Question Why does the overall gaming GPU market treat AMD like they have AIDS?

[VirtualLarry]

I guess I get the (sub-liminal) "The way it's meant to be played" ads from NVidia, along with the recurring FUD tropes about "AMD drivers", but I honestly don't get the sales disparity, especially for the price.

I've owned both NVidia-powered as well as AMD powered GPUs, and IMHO, AMD is (generally) just as good. Maybe 99% as good.

Edit: And I think that there's something to be said about the viability of AMD technologies, when they're in both major console brands.

 

[Heartbreaker]

Bricking your card is not worse than that? Amazing.

You can find individual reports of bricking cards with drivers from both AMD and NVidia users.

Show evidence there was widescale bricking.

 

[tamz_msc]

For those that feel like NVidia is always inherently superior in features:

How do you explain the GTX 970's "3.5GB" memory limitations, and the deceptions about it from NV (at least initially)? I've never seen similar deception from AMD about their cards memory configs.

Isn't that a good example, of the times that NVidia is "truely evil", and that it's a good thing that we have AMD on the market (and now Intel too).

Uh, your example has nothing to do with NVIDIA (or AMD) having superior (or inferior) features. The GTX 970 "3.5 GB + 0.5 GB controversy" was a non-issue in terms of the end-user experience, except in edge cases.

A real example of AMD having inferior features would be something like no ROCm for Windows.

 

[DeathReborn]

For those that feel like NVidia is always inherently superior in features:

How do you explain the GTX 970's "3.5GB" memory limitations, and the deceptions about it from NV (at least initially)? I've never seen similar deception from AMD about their cards memory configs.

Isn't that a good example, of the times that NVidia is "truely evil", and that it's a good thing that we have AMD on the market (and now Intel too).

4GB with 3.5GB at full speed (192GB/s) and 512MB at much slower speed (28GB/s). All 4GB is addressable under Direct3D, CUDA etc. it's just sub-optimal to use the last 512MB.

 

[Hitman928]

You can find individual reports of bricking cards with drivers from both AMD and NVidia users.

Show evidence there was widescale bricking.

Nvidia had to recall their driver and tell people who were on that driver to roll back their driver to the previous version. This was after it was starting to be widely reported that people were having their cards bricked with the latest driver.

Nvidia recalls 3D drivers due to overheating risk

Nvidia responds to reports of most recent graphics card drivers behind fan failures.

www.cnet.com

Edit: it was even a problem in their Linux driver, https://www.phoronix.com/news/ODA0MA.

 

[Heartbreaker]

Nvidia had to recall their driver and tell people who were on that driver to roll back their driver to the previous version. This was after it was starting to be widely reported that people were having their cards bricked with the latest driver.

So 12 years ago, one driver version had to be pulled.

It's a tiny blip, that you could instantly resolve with a rollback.

The 5700 launch had bad drivers with nothing to roll back to. For months 5700 cards were crashing.

Can you see why a long running, unresolved issue has more impact on reputation, than a bad version that can just be rolled back and fixed instantly.

 

[SteveGrabowski]

They ignore the stunts nVidia does where they degrade performance of older cards to make newer cards look better.

I was going to say, Nvidia is known for having good drivers... for the first year or two. Once your card is a couple of years old and there is a new gen it stops receiving the careful optimizations it got when new and performance drops off. Or in the case of Kepler, tanks.

 

[maddie]

So 12 years ago, one driver version had to be pulled.

It's a tiny blip, that you could instantly resolve with a rollback.

The 5700 launch had bad drivers with nothing to roll back to. For months 5700 cards were crashing.

Can you see why a long running, unresolved issue has more impact on reputation, than a bad version that can just be rolled back and fixed instantly.

7yrs, but of course this is irrelevant.

364.72

 

[SteveGrabowski]

For those that feel like NVidia is always inherently superior in features:

How do you explain the GTX 970's "3.5GB" memory limitations, and the deceptions about it from NV (at least initially)? I've never seen similar deception from AMD about their cards memory configs.

Isn't that a good example, of the times that NVidia is "truely evil", and that it's a good thing that we have AMD on the market (and now Intel too).

I'm glad they got sued to hell on that. It was a really good card but they were telling reviewers it was 4GB of RAM at full speed and I probably would have bought an R9 290 or R9 290x had it been known 0.5GB of the VRAM was so much slower. I imagine that lawsuit must have cost Nvidia a ton because I got 10% of my purchase price back on it so I'm guessing the lawyers got even more out of that class action suit.

 

[Hitman928]

So 12 years ago, one driver version had to be pulled.

It's a tiny blip, that you could instantly resolve with a rollback.

The 5700 launch had bad drivers with nothing to roll back to. For months 5700 cards were crashing.

Can you see why a long running, unresolved issue has more impact on reputation, than a bad version that can just be rolled back and fixed instantly.

Could be fixed unless of course your card was bricked before you found out you were supposed to roll back. . .

It's also not the only example, Nvidia has actually recalled more drivers after this for various reasons(see here and here).

Then of course you had major issues as recently as the Ampere launch with cards crashing consistently. Don't you remember the whole drama about the AIB's being blamed for their capacitor choices for causing cards to crash, then it turns out the fix was actually in a new Nvidia driver( here and here)?

Then you also had just plain card issues outside of drivers (or maybe in combination) such as the bumpgate and space invaders issues which AMD has never had as far as I know.

 

[Leeea]

Nvidia cards allowed forcing of visual features for as long as I've been a hardware enthusiast, back in high school.

About that:

That only works sometimes dx9, a bit random: https://www.tomshardware.com/reviews/anti-aliasing-nvidia-geforce-amd-radeon,2868-7.html
... and fails silently when it does not work.

In dx10 it rarely worked, and failed silently when it did not.
On dx11 it almost never worked, and failed silently when it did not.
On dx12 it never works, and still fails silently leaving the user with the impression it does work.

To be fair, AMD does the same thing.

Most recently the RX 5700 cards were plagued with black screen crashes right after launch. That took some time to resolve, and was likely a significant drag on sales.

Raja may be a brilliant guy, but he always over promised and failed on execution. RDNA1 is case in point.

Good news though, Raja is gone!

So 12 years ago, one driver version had to be pulled.

It's a tiny blip, that you could instantly resolve with a rollback.

. . .
Oct 4th, 2021:
https://www.denofgeek.com/games/new-world-bricked-gpu-causes-solutions-prevention-models/

At the moment, the EVGA RTX 3090 FTW Ultra and Gigabyte RTX 3080 Ti seem to be the two GPU models most commonly affected by this issue. That being said, RTX 3090, RTX 3080Ti, RTX 3080, RTX 3060 Ti, and RTX 3000 laptop users have all reported having their GPUs bricked while playing New World.

... and nvidia fixed it with a driver update.

If AMD had done that the nvidia cheer squad would never let is forget it!

 

[maddie]

There Are None So Blind as Those Who Will Not See.

 

[Leeea]

7yrs, but of course this is irrelevant.

364.72

I think he confused the incident you brought up with the spontaneously combusting nvidia GPUs of 2008.


oh, that is not 12 years either, ....
So does anyone remember what nvidia GPUs were spontaneously bricking themselves in 2010?

I know it is such common thing that we tend to get all the incidents blurred together.

 

[Heartbreaker]

If AMD had done that the nvidia cheer squad would never let is forget it!

You seem more interested in defending the AMD faith, than actually understanding why AMD sales lag, which was the point of the thread.

Attacking and disagree with the answers, is not going to improve AMD sales. Especially when you have no answer of your own.

New Worlds bricking some cards was a bizarre edge case.

Even if a driver is used to mitigate something, that doesn't mean a driver caused it. They might be using a driver to protect weaker than expected HW.

EVGA reported solder faults on it's New Worlds bricked cards:

EVGA explains how Amazon’s MMO bricked 24 GeForce RTX 3090s

A bad batch of early cards and a unique power load from Amazon's New World MMO are to blame for two dozen dead RTX 3090 GPUs.

www.pcworld.com

An analysis of dead EVGA GeForce RTX 3090 cards that failed while playing Amazon’s New World game indicate a rare soldering issue limited to a small batch of cards is responsible, a company spokesman told PCWorld.

 

[Stuka87]

So 12 years ago, one driver version had to be pulled.

It's a tiny blip, that you could instantly resolve with a rollback.

The 5700 launch had bad drivers with nothing to roll back to. For months 5700 cards were crashing.

Can you see why a long running, unresolved issue has more impact on reputation, than a bad version that can just be rolled back and fixed instantly.

The 5700 issue was two things. There was a launch day bug that would cause a black screen if hardware acceleration was on for a browser or something while also in a game. That was fixed pretty quickly. The second black screen was actually really rare. Several GPU testing sites/channels were unable to replicate it after the initial fix AMD put out. But people still reported it. And because a handful of people (comparative to the number of cards sold) posted on reddit and such, then the news sites posted it as if every single card constantly crashed. Because bad news for AMD is good for views!

I for instance, have never had a single black screen with my 5700XT. Obviously, this is anecdotal. One report does not make up the whole. But likewise, a bunch of users on redit also doesn't make up the whole.

As for the nVidia drivers bricking cards, this is the one I was referencing:

https://www.reddit.com/r/nvidia/comments/4crp3j

Nvidia Users Beware, Latest Driver May Harm Your PC - Allegedly Plagued With Issues

Nvidia GeForce users beware, the latest 364.72 Game Ready drivers may damage your PC and/or windows installation, driver plagued with issues.

wccftech.com

And then there was the 3090 launch, where the cards were constantly crashing for users. nVidia had to release updated drivers to prevent the issue from happening. The driver update did degrade the performance of the cards. And it was clear the launch day drivers tried to eek out as much performance as possible from the 3090. But this resulted in instability, and caused massive transients that was often kicking off power supplies fault protection, and causing many cards to outright fail.


So, my point is that people will happily forgive nVidia, regardless of how bad an issue may be. But AMD has one issue, and it lives for eternity.

 

[Hitman928]

I think he confused the incident you brought up with the spontaneously combusting nvidia GPUs of 2008.


oh, that is not 12 years either, ....
So does anyone remember what nvidia GPUs were spontaneously bricking themselves in 2010?

I know it is such common thing that we tend to get all the incidents blurred together.

The 2010 issues can be found in my post #154.

 

[Stuka87]

Uh, your example has nothing to do with NVIDIA (or AMD) having superior (or inferior) features. The GTX 970 "3.5 GB + 0.5 GB controversy" was a non-issue in terms of the end-user experience, except in edge cases.

This is not true. It most certainly did impact users. Any game that required more than 3.5GB of VRAM would get bad stuttering due to the obnoxiously slow VRAM for the last 512MB. There were plenty of videos around the web showing the issue happen. Its literally how end users determined there was a hardware issue when more than 3.5GB of RAM was used. If there was no impact to users, nobody would have ever noticed.

nVidia lost the class action lawsuit for this issue. And it wasn't just because of the marketing, but also user impact to users.

 

[Leeea]

You seem more interested in defending the AMD faith, than actually understanding why AMD sales lag, which was the point of the thread.

Attacking the answers, is not going to improve AMD sales.

New Worlds bricking some cards was a bizarre edge case.

Even if a driver is used to mitigate something, that doesn't mean a driver caused it. They might be using a driver to protect weaker than expected HW.

EVGA reported solder faults on it's New Worlds bricked cards:
https://www.pcworld.com/article/395...amazons-mmo-bricked-24-geforce-rtx-3090s.html

We are just pushing back on the claim that AMDs driver experience is worse.

It isn't, it is just it is perceived as worse by Nvidia faithful.


As for New World being an edge case, New World is not the only game with uncapped framerates in the menus:

StarCraft II melting graphics cards

StarCraft II: Wings of Liberty has been melting graphics cards. The - literally hot - Mac and PC game ships with no fra…

www.eurogamer.net

These are big mainstream games, they hardly seem like edge cases.

I imagined it also happened with a lot of smaller games to. Those likely just slipped under the radar.

 

[Heartbreaker]

We are just pushing back on the claim that AMDs driver experience is worse.

It isn't, it is just it is perceived as worse by Nvidia faithful.

So you are just labeling the majority of the market, as the "NVidia Faithful".

IMO, this just like calling Apple buyers iSheep.

It's not productive, and just reveals more about your own biases.

As for New World being an edge case, New World is not the only game with uncapped framerates in the menus:
https://www.eurogamer.net/starcraft-ii-is-melting-graphics-cards
These are big mainstream games, they hardly seem like edge cases.

I imagined it also happened with a lot of smaller games to. Those likely just slipped under the radar.

It's an edge case because it's EXTREMELY rare for a game to kill cards, and it looks like it was only killing poorly manufactured cards. It's not a case of Drivers bricking cards.

 

[maddie]

So you are just labeling the majority of the market, as the "NVidia Faithful".

IMO, this just like calling Apple buyers iSheep.

It's not productive, and just reveals more about your own biases.



It's an edge case because it's EXTREMELY rare for a game to kill cards, and it looks like it was only killing poorly manufactured cards. It's not a case of Drivers bricking cards.

But, but, but. Isn't a video card a Nvidia?

Productive? This argument isn't meant to change the world, but show the truth. As I posted, there are none so blind as those who will not see. NVIDIA has many driver problems that are forgotten almost as soon as they happen. Explain that. Looks a lot like religious faith to me.

 

[DooKey]

But, but, but. Isn't a video card a Nvidia?

Productive? This argument isn't meant to change the world, but show the truth. As I posted, there are none so blind as those who will not see. NVIDIA has many driver problems that are forgotten almost as soon as they happen. Explain that. Looks a lot like religious faith to me.

It's unfortunate that some people have a cult like faith to GPU and CPU companies. AMD/Intel/Nvidia have lots of those. It's not restricted to just one of those companies.

 

[aigomorla]

How do you explain the GTX 970's "3.5GB" memory limitations, and the deceptions about it from NV (at least initially)? I've never seen similar deception from AMD about their cards memory configs.

Larry AMD has also ooffed on the radeon 7 and HBM memory.
There is a reason why they don't use it anymore.
Its also one of the most failed videocards.
Its like owning a porsche.... once that darn Warrenty period ends, your Radeon 7 WILL DIE.

Sure HBM was great memory, but it was not that much of a improvement for gaming over GDDR.
HBM was way more prone to failing, way more expensive, and did they really expect people to pick it over a 1080 which Steve from GN states was the first biggest noticeable improvement that came before the RTX 4090.

Also AMD had really poor choices of AIBs.
Sorry, but Gigabyte is trash, ASUS is corrupt as hell, MSI stands for (Must Scalp It).
Lets not get into Gpu only vendors like Powercolor, Saphire, and XFX.

They lost all the really reputable AIB vendors.... HIS, BFG, (XFX used to be here until there company went south from new management).

Also, you read what the impact of eVGA exiting the market means for NVidia.
eVGA made modifications to circumvent Nvidia, and would actively share it to other AIBs.
Sure it would take a matter of time for Nvidia to go OOPS and fix it, but there was a vendor who was actively doing it and sharing it.
AMD has no such vendor.
Its a real cut throat competition there, with very little customization.
Until reciently, Card A looked almost identical to Card B from another vendor because of the lack of customization.
But Nvidia always had the pretty RGB cards, which smelled like perfume or had cool gundam decals on them for highly customized builds.

I don't remember any driver issue on NVidia side coming close to that.

Here i'll give you one where they still have not addressed, and required the complete replacement of getting a new monitor on a new display port gen to fix.

https://www.nvidia.com/en-us/geforce/forums/discover/252346/monitor-wont-sleep-windows-10-creators-update/

And yes it was fully driver issue with the conflict on microsoft update, which they never fixed even to this day.
And it had no issues on a radeon rx580 card, which the monitors went to.

Nvidia is not less evil then AMD. Infact they are now where Intel sat at the peak of the 7th generation of Intel's Core line.
And i hope they fail hard in the next coming gens like how intel got smashed after the 9th gen.
But that is all up to the RX 7000 series and how it competes against the RTX 4090.
But Steve did say the RTX 4090 is just ludicrous as the performance gain is even better then what the 1080ti brought, which was massively significant.


I do not camp at any band camp for GPU's or CPU's.
I go with that i know will last, and perform at the price i paid for.
I had both flag ships for both gens. So i know the cards.

Spoiler

And this gen, the RTX 3090 was superior then the RX 6900XT.

But if you go down the ladder, the RX6600XT is i think the better budget gamer card then the laughable 3060, or the obnoxiously overpriced 3060ti.

 

[tamz_msc]

This is not true. It most certainly did impact users. Any game that required more than 3.5GB of VRAM would get bad stuttering due to the obnoxiously slow VRAM for the last 512MB. There were plenty of videos around the web showing the issue happen. Its literally how end users determined there was a hardware issue when more than 3.5GB of RAM was used. If there was no impact to users, nobody would have ever noticed.

nVidia lost the class action lawsuit for this issue. And it wasn't just because of the marketing, but also user impact to users.

It is as I said, only noticeable in edge cases - like using 8x MSAA, or when using DX 12 which at the time of the controversy was pretty new and is known to have higher VRAM pressure.

For example, I picked four titles at random from TPU testing game performance that had the GTX 970 tested.

FFXV, BF V, RoTR and COD WWII.

All games can allot more than 3.5 GB memory when run with the settings tested at even the lowest resolution of 1080p, and the only instance where the memory config of the GTX 970 is a problem is in BF V using DX 12. But that is an easy fix - BF V runs better with DX 11, so not only does changing API alleviate the memory usage, it also leads to better performance.

 

[Stuka87]

It is as I said, only noticeable in edge cases - like using 8x MSAA, or when using DX 12 which at the time of the controversy was pretty new and is known to have higher VRAM pressure.

For example, I picked four titles at random from TPU testing game performance that had the GTX 970 tested.

FFXV, BF V, RoTR and COD WWII.

All games can allot more than 3.5 GB memory when run with the settings tested at even the lowest resolution of 1080p, and the only instance where the memory config of the GTX 970 is a problem is in BF V using DX 12. But that is an easy fix - BF V runs better with DX 11, so not only does changing API alleviate the memory usage, it also leads to better performance.

Benchmarks with average FPS don't show the issue. The only way to see stuttering in benchmarks is to look at 99th percentile frames, or to use frame time graphs with FCAT.

And technically, it could happen at anytime. Before the driver changes, memory could be put into that last 512MB at anytime. This is why it took people a while to figure out what was going on. It was really sporadic. For games that use over 3.5GB, its all the time.

The driver changes that nVidia put out after all of this prioritized the first 3.5GB, and it would only use the last 512MB if it absolutely had to.

But the issue itself was *NOT* a corner case. If it was, its likely it never would have become as big of an issue as it did.

 

[Leeea]

EVGA reported solder faults on it's New Worlds bricked cards:

yes.

But the issue was a bit more widespread then just EVGA rtx 3090s.

 

[Heartbreaker]

yes.

But the issue was a bit more widespread then just EVGA rtx 3090s.

That still doesn't make it a driver issue. EVGA analyzed their cards and reported it.

Let me know what you find out about others if it's that important to you.