So Nvidia lied about the GTX 970 for months

[vmem]

so any news on how this is all playing out? word is there are tons of people returning 970s in the EU, and the change.org petition is already well past 5k signatures so things are definitely not working on Nvidia's favor...

5000 signatures is nothing. I can't imagine how many of these got sold considering they sold out almost immediately after getting stocked everywhere for the first two months.

Nvidia will release a Driver for utilizing the 500MB of VRAM. Let`s wait and see this fix.

They already released an official statement saying there's no driver fix coming. they guy who promised a Driver fix on forums probably got fired already

 

[vmem]

Here` a link.

http://www.christiantoday.com/article/nvidia.to.fix.geforce.gtx.970.memory.issue.through.driver.update/47042.htm

http://www.pcworld.com/article/2876802/nvidia-plans-geforce-gtx-970-driver-update-for-memory-performance-concerns.html

 

[iamlegend]

This information is 6 hours ago. Look at the date.

http://www.christiantoday.com/article/nvidia.to.fix.geforce.gtx.970.memory.issue.through.driver.update/47042.htm

 

[ferwindjacks]

They didn't post any other source besides that Kitguru article, which relates back to the representative that wasn't speaking on behalf of Nvidia. Seems like they just took an old story and published it, not knowing the new change in story.

 

[Someone Somewhere]

Also, the memory is already used. It's just slower than the rest of the memory.

Unless they can re-enable bits of the chip via drivers (which normally they try to prevent, because 6950), they can't do anything.

 

[getdamafiaonyou]

Someone please tell me how to unsubscribe from this pointless thread.

 

[Someone Somewhere]

Go to the first page. Look at the first post. There's a 'Stop tracking this thread' link in the bottom left.

 

[getdamafiaonyou]

Go to the first page. Look at the first post. There's a 'Stop tracking this thread' link in the bottom left.

THANK YOU

 

[getdamafiaonyou]

Someone please tell me how to unsubscribe from this pointless thread.

Well there is an issue with the 970s so it's not pointless it's a truth.

I had 2 970's. however this thread has declined into pointlessness.

 

[getdamafiaonyou]

Someone please tell me how to unsubscribe from this pointless thread.

Well there is an issue with the 970s so it's not pointless it's a truth.

I had 2 970's. however this thread has declined into pointlessness.

I am just saying that we should not deny the 970 issue is all.

never claimed to be denying it, just that the thread itself has become pointless.

 

[Someone Somewhere]

You seem not to have actually used the stop tracking button.

 

[getdamafiaonyou]

You seem not to have actually used the stop tracking button.

I actually did. Which is why I'm kinda surprised I keep getting notified.

 

[Someone Somewhere]

Because when you post to the thread, it starts tracking again. You need to click the stop tracking button, then stop posting.

 

[youcanDUit]

Didn't somebody just say that Nvidia has said there will not be a patch for it, and the representative who claimed there would be such a patch wasn't speaking on behalf of the company?

how do you patch hardware? there's no such thing. not for this kind of screw-up.

 

[Someone Somewhere]

It's disabled hardware, not fundamentally missing hardware. Certain cards (6950) have had enthusiasts turn them into faster cards by flashing a new BIOS.

However, more recent cards have traces laser-cut to prevent them being re-enabled.

 

[j0ndafr3ak]

Should they have been honest? What should they have written on the box? 3.5GB of GDDR5 memory + .5 GB of slow ddr3-like vram? Haha come on...

 

[MrJak]

Should they have been honest? What should they have written on the box? 3.5GB of GDDR5 memory + .5 GB of slow ddr3-like vram? Haha come on...

3.5GB GDDR5 + 512MB hyper cache 😛

 

[MrJak]

It's disabled hardware, not fundamentally missing hardware. Certain cards (6950) have had enthusiasts turn them into faster cards by flashing a new BIOS.

However, more recent cards have traces laser-cut to prevent them being re-enabled.

Getting more than what you paid for is more than ok ... getting less is false advertisement and wrong.

The 970, for all intents and purposes, should have maxed at 3GB -- the last 1GB, including the 512MB slower RAM is really just a freebie.

 

[vmem]

It's disabled hardware, not fundamentally missing hardware. Certain cards (6950) have had enthusiasts turn them into faster cards by flashing a new BIOS.

However, more recent cards have traces laser-cut to prevent them being re-enabled.

Getting more than what you paid for is more than ok ... getting less is false advertisement and wrong.

The 970, for all intents and purposes, should have maxed at 3GB -- the last 1GB, including the 512MB slower RAM is really just a freebie.

True, and had they marketed it as such for ~$350 it still would've sold like hotcakes

Unfortunately they tried to hard for the whole "I need 4K and UBER amounts of VRAM" thing and just had slap the 4GB label on there, which is by definition fraudulent, however small the fraud may be. And now they're facing a class-action suite because of it.

Personally I am in that TINY minority who is actually affected by this as my original plan was to go triple SLI for 4K with full custom water cooling. so now I'm looking at $120 in on my first waterblock (thank god these things sold so well that I could only get one), and a 970 set up that I'll probably hang on to for the time being w/ my 1440p monitor and wait to see if 4k/GPUs get better by next year. So overall, I don't want to bash Nvidia too much (though I hate them for my personal situation), and while the issue definitely only affect a tiny minority, I hope this will stop both AMD and Nvidia from pulling such marketing stunts in the near future (we all know they'll never learn in the long run)

 

[MrJak]

The 970, for all intents and purposes, should have maxed at 3GB -- the last 1GB, including the 512MB slower RAM is really just a freebie.

Without being an nvidia apologist we can conclude the the last gig of ram is absolutely not a freebie. At worst you are fetting much less for your money and you will be bottlenecked earlier than cards with a true 4GB Vram buffer in future games.

I'm not a nvidia apologist, I'm just going from the trends that their cards have had.
The 770 had 1GB less than the 780 and had an, overall, worse value, than the 970 does now.
770: 2GB, MSRP: $399
780: 3GB, MSRP: $649

970: 3.5GB, MSRP: $329
980: 4GB, MSRP: $549

I think anyone who experiences issues should return it and get a different gpu, because apparently it, in general, wasn't enough for their needs -- not because the specs weren't "as advertised." Anyone who knows tech should have know that the specs were wrong -- these tech companies have a trend that they keep with evolving tech: minimal improvement on set segments. The 970 is what should have been a 970 ti.

 

[MrJak]

It's disabled hardware, not fundamentally missing hardware. Certain cards (6950) have had enthusiasts turn them into faster cards by flashing a new BIOS.

However, more recent cards have traces laser-cut to prevent them being re-enabled.

Getting more than what you paid for is more than ok ... getting less is false advertisement and wrong.

The 970, for all intents and purposes, should have maxed at 3GB -- the last 1GB, including the 512MB slower RAM is really just a freebie.

True, and had they marketed it as such for ~$350 it still would've sold like hotcakes

Unfortunately they tried to hard for the whole "I need 4K and UBER amounts of VRAM" thing and just had slap the 4GB label on there, which is by definition fraudulent, however small the fraud may be. And now they're facing a class-action suite because of it.

Personally I am in that TINY minority who is actually affected by this as my original plan was to go triple SLI for 4K with full custom water cooling. so now I'm looking at $120 in on my first waterblock (thank god these things sold so well that I could only get one), and a 970 set up that I'll probably hang on to for the time being w/ my 1440p monitor and wait to see if 4k/GPUs get better by next year. So overall, I don't want to bash Nvidia too much (though I hate them for my personal situation), and while the issue definitely only affect a tiny minority, I hope this will stop both AMD and Nvidia from pulling such marketing stunts in the near future (we all know they'll never learn in the long run)

Your situation makes more sense than most, but I don't believe it was ever intended for 4K gaming on any demanding games -- they just gave what they were technically able to do now, ad show off their new tech. I've seen benchmarks that suggest the 970 in SLI can take on the 980 though.

I honestly wish they cut the last 512MB just to simplify the memory structure, even though the extra 512MB can be faster than some PC's system ram.

 

[ferwindjacks]

Return em for a Titan X

 

[cst1992]

Return em for a Titan X

The Titan X is rumoured to cost more than 4x the cost of a 970. And it has so much VRAM, it's not going to be used at all by even the uber-enthusiasts, till the card is obsolete and no one buys it anymore.
If past trends are to be believed, the 1070 and 1080 will launch in 3 months with 6/4 GB of VRAM and cut down GM200(for 1080) and OCd 980 for less(1070). Get that then.

 

[Bossyfins]

Honestly, I think NVIDIA just really did screw up and have mis communication. Think about it, why would a major company try to lie... They know they will fail because of the intense testing.


I own a 970, and I am happy with it. I feel cheated, but it is not as bad as I don't go over 1080p.

 

[renz496]

Honestly, I think NVIDIA just really did screw up and have mis communication. Think about it, why would a major company try to lie... They know they will fail because of the intense testing.


I own a 970, and I am happy with it. I feel cheated, but it is not as bad as I don't go over 1080p.

because discussion conspiracy theories are much more fun? :lol:

but from what i can see so far nvidia only admitting the mistake about how the VRAM work but probably not so much about the amount of L2 cache and ROP count. as others points out that was spec that only given to reviewer to discuss architecture analysis. the box won't advertise them. and some afraid that this debacle might encourage to nvidia to release less info about their future architecture to reviewer. not on Maxwell but probably on Pascal or Volta.