Petitioning for 970 Refund

[{NG}Fidel]

He still is ignoring the fact that they mislead on memory bandwidth. He skipped my entire post, the guy is desperate to make this situation go away and its beginning to raise my suspicions.
He ignores the stutter, stories like Golden Tigers, and the German benchmarks. I am sick of this thread too but I am more sick of people trying to make this a non-issue when it is.

 

[spaceace420]

He still is ignoring the fact that they mislead on memory bandwidth. He skipped my entire post, the guy is desperate to make this situation go away and its beginning to raise my suspicions.
He ignores the stutter, stories like Golden Tigers, and the German benchmarks. I am sick of this thread too but I am more sick of people trying to make this a non-issue when it is.

Best post of the day, That is the only reason I started posting again.

 

[jwcalla]

Quick question, When playing a game and your system memory use jumps up to the point it has to try and use a page file that is set up for 3Gigs and you have 8 Gigs of memory what is eating up the system ram? Do you remember the old days when memory leaks in games were happening quite often? Game played perfectly until that point of getting into the last segment of memory and then began paging to the system memory and then the pagefile. Where do you think the information in the video card comes from? Is it some magical pathway the rest of the world doesn't know about or does it come from system resources. They did not disclose this design that is working as intended until these issues came up right, so lie, misinformation and miscommunication depending on which side of the fence you are on bud.

If a game uses 400 trillion GB of memory, then yeah it's going to hit the page file. But that has nothing to do with the memory design of the 970.

If Windows is crashing with out of memory errors you have problems somewhere else.

He still is ignoring the fact that they mislead on memory bandwidth.

It is true; that was misleading. But in reality it is of no practical consequence.

He ignores the stutter, stories like Golden Tigers, and the German benchmarks.

I didn't ignore anything. I responded to his "stories" earlier in the thread.

 

[jwcalla]

He skipped my entire post, the guy is desperate to make this situation go away and its beginning to raise my suspicions.

Raise your suspicions? About what? No, I'm not employed by or affiliated with nvidia in any way, if that's what you want to know. I haven't bought an nvidia card since 2010. And unlike many people posting here in these threads, I've been posting here regularly for years and didn't just make an appearance when this brouhaha came up.

Isn't it possible in your mind that maybe I'm just telling you what I really think -- i.e., that people are blowing this way out of proportion -- and that I have no ulterior motive? Or is it so hard to believe that somebody couldn't get all spun up over what amounts to a non-issue? Or is it hard to believe that there could be somebody out there who doesn't believe that everyone is a victim of some corporate evil overlord trying to screw them over?

And I responded to your post, I just didn't quote it.

 

[Dayaks]

Dude stop lying to people...It does not perform as advertised...The 256Bit Bus? You want to tell me its performing at 224GB/s? You want to tell people that it was advertised to only have 24GB/s bandwidth in the last .5GB? I cannot believe that people think we are stupid enough to fall for this shit Stop defending them its done, stop white-knighting this company like your paycheck depends on it.

“ROPs (original) – 64
ROPs (actual) – 56

L2 cache (original) – 2MB
L2 cache (actual) – 1.75MB

Memory Bandwidth (original) – 224 GB/s for 4GB VRAM
Memory Bandwidth (actual) – 196 GB/s for 7 memory chips totalling 3.5 GB, and 28 GB/s for the remaining 512 MB 8th memory chip.”

IT CANNOT MAGICALLY PERFORM AT 224GB/s! THAT'S FALSE ADVERTISEMENT CUT AND DRY JUST ACCEPT IT!
Memory bandwidth affects performance, or do you never use AA, DSR, or any bandwidth intensive features?

I was really hoping this post was going to end this thread.

I am going to invest in hybrid SSD/HDDs, mark up the price and relabel as SSDs and sell them to the folks on this thread...

 

[doz]

Lets allow NV to do this. Fuck it. Then, when Intel releases their new 8 core CPU with 1 core that runs half speed, we wont feel so violated.

Anyone defending NV at this point is a NV troll. No 2 ways about it.

 

[{NG}Fidel]

If its such a non issue then what happened with GoldenTiger?

 

[trick0502]

Let the lawsuits begin

http://wccftech.com/nvidia-face-lawsuit-gtx-970-false-advertising/

 

[jwcalla]

If its such a non issue then what happened with GoldenTiger?

The cards probably just don't perform well at the 4k resolution he was using, and SLI just doesn't scale well.

People have problems with SLI all the time.

Like I said before, if you're using >3.5GB to generate frame after frame after frame (which is probably the case in most of these scenarios including GT's), then you're not going to get stutter, you're going to get a lower average framerate. If you're getting stutter, it's a different issue. If you don't understand why this is, then I give up.

 

[Rav3n]

Let the lawsuits begin

http://wccftech.com/nvidia-face-lawsuit-gtx-970-false-advertising/

WEll technically it hasn't become a class action cuit yet. But time will tell if it turns into that.

 

[{NG}Fidel]

Like I said before, if you're using >3.5GB to generate frame after frame after frame (which is probably the case in most of these scenarios including GT's), then you're not going to get stutter, you're going to get a lower average framerate. If you're getting stutter, it's a different issue. If you don't understand why this is, then I give up.

Yeah I give up with you if you do not understand that the memory limitation is affecting higher resolutions. You want to blame anything but the issue, so I will be done here.

 

[Kalmado]

Yeah I give up with you if you do not understand that the memory limitation is affecting higher resolutions. You want to blame anything but the issue, so I will be done here.

That's why he is now on my ignore list. He's too ignorant to have a constructive conversation. It frustrates me too that a person can be so blind, but I'm not going to let him get me riled up over his blind allegiance.

 

[Despotes]

Seems cut and dry to me. Actual performance is irrelevant.
64 ROPs advertised. 54 in reality.
2048KB L2 cache advertised. 1792KB in reality.
4GB of VRAM advertised (assumed fully operational at spec speeds.) In reality, only 3.5BG VRAM fully operational at spec.
It doesn't matter if you experience a performance loss or not. The marketing specs are inaccurate.
I'm curious what the settlement will entail.

 

[MelonSplitter]

Seems cut and dry to me. Actual performance is irrelevant.
64 ROPs advertised. 54 in reality.
2048KB L2 cache advertised. 1792KB in reality.
4GB of VRAM advertised (assumed fully operational at spec speeds.) In reality, only 3.5BG VRAM fully operational at spec.
It doesn't matter if you experience a performance loss or not. The marketing specs are inaccurate.
I'm curious what the settlement will entail.

This. Period. Performance gain or not.

 

[AceGoober]

Uh...

...simply because he had purchased two GTX 970 cards from Gigabyte.

Gigabyte, as far as I'm aware, doesn't sell directly to end-users.

 

[Tgrove]

Quick question, When playing a game and your system memory use jumps up to the point it has to try and use a page file that is set up for 3Gigs and you have 8 Gigs of memory what is eating up the system ram? Do you remember the old days when memory leaks in games were happening quite often? Game played perfectly until that point of getting into the last segment of memory and then began paging to the system memory and then the pagefile. Where do you think the information in the video card comes from? Is it some magical pathway the rest of the world doesn't know about or does it come from system resources. They did not disclose this design that is working as intended until these issues came up right, so lie, misinformation and miscommunication depending on which side of the fence you are on bud.

I noticed this as well. Anytime the vram usage gets high on my 970s my ram goes up around 12gb usually (16 total).

 

[jwcalla]

In order to be awarded damages you have to actually demonstrate losses. If there's no performance difference between 64 ROPs and 56 ROPs then what could one possibly claim they're due (in this particular instance)? Do they think they should be awarded money because they "feel" hurt by only having 56 ROPs? Even though there is no actual loss?

This is all very basic stuff here.

This all corroborates my point that people are blowing this way out of proportion because of some innate desire to feel offended, outraged, and victimized. "64 vs 56 ROPs makes no difference? Doesn't matter! PAY!"

 

[{NG}Fidel]

You always ignore memory bandwidth which is a demonstrable loss that impacts performance.

 

[LurkerLito]

In order to be awarded damages you have to actually demonstrate losses. If there's no performance difference between 64 ROPs and 56 ROPs then what could one possibly claim they're due (in this particular instance)? Do they think they should be awarded money because they "feel" hurt by only having 56 ROPs? Even though there is no actual loss?

This is all very basic stuff here.

This all corroborates my point that people are blowing this way out of proportion because of some innate desire to feel offended, outraged, and victimized. "64 vs 56 ROPs makes no difference? Doesn't matter! PAY!"

False advertising is already considered a 100% loss so I don't see what the issue is. There is an actual loss. The loss of the entire purchase price of the card because it doesn't have what they said it is suppose to have. Performance is again irrelevant. You can't advertise a car with a 300 HP engine then deliver a car with 275 HP. It doesn't matter if 99.99999% of the buyers will never use that extra 25 HP, the buyer was lied to and the manufacturer is obliged to either fix it so it does deliver what was advertised or to offer full refund if the buyer doesn't like the delivered specifications. Same applies here.

 

[spaceace420]

I noticed this as well. Anytime the vram usage gets high on my 970s my ram goes up around 12gb usually (16 total).

Thank you sir that is what I was getting at typing with an onscreen keyboard lol. I only have an 8 gb stick so it did turn into an issue that one single time with a crash, Only stuttering when it has happened since. Now with the new driver I haven't been able to get into the last playing som.

 

[Red Falcon]

If its such a non issue then what happened with GoldenTiger?

He whooped ass in this thread and proved the NV trolls and NVIDIA wrong.
This is a major issue, perhaps not for performance as much as it is a trust issue.

Everything NVIDIA releases from now on is going to be questioned, and rightfully so.

 

[Tgrove]

If you bought a monitor that was advertised as 2560x1440p, but upon further investigation it only does 1920x1080p....how would you feel? You bought it based on advertised specs, and received something different. What would you do?

 

[mope54]

If you bought a monitor that was advertised as 2560x1440p, but upon further investigation it only does 1920x1080p....how would you feel? You bought it based on advertised specs, and received something different. What would you do?

wouldn't the analogy be you bought a 2560x1440 60hz panel but it only refreshes at 59.9hz?

or hey, what about this one: you buy a burger and eat some of it, notice it is pink in the middle even though you asked for well-done. but it tastes ok so you go ahead and eat 7/8th of it then bitch out the waitress so badly that your new burger comes out of her paycheck because that's really how the people in the last few pages have been acting

 

[Climber]

In order to be awarded damages you have to actually demonstrate losses. If there's no performance difference between 64 ROPs and 56 ROPs then what could one possibly claim they're due (in this particular instance)? Do they think they should be awarded money because they "feel" hurt by only having 56 ROPs? Even though there is no actual loss?

This is all very basic stuff here.

This all corroborates my point that people are blowing this way out of proportion because of some innate desire to feel offended, outraged, and victimized. "64 vs 56 ROPs makes no difference? Doesn't matter! PAY!"

Your ignorance of the legalities does not in any way mean you are correct.

 

[Factum]

Just ordered a GTX970 for my daugthers PC.
Won't be asking for a refund, it's performance is the same as advertised.
Performance is my metric.

 

[xorbe]

The whole "it seems to perform fine for the price, so I don't care if ${major_corp} was misleading about key product metrics" is astonishing! Give them an inch, and they will take a mile. It's not just about this incident, but standards going forward.

 

[SeegsElite]

Just ordered a GTX970 for my daugthers PC.
Won't be asking for a refund, it's performance is the same as advertised.
Performance is my metric.

I like this attitude. It's refreshing to hear a statement like this in a sea of melodramatic posts as of late.

 

[Kalmado]

Just ordered a GTX970 for my daugthers PC.
Won't be asking for a refund, it's performance is the same as advertised.
Performance is my metric.

I like this attitude. It's refreshing to hear a statement like this in a sea of melodramatic posts as of late.

So the two of you support a company that falsely advertised products? It doesn't bother you only 3.5gb vram is usable? You are ok with that? Would you be ok buying a car that's advertised 35mpg but it only gets 28mpg? If you're not ok with that scenario, why are you ok with Nvidia's scenario?

 

[Factum]

So the two of you support a company that falsely advertised products? It doesn't bother you only 3.5gb vram is usable? You are ok with that? Would you be ok buying a car that's advertised 35mpg but it only gets 28mpg? If you're not ok with that scenario, why are you ok with Nvidia's scenario?

Why do you make false claims?
You can use 4GB of RAM RAM on the SKU before going into the page file in Windows.
Are you sure you understand topic

 

[Kalmado]

Why do you make false claims?
You can use 4GB of RAM RAM on the SKU before going into the page file in Windows.
Are you sure you understand topic

http://www.pcper.com/news/Graphics-C...B-Memory-Issue

From the article: "Over the past couple of weeks users of GeForce GTX 970 cards have noticed and started researching a problem with memory allocation in memory-heavy gaming. Essentially, gamers noticed that the GTX 970 with its 4GB of system memory was only ever accessing 3.5GB of that memory. When it did attempt to access the final 500MB of memory, performance

If anybody wants a full, technical explanation of the 970's memory subsystem hit up AnandTech.

The short of it is this:

Quote:
This in turn is why the 224GB/sec memory bandwidth number for the GTX 970 is technically correct and yet still not entirely useful as we move past the memory controllers, as it is not possible to actually get that much bandwidth at once when doing a pure read or a pure write. In the case of pure reads for example, GTX 970 can read the 3.5GB segment at 196GB/sec (7GHz * 7 ports * 32-bits), or it can read the 512MB segment at 28GB/sec, but it cannot read from both at once; it is a true XOR situation. The same is also true for writes, as only one segment can be written to at a time.

Unfortunately what this means is that accessing the weaker 512MB segment blocks access to the stronger 3.5GB segment if both memory operations are identical; or put another way, using the 512MB segment can harm the performance of the 3.5GB segment. For example, if we want to issue reads to both segments at once, reading the 512MB segment blocks any other reads to the 3.5GB segment for that cycle. If the 3.5GB segment is blocked in this fashion and doesn't have a non-blocking write to work on instead, it would have to go idle for that cycle, which would reduce the effective memory bandwidth of the 3.5GB segment. This means that taken over time in our example, the larger the percentage of the time the crossbar is reading the 512MB segment, the lower the effective read memory bandwidth would be from the 3.5GB segment.

Go ahead, keep burying your head in the sand. I'm sure Nvidia thanks mopes like you.

 

[Factum]

Go ahead, keep burying your head in the sand. I'm sure Nvidia thanks mopes like you.

The GTX970 can adress 4GB on the SKU.
It cannot be said more simple.
I suggest you do some reading up:
http://www.pcper.com/reviews/Graphi...Full-Memory-Structure-and-Limitations-GTX-970

I wanted to clarify a point on the GTX 970's ability to access both the 3.5GB and 0.5GB pools of data at the same. Despite some other outlets reporting that the GPU cannot do that, Alben confirmed to me that because the L2 has multiple request busses, the 7th L2 can indeed access both memories that are attached to it at the same time

For users that are attempting to measure the impact of this issue you should be aware that in some cases the software you are using report the in-use graphics memory could be wrong. Some applications are only aware of the first "pool" of memory and may only ever show up to 3.5GB in use for a game. Other applications, including MSI Afterburner as an example, do properly report total memory usage of up to 4GB. Because of the unique allocation of memory in the system, the OS and driver and monitoring application may not always be on the page. Many users, like bootski over at NeoGAF have done a job of compiling examples where the memory issue occurs, so look around for the right tools to use to test your own GTX 970. (Side note: we are going to try to do some of our own testing this afternoon.)

back to you.

(and could you stop with the personal attacks, I don't attack you for not understanding the topic, thank you)

 

[Ticker305]

Too bad some of you guys aren't mods on here, I would love to sell my gtx670 2gb in the FS section and advertise it as a gtx680 2gb card. Then when the buyer complains reply, well they both have 2gb and the gtx670 performs just like a gtx680 or better if you overclock it, so essentially its a gtx680. Who cares about specs as long as they both perform the same, thats all that ultimately matters in the end

 

[SeegsElite]

So the two of you support a company that falsely advertised products? It doesn't bother you only 3.5gb vram is usable? You are ok with that? Would you be ok buying a car that's advertised 35mpg but it only gets 28mpg? If you're not ok with that scenario, why are you ok with Nvidia's scenario?

Are you sure your TV is getting it's claimed contrast ratio? Most TVs claim millions to one contrast and fall millions short. Cars claim X amount of MPG and fall short. I can name 30 other products that claim one thing and deliver something different. Why is there so little outrage about those products but so much about this one?

 

[trick0502]

The GTX970 can adress 4GB on the SKU.
It cannot be said more simple.
I suggest you do some reading up:
http://www.pcper.com/reviews/Graphi...Full-Memory-Structure-and-Limitations-GTX-970




back to you.

(and could you stop with the personal attacks, I don't attack you for not understanding the topic, thank you)

Anytime a gpu has to do "tricks" to access all of its memory, that can't be good.

Back to the car analogy. If a car is marketed at 300hp at 35mpg, but it can only do 250hp at 35mpg, or 300hp at 27mpg, would that lying?

Looking at the price of the card compared to the 980, should have been a clue. 329 compared to 649. That is a huge price difference for a few fused off shaders.

 

[Factum]

Are you sure your TV is getting it's claimed contrast ratio? Most TVs claim millions to one contrast and fall millions short. Cars claim X amount of MPG and fall short. I can name 30 other products that claim one thing and deliver something different. Why is there so little outrage about those products but so much about this one?

But he is making a false claim, because he didn't follow up on this.
Hence why I posted a more updated link from the same site as he posted.

One would expect people that complain about a topic at least have the basic understading of the topic, this does not seem to be the case here.

There are 4GB usable RAM, not 3.5GB.

 

[Factum]

Anytime a gpu has to do "tricks" to access all of its memory, that can't be good.

Back to the car analogy. If a car is marketed at 300hp at 35mpg, but it can only do 250hp at 35mpg, or 300hp at 27mpg, would that lying?

An interconnect is not a trick?
It's used on hardware all over the sprectrum and manufactors?

The cards performance is as advertised, there was no rigged benchmarking with "fake" performance, hence you "analogy" fails.

And prices fall fast, when you disable units:
GTX780 $499
GTX780 Ti $699

 

[trick0502]

But he is making a false claim, because he didn't follow up on this.
Hence why I posted a more updated link from the same site as he posted.

One would expect people that complain about a topic at least have the basic understading of the topic, this does not seem to be the case here.

There are 4GB usable RAM, not 3.5GB.

3.5gb at 192gb/s and 0.5 at 24gb/s. Can it use all 4gb, yes. Can it use all 4gb at the advertised speed, no. Honestly, it can't even use the 3.5gb at the advertised speed.

 

[trick0502]

An interconnect is not a trick?
It's used on hardware all over the sprectrum and manufactors?

The cards performance is as advertised, there was no rigged benchmarking with "fake" performance, hence you "analogy" fails.

And prices fall fast, when you disable units:
GTX780 $499
GTX780 Ti $699

http://www.pcper.com/image/view/52241?return=node/62218

Look at the diagram. Look at the last l2 and mcs. There is an extra connection there to access the last 0.5gb. Yeah, that's a hack.

 

[Factum]

3.5gb at 192gb/s and 0.5 at 24gb/s. Can it use all 4gb, yes. Can it use all 4gb at the advertised speed, no. Honestly, it can't even use the 3.5gb at the advertised speed.

It has the perfomance as benched right?
Infact I think the performance since launch have improved.
So it's working better than advertised.

But do you have any real world tests to show me where this becomes an issue?

 

[Factum]

http://www.pcper.com/image/view/52241?return=node/62218

Look at the diagram. Look at the last l2 and mcs. There is an extra connection there to access the last 0.5gb. Yeah, that's a hack.

Look at the diagram again.
There are interconnects between all pairs of memory controllers.
(By design you do not know wich part of you die will have defects, so you design all units alike so you have flexability)
But I take it that you do not use the false claim of 3.5GB more, so we can put that part to rest?