Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
It's still a good card for the price if you need cuda and don't have the money for a 980, works well for me. ¯\_(ツ)_/¯
 
Yet another thing that shows people have too much time on their hands.

AMD had to lower price on their top card to compete with Nvidia's second rung card.

AMD needs to get off their butts and compete.
 
I'll admit that I haven't sought it out since the early days, but I've yet to see any tangible real-world examples of this problem manifesting itself. I feel the card was priced with this in mind to begin with.
 
The throttling would mostly affect higher resolutions wouldn't it? Isnt 3.5gb of memory overkill for 1080p resolution, and it would mostly manifest at 2k or 4k. Regardless, the benchmarks are what they are - I don't see how knowing why the benchmarks are lower makes much difference. I for one am thrilled at the prospect of a 140-150 watt TDP card with that level of performance.
 
The throttling would mostly affect higher resolutions wouldn't it? Isnt 3.5gb of memory overkill for 1080p resolution, and it would mostly manifest at 2k or 4k. Regardless, the benchmarks are what they are - I don't see how knowing why the benchmarks are lower makes much difference. I for one am thrilled at the prospect of a 140-150 watt TDP card with that level of performance.

While I'm not implying you're endorsing what they did, it sounds like you don't care, or don't think its a big deal?

I think it's sad and shouldn't be allowed to happen otherwise how do we ever trust that what we're actually getting is what is advertised? They need to be held accountable regardless of whether it makes a difference or not even in edge cases.

At the very least it's shameful and what's most bizarre is that it was completely unnecessary.
 
While I'm not implying you're endorsing what they did, it sounds like you don't care, or don't think its a big deal?

I think it's sad and shouldn't be allowed to happen otherwise how do we ever trust that what we're actually getting is what is advertised? They need to be held accountable regardless of whether it makes a difference or not even in edge cases.

At the very least it's shameful and what's most bizarre is that it was completely unnecessary.

I would agree that early specs of 970 being wrong is suspect. But there is no way to fool a benchmark. Certainly no way to fool them all.

So a 970 will have whatever performance whether you call it a 9,700,000 or say it has a light-energy engine. The end result is whatever it DOES.

Sort of like Apple renaming a 7700 a "D300" or a 7870 into a "D500" to make mid-grade old-tech cards palatable.

At the end of the day, it's what the card does, and what it cost.
 
I'll admit that I haven't sought it out since the early days, but I've yet to see any tangible real-world examples of this problem manifesting itself. I feel the card was priced with this in mind to begin with.

i get choppy graphics in Saints row 4 and Far cry 4 if i max the settings at 1440p, this it with 970 in sli, they shouldn't be any choppiness at all. even my gtx 680 didn't have an issue.

so yes, this is a problem.
 
While I'm not implying you're endorsing what they did, it sounds like you don't care, or don't think its a big deal?

That's probably fair criticism. I can only point out that we have seen permutations of this story play out many times before, because the modus operandi for the GPU makers has always been to design the flagship card first (I can't argue with that), then ask "how do we skimp on features for our 2nd tier card?" We have seen everything from lazy and/or benignly neglectful down-design to ham-handed kneecapping, and this is just the latest example.

For what it's worth, for every person who is concerned about the .5gb of partitioned, slow-boat-to-China memory, there are other people who are reviewing the card positively and are pleased with its performance.

Maybe a better question to ask, and discussion to have, would be to compare the GTX 970 to 2nd-tier cards of the past so that we are comparing apples-to-apples. Is it any more or less hobbled than past 2nd-tier cards, and is it any more or less of a value at its lower price point relative to the flagship card than other 2nd-tier cards have been in the past.
 
Yeah. I think this is all kinda dopey. It performs how it performs, and people who go out of their way to read reviews should understand that. Meh.

Well, they're still advertising it as a 4GB card when it's incapable of actually performing like one. It'd be like Apple offering 4GB of RAM but it couldn't address all of it.

If the card had just been 3.5GB, it wouldn't have been a huge issue, but because there's 4GB on the card people are hitting the massive slowdowns because games are still trying to use that memory.

Yet another thing that shows people have too much time on their hands.

AMD had to lower price on their top card to compete with Nvidia's second rung card.

AMD needs to get off their butts and compete.

And of course the guy selling graphics cards would say it's no big deal, everything's fine, by the way, we have a special on 970s flashed for Mac!
 
http://blogs.nvidia.com/blog/2015/02/24/gtx-970/

I just fallen of my chair, laughing my a** off.

"Feature", "New memory architecture"...

This is like putting insult to injury. Like someone would believe that all the engineers at NVidia did not know what the results of limiting the memory bandwidth to 1/8 for 12.5% of the memory would be.

Speaking of failed communication is a joke. Either the company is a dictatorship and engineers are not allowed to speak their concerns—or marketing and management simply decided to ignore the facts. The latter is more likely, as anyone in the tech industry knows.
 
Or if like Apple advertised a 64 bit computer then pulled OS support because of a 32 bit EFI.....whoops !

Proof's in the pudding.

Which is why I don't understand why you're minimizing Nvidia's actions here, especially after you were railing against Apple for their failure to properly support 4K on Nvidia cards.
 
Which is why I don't understand why you're minimizing Nvidia's actions here, especially after you were railing against Apple for their failure to properly support 4K on Nvidia cards.

Already explained this once.

I'll try again.

You can call something anything you like in computer world. You can make any claims you want. (at the peril of your reputation)

But at the end of the day, you either deliver the goods, or not.

If this problem was REALLY some huge revelation ("MemoryGate"?) then why did it take months to be figured out?

Because for most uses, the issue NEVER COMES UP. And when it does, the benches show it, Nvidia can't hide from reviews and benchmarks. And they knew this going in. They weren't going to have reviewers hallucinate better scores than card could do.

So the card is what it is. In use cases where this last 512MB is needed, the card slows down. And in places where it isn't needed, the card is insanely fast and power efficient and an incredible value for the money.

In the next couple days I will send Barefeats 2 of our EFI 970s. I'm not scared of what he will find, just like Nvidia wasn't when they started selling them.

Speaking of which, did anyone notice his last round of Dual GPU tests? The 2010 Hex Core with Dual 980s left a nMP D700 for dead in both the DaVinci Resolve Candle test as well as Lexmark OpenCl. My guess is that this 970 noise won't stop the 970s from doing rather well in his tests. Guaranteed to beat the D300 and D500 slugs, may give D700s a run for the money.
 
From Nvidia, the "apology".

From Techpowerup http://www.techpowerup.com/210149/i...eo-breaks-silence-on-gtx-970-controversy.html as of today.

In the wake of bad PR, and a potentially expensive class-action lawsuit over the GeForce GTX 970 memory controversy, NVIDIA CEO Jen-Hsun Huang wrote a candid letter addressed to everyone concerned, explaining in the simplest possible language what went wrong with designing and marketing the chip, how it doesn't affect the design-goals of the product, its quality or stability, and how it could be misconstrued in a whole different ways.

Huang's explanation of the issue isn't much different from the one we already have, but bears the final stamp of authority from the company, especially with the spate of discrepancies between what NVIDIA representatives post on GeForce forums, and what ends up being the company's position on certain things. Huang's letter signs off with "we won't let this happen again. We'll do a better job next time."

The transcript of Huang's letter follows.

Hey everyone,

Some of you are disappointed that we didn't clearly describe the segmented memory of GeForce GTX 970 when we launched it. I can see why, so let me address it.

We invented a new memory architecture in Maxwell. This new capability was created so that reduced-configurations of Maxwell can have a larger framebuffer - i.e., so that GTX 970 is not limited to 3GB, and can have an additional 1GB.

GTX 970 is a 4GB card. However, the upper 512MB of the additional 1GB is segmented and has reduced bandwidth. This is a good design because we were able to add an additional 1GB for GTX 970 and our software engineers can keep less frequently used data in the 512MB segment.

Unfortunately, we failed to communicate this internally to our marketing team, and externally to reviewers at launch.

Since then, Jonah Alben, our senior vice president of hardware engineering, provided a technical description of the design, which was captured well by several editors. Here's one example from The Tech Report.

Instead of being excited that we invented a way to increase memory of the GTX 970 from 3GB to 4GB, some were disappointed that we didn't better describe the segmented nature of the architecture for that last 1GB of memory.

This is understandable. But, let me be clear: Our only intention was to create the best GPU for you. We wanted GTX 970 to have 4GB of memory, as games are using more memory than ever.

The 4GB of memory on GTX 970 is used and useful to achieve the performance you are enjoying. And as ever, our engineers will continue to enhance game performance that you can regularly download using GeForce Experience.

This new feature of Maxwell should have been clearly detailed from the beginning.

We won't let this happen again. We'll do a better job next time.

Jen-Hsun


Fundamentally the matter is simple. There are two groups of users. Those who bought the card because of its perceived performance and those who bought the card because of its specs. Both groups have different intentions and reactions. Those who bought the card for the performance knew what they were getting into and largely didn't care what the specs were. Those who bought the card for the specs because those are assuring things for the long-term longevity of the card in future games (like 4GB simultaneously useable, full-speed VRAM) cares about the specs and thus feels cheated. This is of course does not including hardcore fanatics like fanboys.

Either way, I think it is a pretty big "miscommunication" if what the CEO said is true. If it is true, then it is an even bigger issue as Nvidia has apparently no oversight over how their partners market the card and no Nvidia engineers read reviews to catch on the mistake from day one. A lot can be read between the lines depending on how much you believe them/their PR.
 
Well, they're still advertising it as a 4GB card when it's incapable of actually performing like one. It'd be like Apple offering 4GB of RAM but it couldn't address all of it.

What does a "4GB card" perform like? You can get 4GB on a GTX 760, but the 970 - warts and all - runs circles around it. The actual performance is there, and that's why all this circle jerk is dumb. It doesn't perform any differently now compared to when it launched.
 
For what it worth, I'm reading several bad report about this card using 3d softwares. As soon as the GPU approach the 3.5GB limits users are experiencing unusual freeze and crash not seen on other Nvidia cards. This is happening with different brands of 970 and testing many different drivers.
 
For what it worth, I'm reading several bad report about this card using 3d softwares. As soon as the GPU approach the 3.5GB limits users are experiencing unusual freeze and crash not seen on other Nvidia cards. This is happening with different brands of 970 and testing many different drivers.

Because of the unusual memory architecture in the split partition, this card will always have compatibility issues in terms of performance in games and in professional workloads as you suggested. Remember when Nvidia promised driver improvements to diminish the effect of the memory partition a short while ago ? I doubt they will do this hastely for the professional market, because in their defense, this card was never intended for that market, so for those users, they may be out of luck. Games wise, you are at the mercy of Nvidia to promptly deliver optimizations. I don't think this will be a big issue for games, as everyone optimizes their drivers for each GPU release anyway. When the card finally becomes old enough to be abandoned in driver updates, then it won't matter anymore anyway.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.