Will VRAM-gate hurt Nvidia?

[Dabombinable]

What Nvidia did was like trying to sell a 1 TB SDD that was one part 900 GB TB and One part glued together 5400rpm 100 GB HDD.

A better analogy: 3.5GB of L2 cache, 500MB of L3. That's what effectively happened with the GTX 970. But its still a lot faster than system memory.

[GrimNeo]

http://www.guru3d.com/articles_pages/geforce_gtx_970_sli_review,18.html gtx 970 review.

http://www.anandtech.com/show/8568/the-geforce-gtx-970-review-feat-evga The card still performs the same as it did before. Please list over 50 aaa games that use more than 3.5GBs of ram at 1080p very high settings 2x aa.

[zappian]

A better analogy: 3.5GB of L2 cache, 500MB of L3. That's what effectively happened with the GTX 970. But its still a lot faster than system memory.

 

I guess , its still much slower than the 3.5 GB pool of memory.

Thats how people found out the problem, gaming at 4k and getting stutter.

[Dabombinable]

http://www.guru3d.com/articles_pages/geforce_gtx_970_sli_review,18.html gtx 970 review.

http://www.anandtech.com/show/8568/the-geforce-gtx-970-review-feat-evga The card still performs the same as it did before. Please list over 50 aaa games that use more than 3.5GBs of ram at 1080p very high settings 2x aa.

I know I haven't found any yet.

 

I guess , its still much slower than the 3.5 GB pool of memory.

Thats how people found out the problem, gaming at 4k and getting stutter.

For crying out loud, its not designed for 4K gaming, people who think it is are idiots. We are still a fair way off a single GPU solution that handles 4K properly. Even a GTX 980 will see the same problems.

[Clarkey]

Why the hell are controversies called ___gate.

Because they are likening it to Watergate.

[zappian]

http://www.guru3d.com/articles_pages/geforce_gtx_970_sli_review,18.html gtx 970 review.

http://www.anandtech.com/show/8568/the-geforce-gtx-970-review-feat-evga The card still performs the same as it did before. Please list over 50 aaa games that use more than 3.5GBs of ram at 1080p very high settings 2x aa.

 

What about people running two 970 in SLI for 4k usage that were deeply affected by this issue?

 

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

[Dabombinable]

What about people running two 970 in SLI for 4k usage that were deeply affected by this issue?

 

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

If you read all of the information pertaining to the GTX 970, its quite obvious that it was never capable of running all 4GB of vRAM at the full bandwidth. Those who don't look in-depth into a product (and if people did look in-depth they would have seen this 'problem' with the vRAM) have no one to blame but themselves for choosing a product that doesn't fully suit their needs. I know about the misleading ROPS and slightly lower L2 cache, but again looking at it in-depth it was noticeable that it wouldn't be quite the same as a GTX 980, with lower CUDA cores never being the only difference between it and a GTX 970,  people should have known that it wasn't that simple.

[Takata]

I don't think this will hurt Nvidia much. However, if I was in the market for a video card (and I'm not), this would cause me to avoid Nvidia for about half a year.

 

Some would argue that the 970 still performs extremely well for its price, and that most users are unlikely to ever use the last 0.5GB, so this doesn't matter. However, I am willing to pay a little extra, or lose a little performance, to do business with the company that did not recently publish intentionally(?) misleading, inflated parameters. The inflated parameter being unimportant doesn't excuse Nvidia, in my opinion. It was advertised, so it should be met.

 

Sometimes, companies put out products that don't perform as well as expected, and that's okay. Occasional sub-par engineering is inevitable. However, I find intentional lies or inflated figures to be harder to forgive. (Although we'll probably never know how intentional the misleading 4GB figure was. Nvidia fans may say it was a misunderstanding between the engineers and the marketers. Nvidia haters may say everyone at Nvidia knew from the start and hoped no-one would find out.)

[han_han08]

well, everybody is surprised with price the 970 started with..perhaps the price was intended that way for the users to be compensated with this vram limit.lol

[Andrewf]

Are we still talking about this? Didn't Linus himself put this to rest? I can't seem to find the video at the moment unfortunately, but I do remember his face, and some explanation of 3.5GB and another 0.5GB in a card.

[Briggsy]

Are we still talking about this? Didn't Linus himself put this to rest? I can't seem to find the video at the moment unfortunately, but I do remember his face, and some explanation of 3.5GB and another 0.5GB in a card.

 

yeah, the 0.5GB partition and 3.5GB partition can't be accessed during the same read/write cycle, and its basically a 3.5 GB card with a 0.5GB partition of Ram thats a bit faster than System Memory (although system memory is faster latency-wise so windows uses it first... most of the time).

 

Some people are still finding out about this, so it will take time for it to die down.