Official Nvidia GTX 970 Discussion Thread

[thebennyboy]

No, I'm genuinely asking because I don't know how you can say Victorious is being a fanboy when he, like the very few people in this thread, are looking at this from a rational standpoint and have attempted multiple times to show you people what is and what isn't. Obviously that isn't working, so that's why I asked: do you have any reading and comprehension skills?

 

My definition of "rational" is obviously much different to yours. Clearly being rational means you flame anyone who disagrees with you or has issues with what Nvidia have done.

[Guest Kloaked]

My definition of "rational" is obviously much different to yours. Clearly being rational means you flame anyone who disagrees with you or has issues with what Nvidia have done.

I'd do the same thing if you people were being dumb about what any company is doing - I have no company bias.

[thebennyboy]

I'd do the same thing if you people were being dumb about what any company is doing - I have no company bias.

 

So in that case he's not being rational. I have nothing more to say on the matter.

[Victorious Secret]

I'd do the same thing if you people were being dumb about what any company is doing - I have no company bias.

 

Its amusing, people have issues with being told to calm down and look at things rationally. I don't really care if you're Red or Green team, be rational about it and avoid descending into doomsday levels of nonsense about everything. 

But its easier to be knee jerking every reaction than it is to think things through. Then again, this site doesn't have a stellar population of grown ups who take things more slowly so I'm not surprised. Its funny what "issues" people chose to get invested in yet ignore the things that are far more detrimental to their PC use/gaming experience. 

[patrickjp93]

we never were. but i dont hold grudges, so if you realise youre the idiot in this thread, i will treat you like everyone else again.

hell if @patrickjp93 managed to get me to treat him like everyone else again, im sure you can (no offense patrick)

How dare you treat me like one of the other plebeians on this horrid excuse for a tech and computing forum. Who do you think you are?!

[TroubleKlef]

My my this thread has blown up into a giant e-peen contest. 

 

Are we talking about Nvidia misleading consumers or who has the biggest ego? Holy cow. 

[Shahnewaz]

I like how your trying to evade the topic at hand since you have no valid response.

But.... but I thought you were the Tech Geek!

Here is a hint: Arch____ture impro______ts

 

But I guess architectures, implementations, improvements, specifications and all that lengthy non-techie BS is irrelevant, right?

[79wjd]

But.... but I thought you were the Tech Geek!

Here is a hint: Arch____ture impro______ts

 

But I guess architectures, implementations, improvements, specifications and all that lengthy non-techie BS is irrelevant, right?

Which is exaclty why specs are meaningless. Specs don't tell you anything about how different architectures compare. So again, how does reading off a spec list tell you anything of any importance? The only reason you can compare different architectures is because benchmarks exist -- unless you happen to be some sort of savant who can map out circuits and draw conclusions of the power/efficiency of an architecture with nothing but an architectural map. 

 

But hey, if you can, then more power to you. 

[TroubleKlef]

Which is exaclty why specs are meaningless. Specs don't tell you anything about how different architectures compare. So again, how does reading off a spec list tell you anything of any importance? The only reason you can compare different architectures is because benchmarks exist -- unless you happen to be some sort of savant who can map out circuits and draw conclusions of the power/efficiency of an architecture with nothing but an architectural map. 

 

But hey, if you can, then more power to you. 

 

Agreed, been saying this for awhile now. Specs=/=performance but is simply an indicator of performance. 

 

That being said. If the spec sheet said that I was paying for X amount of something. I better have that in my product irregardless if that product still performs admirably. 

[Zulkkis]

 

 

 

The 970 and 980 still performs exactly how it does in all the benchmarks before people knew about any of this, so I don't see any of your point. 970 SLI is fine (most people knew it was a superior alternative to a single 980):

 

 

There are plenty of uses for these cards outside of the benchmarks, and these particular benchmarks don't show what might happen should you break the 3.5GB limit. The specs were false, nVidia chose not to come forward with it until people started noticing.

 

I'm not saying this is the end of the world, but a dick move? Very much so.

 

I can understand your side of the argument, had I bought the cards for 1080p gaming I would not care. However, I do rendering myself, and it'd be interesting to know what happens with the card once you break that 3.5GB limit. I'd expect consistent performance. Sure the counterargument of "professional cards lol" is there, but frankly the GTX series of cards are faster in that use, and far more cost efficient as a result. There are only very limited uses where the drivers of a quadro series card might pull ahead, and they aren't my concern.

 

Nobody runs the benchmarks for full complex 4GB time because that happens to take disturbingly long periods of time and nobody expects the memory performance to magically drop halfway through.

 

However I do think that a customer who bought the card without this knowledge, should be entitled to return a functional card and be refunded if he chooses to do so. The card might have technically have 4GB of RAM, but at least the amount of cache and RoPs is very much a lie. However, the difference in memory is what concerns people the most, and rightfully so. A high-end card released when it was has a pretty low amount of VRAM tbh, and having any less effective memory (and it does show up in some uses, otherwise people wouldn't have noticed it in the first place) is a concern some people might have.

 

Again, I wouldn't be too pissed had I only bought one card for 1080p use like most folks. If I'd sli a couple of them for higher resolutions (where it was with the old specs, and unquestionable choice unless you can afford not to budget) I'd be slightly pissed. If I'd get a couple for a mixture of gaming and rendering (pure rendering people would most probably get a pair of 780 6GB's) I'd be pissed.

[79wjd]

Agreed, been saying this for awhile now. Specs=/=performance but is simply an indicator of performance. 

 

That being said. If the spec sheet said that I was paying for X amount of something. I better have that in my product irregardless if that product still performs admirably. 

True, and I think Nvidia deserves all the flak they get for posting inaccurate specifications. But that doesn't change the fact that the 970's performance is exactly the same as it has always been so anyone complaining about the 970 being stubborn. 

 

Also, ew you're one of those people...'irregardless' :unsure:

[Shahnewaz]

Which is exaclty why specs are meaningless. Specs don't tell you anything about how different architectures compare. So again, how does reading off a spec list tell you anything of any importance?

IF they WERE meaningless then how could you distinguish the 970 from a 980?

 

The only reason you can compare different architectures is because benchmarks exist -- unless you happen to be some sort of savant who can map out circuits and draw conclusions of the power/efficiency of an architecture with nothing but an architectural map.

So that means I cannot compare the 970 with the 980. They have the same architectures.

[TroubleKlef]

True, and I think Nvidia deserves all the flak they get for posting inaccurate specifications. But that doesn't change the fact that the 970's performance is exactly the same as it has always been so anyone complaining about the 970 being a crippled card is just being naive. 

 

Also, ew you're one of those people...'irregardless' :unsure:

 

Mmmm... who knew irregardless was improper usage of the English language  ? Being that this is an internet forum, I think English etiquette is out of the window  :lol: .

 

I understand that the 970 is still the same card before and after the information leak but its definitely not the same card you paid for. If that makes any sense. 

[HippY]

True, and I think Nvidia deserves all the flak they get for posting inaccurate specifications. But that doesn't change the fact that the 970's performance is exactly the same as it has always been so anyone complaining about the 970 being stubborn. 

 

Also, ew you're one of those people...'irregardless' :unsure:

The problem is that they should

 

The reason why I am upset of this is simple.

 

Technically there are 4 GB of RAM chips on the card and yes, the card can use all of them

But it was made in such a way, that when the system access the last chip, the bandwidth drops.

 

So what is the problem for me is that they are trying to walk on the very very edge of false advertising, and if this company (or any company for that matter) who starts to walk on this route must be stopped, unless you want false advertising to take place, or borderline false advertising.

 

 

This GM204 can handle 3.5 GB of RAM by its architecture, but they sorted out that it gets the last one with performance drop. It is not true advertising in my eye

[mr moose]

Agreed, been saying this for awhile now. Specs=/=performance but is simply an indicator of performance. 

 

That being said. If the spec sheet said that I was paying for X amount of something. I better have that in my product irregardless if that product still performs admirably. 

 

This I agree with, one caveat though: specs can only indicate performance if you have an understanding of the product history and what each spec means within that environment.  Ergo you'd have a hard time working out if a card is better or worse based on the clock if you don't know how many cores the card has etc. Much like the clock spec for processors is only relative to a few CPUS of the same model let alone brand.  This makes looking at specs a waste of time unless you really know what you are looking at (enthusiast).

 

 

IF they WERE meaningless then how could you distinguish the 970 from a 980?

 

So that means I cannot compare the 970 with the 980. They have the same architectures.

 

Personally, I ascertain what product I am buying by looking at the serial on the chip, the name on the heatsink/shroud and the printing on the box. I figure reviewers and product testers do the same.

[GoodBytes]

This GM204 can handle 3.5 GB of RAM by its architecture, but they sorted out that it gets the last one with performance drop. It is not true advertising in my eye

But one can argue that: isn't it the reviewer fault for not noticing? It is their duty to give us all the information. Some sites write large articles on new architectures, yet they don't know when they do reviews?

 

Playing devil advocate here.

[Colonel_Gerdauf]

IF they WERE meaningless then how could you distinguish the 970 from a 980?

 

So that means I cannot compare the 970 with the 980. They have the same architectures.

 

One word; benchmarks. Only benchmarks can tell you what is a more powerful card over the other, because it uses a real-life case scenario. On it's own, the GPU specs are next to completely useless, because it does not tell us how it will be utilized by the programs and games. I have said this before, I'll say it again, the benchmarks from before and after the incident did not change, because NVidia did not change the physical unit (unlike Kingston). All that changed was some numbers on a piece of paper.

[79wjd]

IF they WERE meaningless then how could you distinguish the 970 from a 980?

 

So that means I cannot compare the 970 with the 980. They have the same architectures. 

Within the same architechture specs aren't entirely useless, but performance doesn't always scale linearly in all applications. So just because one card has x more cores and y higher clock speed doesn't mean it will by x/y faster. 

 

Mmmm... who knew irregardless was improper usage of the English language   ? Being that this is an internet forum, I think English etiquette is out of the window   :lol: .

 

I understand that the 970 is still the same card before and after the information leak but its definitely not the same card you paid for. If that makes any sense. 

It's not improper, it's just redundant  

 

/I agree with you there too, the internet/forums are not a place where grammar is really important especially when the forums are international. 

 

And yes, it makes sense, but at the same time, you should base a decision on (realistic) benchmarks/real-world performance more than numbers. 

 

The problem is that they should

 

The reason why I am upset of this is simple.

 

Technically there are 4 GB of RAM chips on the card and yes, the card can use all of them

But it was made in such a way, that when the system access the last chip, the bandwidth drops.

 

So what is the problem for me is that they are trying to walk on the very very edge of false advertising, and if this company (or any company for that matter) who starts to walk on this route must be stopped, unless you want false advertising to take place, or borderline false advertising.

 

This GM204 can handle 3.5 GB of RAM by its architecture, but they sorted out that it gets the last one with performance drop. It is not true advertising in my eye

Like I said, Nvidia deserves to take flak for this, but thats not really whats going on here. 

[Faa]

Who is ever 4" away from their GPU??

 

Whoever or wherever told you that must clearly be mistaken...

That's measured from a meter away, lots of accuracy is lost. A 290x reference is 40 dBa at quiet mode, a tiny bit higher than a 970 which is also BS, feel free to find people who claimed their reference to be quiet, a tiny bit louder than a 970 is also BS. 780 reference is quieter than the VaporX according to your graph, I've owned them and they're freaking loud hence why I upgraded to a 970 Strix.

 

[TroubleKlef]

The problem is that they should

 

The reason why I am upset of this is simple.

 

Technically there are 4 GB of RAM chips on the card and yes, the card can use all of them

But it was made in such a way, that when the system access the last chip, the bandwidth drops.

 

So what is the problem for me is that they are trying to walk on the very very edge of false advertising, and if this company (or any company for that matter) who starts to walk on this route must be stopped, unless you want false advertising to take place, or borderline false advertising.

 

 

This GM204 can handle 3.5 GB of RAM by its architecture, but they sorted out that it gets the last one with performance drop. It is not true advertising in my eye

 

Whole heartedly agree.

 

Consumers should not be complacent to tactics such as false advertising, both borderline and blatant. Businesses are more than just the transfer of goods between two parties. It's a relationship between the consumer and seller. Allowing things like this to happen ruins the dynamic of this relationship. 

[jaggysnake57]

Performance is NOT THE SAME at 4k.

 

http://www.reddit.com/r/buildapc/comments/2tu86z/discussion_i_benchmarked_gtx_970s_in_sli_at_1440p/

 

it is the same....it couldnt do 4k before and it still cant now. thatpile of turd you linked even showed in battlefield 4 @ 4k average fps was 30fps...not playable regardless of memory issues. things really went to shit though when he used watch dogs and farcry 4 as testing scenarios, both of which are known to have all sorts of issues to do with optimisation. the facts remain the same that the 970 was never a 4k card 1440p is it max and TBH its ok @ that nothing mind blowing.  seriously its getting embarrassing how badly this is being blown out of all proportion. yes information has been misrepresented but you lot are acting like they advertised a 980 and sold you a 750ti 

[HippY]

But one can argue that: isn't it the reviewer fault for not noticing? It is their duty to give us all the information. Some sites write large articles on new architectures, yet they don't know when they do reviews?

 

Playing devil advocate here.

no it is not, as some has noticed at SLI the frame droppings, and games today rarely use more than 3.5GB of RAM

It is just purely nVidia`s fault about dancing on the edge of the false advertising

 

@LukaP it does not matter whether a game is well written or not in a consumer stand point, as you would have a card with 4GB of RAM, so it should handle

[BiG StroOnZ]

 

There are plenty of uses for these cards outside of the benchmarks, and these particular benchmarks don't show what might happen should you break the 3.5GB limit. The specs were false, nVidia chose not to come forward with it until people started noticing.

 

I'm not saying this is the end of the world, but a dick move? Very much so.

 

I can understand your side of the argument, had I bought the cards for 1080p gaming I would not care. However, I do rendering myself, and it'd be interesting to know what happens with the card once you break that 3.5GB limit. I'd expect consistent performance. Sure the counterargument of "professional cards lol" is there, but frankly the GTX series of cards are faster in that use, and far more cost efficient as a result. There are only very limited uses where the drivers of a quadro series card might pull ahead, and they aren't my concern.

 

Nobody runs the benchmarks for full complex 4GB time because that happens to take disturbingly long periods of time and nobody expects the memory performance to magically drop halfway through.

 

However I do think that a customer who bought the card without this knowledge, should be entitled to return a functional card and be refunded if he chooses to do so. The card might have technically have 4GB of RAM, but at least the amount of cache and RoPs is very much a lie. However, the difference in memory is what concerns people the most, and rightfully so. A high-end card released when it was has a pretty low amount of VRAM tbh, and having any less effective memory (and it does show up in some uses, otherwise people wouldn't have noticed it in the first place) is a concern some people might have.

 

Again, I wouldn't be too pissed had I only bought one card for 1080p use like most folks. If I'd sli a couple of them for higher resolutions (where it was with the old specs, and unquestionable choice unless you can afford not to budget) I'd be slightly pissed. If I'd get a couple for a mixture of gaming and rendering (pure rendering people would most probably get a pair of 780 6GB's) I'd be pissed.

 

The reason why they don't show what might happen when you break past the 3.5GB barrier is because, the card is not powerful enough to drive the settings required to push past the 3.5 GB barrier. That's why when running 4K even in SLI you see reviewers disabling AA. Because the card is not capable of doing such things. All of a sudden, people are getting distraught over the fact that their card has difficulty running 4K DSR, 8X MSAA, Ultra Textures, along with every other setting maxed out, and it's suddenly not performing perfectly. This type of scenario would happen on any card aside from the 970. Increasing settings to achieve over 3.5GB of VRAM usage will hurt performance anyway. So trying to prove whether or not that is directly related to the .5 GB memory section is going to be difficult to determine. They might have misrepresented the ROP count, or separated the full 4GB of memory. But why is everyone "up in arms" over it now. It makes no sense. We knew what the card was capable of doing before this information surfaced, now we just know why it is as so, it suddenly makes it a problem? It's like trying to sue, or demand a refund when the 3GB GTX 780 struggled in certain games that demanded more VRAM, but the 6GB GTX 780 had no problem with those particular games. Then trying to demand NVIDIA and other manufactures to replace their 3GB GTX 780 with a 6GB GTX 780. Because an individual is trying to run settings that we knew the card wasn't capable of doing before hand.

 

It might be a dick move, but there were many dick moves in this industry over the past couple of years. Remember the mining craze? When retailers were charging $800 for an R9 290X? The card retailed for $550. Now that card is going for $300. Where's everyone getting ready to sue over price gouging? Demanding a difference in the money they spent? What about when AMD released a cooler that didn't effectively cool the unit for the reference 290's and 290x's? In order for the cooler to keep the card from throttling you had to have it at 100% fan speed and which made it sound like a jet engine. Why didn't anyone complain about a functionality issue then? I mean they did complain, but not to the extent where they demanded refunds or their money back in hoards. This was obviously a hardware issue design, where the reference cooler just wasn't built properly for the card. Tons of people complained, but I don't recall their being an uproar of people demanding their money back to the extent of this issue.

 

Have you tested the card in rendering when you go over the 3.5GB limit? I mean, it's not exactly an issue until you test it and discover it is an issue. For all you know it might not be an issue at all and even mentioning it to use as a argument against them is pointless.

 

Except I've seen a handful of people test a decent amount games up to 4GB of usage reporting no problems at all. Then another handful reporting problems. So who is right? When there is a 50/50 stance on the issue?

 

People think they are entitled to a lot of stuff. However, I think people aren't giving NVIDIA the benefit of the doubt. As if they aren't going to make this right. They will. They have no choice. They currently hold market share over AMD. They can't afford to lose all that market share because of one mishap. If they do, they will be in trouble. So before anyone says, I demand this or I demand that or I'm expected to see this happen. They should wait and see what NVIDIA is going to do or what the manufacturers of the cards themselves are going to do. Because you have to believe that this is all being talked about right now behind closed doors. There is going to be a solution, there just isn't one yet. These type of things take time, and people aren't having any patience over it. 

 

Well this is where we differ, I don't see why people should be pissed. All the benchmarks for the card is available in troves from many different sources. If you researched your card before hand, and knew what it was capable of. You wouldn't be pissed, since this information was available for many months now. If you expected anything different, I really don't understand. This information was available. The only difference between now and then, is we actually know why it is the case. Which doesn't change the facts of how it performed, just adds to the understanding of why it performs this way. 

[TroubleKlef]

snip

 

I think many of us here are trying to say that this is an Nvidia problem not a 970 problem. You may link as many benchmarks and performance figures as you'd like but that still doesn't rectify the problem of misleading information. 

[Brainless906]

Yes, you and people like you need to grow up. If those "real" specs were never researched and released, none of you would be bitching about the 970. Why? Because it's a good card. Nothing has changed since it's release. Every contracted reviewer gave it a thumbs up due to its performance per watt, overall performance in general and it's price point.

How are you going to tell me I'm defending a company that doesn't care about me? I'm looking at this rationally; so don't call me a fanboy or someone defending a company.

For fuck sake, I was supposedly "defending Ubisoft" for doing something good despite the p.o.s. game they released (Unity). You can be objective and rational and not be a "fanboy" or a company's white Knight.

Edit: I also doubt that they're throwing their marketing team under the bus. When you read something like that on the internet with little context, one would over sensationalize it's actual meaning. It seems like a simple miscommunication, possibly. It happens.

 

 

I'd accept that argument if the performance didn't fall to shit after hitting 3.5gb+ on the card. From most of the content I've seen the gpu starts to stutter and glitch when pushing over 3.5gb of VRAM usage even though when its not stuttering and glitching the game would be completely playable 30+fps.

 

My argument is they lied. It doesn't actually work as intended, you push the card to its limits to play your favorite games and it breaks. You seem to believe its ok just because most games are playable at 1080p? I call some mad bullshit yo.