Official Nvidia GTX 970 Discussion Thread

[zappian]

 

 

 

Why the hell should I care what they are priced in one foreign country? I don't live in the US. That in no way affects my purchasing decision.

 

 

Maybe sometime in the future, perhaps. Right now the 970s's 4GB is absolutely fine except for a couple of terrible ports (including AC:U) so the standard 290X would be absolutely fine for this too. Which is good, because the 8GB version still costs about £400, which is stupid. 

 

 

There is no link associated with that picture. There is no testing methodology, no reference to quality beyond "high", there is no reference to how much AA they are using. All we know is that they are testing what is notorious for being one of the most demanding games in existence at 4K and having it cripple basically every dual GPU solution that you could possibly test it on. If you are claiming this is representative, then your point should be that 290Xs are bad at 4K because if they could be expected to throw out a maximum of 40 fps generally they would also be bad.

 

 

So your argument is that all benchmarks are wrong.

 

 

 

 

 

And the 970 wins on a gameworks title which is to be expected

 

 

 

 

Now granted, for the most part, AMD Radeon R9 290X CrossFire was the fastest video card configuration allowing the best gameplay experience across the board 

[othertomperson]

So your argument is that all benchmarks are wrong.

 

If you have to rely on benchmarks that use SMAA just to support a claim that a GPU cannot handle 4K then you clearly have no idea what you're talking about. That's why I pointed out that the previous Crysis 3 picture had no source associated with it, because stuff like this is vital to asserting whether or not you can get a decent gaming experience at 4K, which you are going out of your way to ignore.

 

Jesus Christ, I just showed your two videos of what gaming at maximum presets without AA looks like at 4K, and you are ignoring it because the facts don't align with your preconceptions.

 

By all means laud the 290X at this resolution. It's a very competent GPU and my recommendation is only determined by how good a price you can get. But your assertion that 970s are a bad solution, or are incapable, is just flat out wrong.

 

 

4GB>3,5GB 

He was just proven wrong ._. i mean i don't even...

 

There isn't a GPU in production right now with 3.5GB of RAM so that's irrelevant. Maybe you should actually read articles instead of just their sensationalist headlines.

 

And no I wasn't. Not remotely.

[zappian]

If you have to rely on benchmarks that use SMAA just to support a claim that a GPU cannot handle 4K then you clearly have no idea what you're talking about. That's why I pointed out that the previous Crysis 3 picture had no source associated with it, because stuff like this is vital to asserting whether or not you can get a decent gaming experience at 4K, which you are going out of your way to ignore.

 

Its impossible to argue with someone who ignores hard proof.

The burden of proof was on me and I delivered so. 

[Nikolithebear]

I think what we can all take away from this discussion is that the 970 has a flaw.  If it affects you, then change it out for something else you want.  If you have no problems with you 970, then dont change it.  I learned that upcoming I might change depending if my problems continue but as most of you have said, turn down the settings abit, and if that's what I'm going to have to do thats what I'm going to do.  

[othertomperson]

Its impossible to argue with someone who ignores hard proof.

The burden of proof was on me and I delivered so. 

 

Have you actually looked at the benchmarks you have quoted? There difference between the 290X and the 970 is only significant in one game. The others are near identical, and were hovering around 30 for both. Your source doesn't even back up the point you are trying to make.

 

If your point was to claim that 4K gaming wasn't ready for any GPU, this would be one way of showing it. It would be a crap way of showing it, since turning off AA would be barely noticable and would have a dramatic impact on framerates (as shown in the Thief videos).

 

If anything you've just reiterated what I've been saying all along that generally speaking differences between the 970 and 290X are tiny.

 

You don't seem to understand the concept of "hard proof." How many times have I pointed out that 970s at 4k is my actual gaming experience. It is literally how I play every game I have. You don't have to prove to me how good or bad 970s are at this, it's something I am a lot more experienced with than you are.

 

I'll also add that the numbers for your Crysis 3 benchmarks in the two sources you have provided are dramatically different. To the extent where you could get completely different impressions on its playability as far as 290X corssfire goes. This is why linking to the articles themselves, instead of just the pictures, and ascertaining their methodology is very important.

[Lewellyn]

There is no link associated with that picture. There is no testing methodology, no reference to quality beyond "high", there is no reference to how much AA they are using. All we know is that they are testing what is notorious for being one of the most demanding games in existence at 4K and having it cripple basically every dual GPU solution that you could possibly test it on. If you are claiming this is representative, then your point should be that 290Xs are bad at 4K because if they could be expected to throw out a maximum of 40 fps they wouldn't be doing much better than 970s.

 

The age old fall back of asking about testing methodology. That really wasn't the point, you were all gun-hoe about how someone should look at benchmarks until you made the mistake of not looking at them yourself. Now you're just grasping at straws. I don't care about 4k gaming and I don't think it's here yet, but there is a known issue with the 970's peak VRAM usage and higher pixel counts drastically increase memory far more than any other variable. I'm not recommending any other card in particular, if it was a radeon something or other crossfired or 980 sli I don't really care.

 

edit: And it's not just average framerates that are affected, it's frame drops and stuttering that can be a problem - you're already playing at a lower framerate at 4k the stuttering would be unbearable.

[xAcid9]

And the 970 wins on a gameworks title which is to be expected

 

 

 

Medium vs Ultra. 

[zappian]

Medium vs Ultra. 

 

Lol handt even noticed that

 

So the only game it won the settings werent the same .

[othertomperson]

The age old fall back of asking about testing methodology. That really wasn't the point, you were all gun-hoe about how someone should look at benchmarks until you made the mistake of not looking at them yourself. Now you're just grasping at straws. I don't care about 4k gaming and I don't think it's here yet, but there is a known issue with the 970's peak VRAM usage and higher pixel counts drastically increase memory far more than any other variable. I'm not recommending any other card in particular, if it was a radeon something or other crossfired or 980 sli I don't really care.

 

edit: And it's not just average framerates that are affected, it's frame drops and stuttering that can be a problem - you're already playing at a lower framerate at 4k the stuttering would be unbearable.

 

The testing methodology would be irrelevant if the point being discussed were "the 290X performs better than the 970s at 4K". This isn't the point being discussed. The point being presented is "The 290X offers a playable experience, the 970s is bad." The testing methodology becomes vital in this scenario. I could just use 8x Supersampling to make any card look shit here.

 

You don't think we're here yet because you are ignoring anything that contradicts your preconception that we aren't by people who already do. Someone telling you that the dpi is high enough for AA not to matter anymore, and so needlessly limiting your FPS to deal with it is "grasping at straws" to you.

 

It is true that the GTX 970 suffers a 3% performance dip as its vram gets close to saturation. You can go on about the sky falling as much as you like, but you are massively overstating the impact, even at 4K.

[Justhavocman]

The testing methodology would be irrelevant if the point being discussed were "the 290X performs better than the 970s at 4K". This isn't the point being discussed. The point being presented is "The 290X offers a playable experience, the 970s is bad." The testing methodology becomes vital in this scenario. I could just use 8x Supersampling to make any card look shit here.

 

You don't think we're here yet because you are ignoring anything that contradicts your preconception that we aren't by people who already do. Someone telling you that the dpi is high enough for AA not to matter anymore, and so needlessly limiting your FPS to deal with it is "grasping at straws" to you.

 

It is true that the GTX 970 suffers a 3% performance dip as its vram gets close to saturation. You can go on about the sky falling as much as you like, but you are massively overstating the impact, even at 4K.

To be honest i'm just saying that the 970 was not marketed to be a killer 4K card , it was marketed to be a killer 1080p card. Ofcourse you can play at 4K , my 670 can play at 4K on mainstream games like League of legends , Dota 2 etc , the point of discussion is whether the 970 will be a viable 4K solution for future use , to which the answer is : hardly. As mentioned before an 8Gig 290X might be , but the 970 is not , and it's ok , i mean it's ok that it isn't  , at that price it's a really good card , just not for 4k.

[othertomperson]

To be honest i'm just saying that the 970 was not marketed to be a killer 4K card , it was marketed to be a killer 1080p card. Ofcourse you can play at 4K , my 670 can play at 4K on mainstream games like League of legends , Dota 2 etc , the point of discussion is whether the 970 will be a viable 4K solution for future use , to which the answer is : hardly. As mentioned before an 8Gig 290X might be , but the 970 is not , and it's ok , i mean it's ok that it isn't  , at that price it's a really good card , just not for 4k.

 

One of them, sure. There isn't a GPU on the market now that is sufficiently powerful for AAA 4K gaming. And that's an important distinction for a lot of people. If you only want one GPU right now you're pretty much limited to a max res of 1440p, unless you exclusively play games like LoL as you mention.

 

But two GPUs makes a tonne of sense now. Whether it's two 970s, two 290Xs (or a 295X2). Two 980s is stupidly expensive for very little gain over 970s or 290Xs so I don't recommend that, and I have to throw "Titan Z" out there for a laugh.

 

I don't think the 8GB 290X makes much sense either. It's expensive and it might benefit you in two or three games but no more. That might change in the future, but I don't see that being for a little while.

 

But if you're willing to double up, whether it be SLI or Crossfire, 4K is very doable right now.

[Lewellyn]

The OP already selected an answer....

[Butterfish]

Im talking about 4k and higher .

everyone besides you might not know that.

[Mick Naughty]

How are you even able to use that much vram? I cant get bf4 to use that much.

[Lewellyn]

How are you even able to use that much vram? I cant get bf4 to use that much.

 

You can read about what impacts vram on this thread.

 

"Rendered resolution of a game has about 5% of an impact on how much vRAM the game is using. The only game I've seen make a large jump in used vRAM from a resolution increase is Watch Dogs, which (according to benchmarks) goes from ~3100MB vRAM to ~3800MB vRAM bumping from 1080p to 4k."

 

Increasing the pixels by 4 times doesn't mean 4 times the memory usage so it's not 1:1 but it can create a significant increase. 

[Blade of Grass]

Hello!

 

This is the official thread where people can discuss their issues (or non-issues) with the 970.

 

Notice:

Additional threads about issues with the 970 will be locked.

Any thread about 970 will be merged with this one.

[CoolaxGaming]

The GTX 970 DOES have 4 gb everyone. Its not 3.5... It just has less ROP's, less l2 cache because of that, the last 0.5 gigs is SLOW. 1/7 of performance. I dont know why no one has sued Nvidia over false advertising yet...

 

Anyone saying 3.5+ is unnecessary see this  = , Titanfall @ 1080p with Insane Grazphics = 3.5 - 3.7 gigs.

[Interfectorem]

I broke the two pins on my G1 Gigabyte GTX 970 that provide power to the "Windforce" LED whilst disassembling the card. Talk about screw ups.

[Laputacake]

I feel sorry for this thread already

[BuckGup]

I find it hilarious that everyone was buying the GTX970 like hot cakes and saying they are amazing for 4K and 1080p flawless. Now after they said the true spec and memory bug everyone says they suck and aren't that good and can't even play 4k. I think it is perfectly fine, I just can't wait to see if AMDs R9 300 series is as good as they say or if there is some serious heat issues due to 3 dimensional memory. 

[TheSLSAMG]

I care about the issue, but I am not disappointed in the card. I am a little distrustful of NVIDIA for straight up lying about some specs of the card, but I will not stop buying their cards because of it.

 

And good luck Blade, this thread will probably turn sour fast without swift moderation.

[Syntaxvgm]

http://linustechtips.com/main/topic/303145-i-found-a-problem-with-the-970-memory-test-im-examining-nais-benchmark-and-possibly-making-my-own/

^^

About Nai's test

[Dabombinable]

I ended up with a GTX 970 that takes +65mV to hit +153MHz stable, and +50mV to hit +140MHz stable. No way in hell has the GPU gone through the 'Super Overclock GPU Gauntlet' when even standard Windforce cards can hit +140MHz with no voltage increase.

[wyattzx]

inb4 interview video. 

 

I think they should cut the price of the 970, and lower its VRAM to 3GB. Maybe a solid $300? Or go to an actual 4GB card and keep the current price. 

[Razzaa]

Im talking about 4k and higher .

I didnt realize people were playing @ resolution higher than 4K. That must be a great experience...............