 'thicker' compared to the titan.

now looking at the avg FPS the 690 looks "faster" than the TITAN by 12 fps.

lets see how much time is spent rendering the avg fps:

ok both are relatively close . . the 690 does look better but, is it?

but here is where you see "stutter" or "lag" (depending on your preferred terminology)
@60 fps (16.6fps)

well of course the TITAN looks bad because the avg is 41 fps (around 40ms) and the 690 is 53 (close to the 60 fps of 16.6ms)

@30fps (33.3ms)

WHOA! what is that?!?! the TITAN spends more time at its avg of ~30 fps than the 690?? and the 690 gets 53 fps?? (a 20fps DIP)

now for the sad panda time @20fps (50ms)

the 690 spends more than 5 times as much time putting out frames slower than the TITAN when it has a ~17% higher fps avg.
(edit i mistakenly first looked at the 680 of twice as much)
now ask yourself where is the stutter?

 

[bunnyfubbles]

wait, what? i thought microstutter only occurs with SLI or Xfire? its a problem that manifests itself when using 2 cards, not a single card, that's not correct ur saying?

no, microstutter occurs in all GPUs, its only come to the forefront with multi GPUs because of how much worse they are

and of course recently the whole GCN microstutter issue that's been plaguing AMD is a perfect example of single GPU stutter

fortunately most people don't seem to be bothered by it, but for those of us that are bothered by it, well it severely limits our options

 

[3DVagabond]

no, microstutter occurs in all GPUs, its only come to the forefront with multi GPUs because of how much worse they are

and of course recently the whole GCN microstutter issue that's been plaguing AMD is a perfect example of single GPU stutter

fortunately most people don't seem to be bothered by it, but for those of us that are bothered by it, well it severely limits our options

Also, a word on words. Although I'm reading a Google translation, I can see that they used the word "microstuttering" to describe the frame latency issues on the Radeon. For what it's worth, I prefer to reserve the term "microstuttering" for the peculiar sort of problem often encountered in multi-GPU setups where frame times oscillate in a tight, alternating pattern. That, to me, is "jitter," too. Intermittent latency spikes are problematic, of course, but aren't necessarily microstuttering. I expect to fail in enforcing this preference anywhere beyond TR, of course.

Even TR says not to confuse multi GPU microstutter with their measurements. People really need to stop doing this. They have not been shown to be the same thing.