[AMD_Robert] Concerning the AOTS image quality controversy

[Minkoff]

"massive rendering error"? Have you looked at the pictures? The snow shader is destroying information which has been rendered by the GPU in a previous step.
The snow shader has no positiv impact on the picture. Applying a snow shader after the rendering of the terrain has no benefit at all and cost only performance.

So yes, it is no big deal because the picture looks (much) and performs better.

There are simple answers to all your question, and I mean all:

Object Space Lighting - Oxide Presentation
http://oxidegames.com/wp-content/uploads/2016/03/Object-Space-Lighting-Rev-21.pptx
"Could never afford to place the entire terrain into a texture – to much offscreen unused space"
...and nothing is destroyed, as there is nothing to be destroyed. The presentation even has pictures, so it should not be a problem.

You have those options in your driver. And sometimes AMD is taking care of it for you.

Don't tell me, tell it those that make game reviews but is it a game menu option or is it clever thinking (on AMD part)

 

[poofyhairguy]

So knowing this, nVidia definitely focuses on this game's performance.

That can't be debated anymore. It's obvious no matter what happens with this issue Nvidia cares about how a relatively unpopular RTS runs on their cards. They have done a lot of work specific to this game. Ashes is the Direct 12 battleground right now, popularity or not.

 

[Silverforce11]

So the 1080's better showing over the 980Ti/Titan X in DX12 is due to a dodgy driver that skips rendering shaders??

Wow. Cheating busted, where's the tech press up in arms about IHVs cheating with drivers?

*crickets chirping*

That's right. USA tech sites, all bought out. Leave it to EU tech sites to cover this..

 

[3DVagabond]

Nice thanks

https://twitter.com/PellyNV/status/739890789208055808

Funny that it was only broken in a press only driver, one that regular users wouldn't be able to validate... Wonder why.

So nVidia sends drivers to the press that render at lower IQ settings?

 

[3DVagabond]

I was totally expecting you to see some malice, I wonder why. Before making accusations without proof shouldn't we wait to see how performance compares with bad and good drivers?

Also why did AMD use a driver that they were not even supposed to have?

If performance with official drivers is actually higher are you going to say "I wonder why?" maliciously accusing AMD of cheating? I bet not.

Whatever drivers do now is no absolvement for this. They could have limited the IQ at the time because there was a performance issue that's since been fixed.

 

[Zstream]

Sure it is. When a developer uses the most inefficient way to render terrain (why do they need to use multple shaders instead of a texture?!) then i would trade it off for more performance. It is funny that Oxide doesnt give me this option. But i guess they are trying very hard to do so much work in every frame to put AMD at the number one spot.

And yes: Even as a Maxwell user i want this option. Snow or no snow - it doesnt matter because graphics is not important here.

I don't even know what to say...

 

[3DVagabond]

Thumbs up for totally ignoring the fact you have no proof of what you said. Waiting for facts, not malicious wishful thinking.

Fact: nVidia press only drivers rendered at lower IQ. Does that sound above board to you?

 

[renderstate]

Fact: nVidia press only drivers rendered at lower IQ. Does that sound above board to you?

So what? You never seen a driver bug in your life?

 

[sirmo]

So what? You never seen a driver bug in your life?

I have, but I've also seen Nvidia cheat and misrepresent things as well. Like for instance a demo of a 1080 FE @2100Mhz running cool as a cucumber.

 

[Silverforce11]

I don't even know what to say...

Some people have a habit of defending the indefensible.

It's simply more honest to just call them out when they make such mistakes.

Selling a throttling 1080 for a premium? That's good marketing.

Driver cheating in DX12? That's just cheating. Unacceptable!

 

[Slaughterem]

So what? You never seen a driver bug in your life?

A driver bug that they fixed for the public release driver even before it became apparent??? LMAO

 

[renderstate]

1) We currently have no proof the bad driver was improving perf.
Dan Baker says he'd be surprised if perf was very different (https://twitter.com/dankbaker/status/739894485878984705) but that won't stop some individuals claiming NVIDIA cheated.

2) Drivers are software and software is buggy. According the aforementioned individuals upon discovering a bug in their drivers NVIDIA shouldn't have fixed it.. God forbid! Because we all know that video drivers are a bug free and immutable piece of software made of millions of lines of code. Once NVIDIA release a driver to reviewers all development should be halted, time to move to the next GPU. This brilliant observation comes from the same guys that can't stop saying NVIDIA stops optimizing for older GPUs. How can one possibly argue against such strong points?

We don't even know if NVIDIA was aware this bug was impacting AotS. It could very well be they fixed it because of another title. Of course we cannot let logic ruin a nice conspiracy theory. Someone mentioned some AMD driver broke image quality on AotS for some GCN GPUs but it was eventually fixed. Bad AMD, don't ever do that again

3) Company A publicly compared their product to company B product using outdated software they shouldn't even be in their possession. The up to date software sits 2 clicks away and it has been available for almost a whole week. This is a fact, not a conspiracy theory. Is this bad? Good? Borderline unethical? I am truly not sure and it would be a nice debate to have if done without preconceived notions and partisanisms. What I am sure is that all the individuals above will reply with more logic defiant arguments, ignoring the facts they don't like and creating new ones out of thin air to make some more baseless accusations.

 

[sirmo]

1) We currently have no proof the bad driver was improving perf.
Dan Baker says he'd be surprised if perf was any different (https://twitter.com/dankbaker/status/739894485878984705)

You can't even quote the tweet without spin. Haha

He said, doubt it would be very different. You wrote that he said: doubt it would be any different. Those are two entirely different meanings.

Can't really take anything you say at face value. Every single one of your posts is spin.

 

[renderstate]

Apologies for the misquote, I was a genuine mistake (otherwise I wouldn't have linked the original!). Anyway, I see no comments about the rest.

 

[sirmo]

Apologies for the misquote, I was a genuine mistake (otherwise I wouldn't have linked the original!). Anyway, I see no comments about the rest.

The rest of your post is no different, all conjecture.

 

[Sweepr]

The rest of your post is no different, all conjecture.

Still waiting for your arguments. Do you have any evidence that contradicts what he says?

 

[sirmo]

Still waiting for your arguments. Do you have any evidence that contradicts what he says?

Read his post. It's a rant. Zero arguments. There is nothing of substance to respond to.

 

[dogen1]

The rest of your post is no different, all conjecture.

Are you serious right now?

His post was by definition not conjecture.

 

[sirmo]

Are you serious right now?

His post was by definition not conjecture.

We don't even know if NVIDIA was aware this bug was impacting AotS. It could very well be they fixed it because of another title.

Conjecture

 

[dogen1]

Conjecture

Heh.

You didn't even bother looking the definition up?

Nice one.

 

[sirmo]

Heh.

You didn't even bother looking the definition up?

Nice one.

Do you need another example from that post, because I can give you another example from that post.

3) Company A publicly compared their product to company B product using outdated software they shouldn't even be in their possession. The up to date software sits 2 clicks away and it has been available for almost a whole week. This is a fact, not a conspiracy theory. Is this bad? Good? Borderline unethical? I am truly not sure and it would be a nice debate to have if done without preconceived notions and partisanisms. What I am sure is that all the individuals above will reply with more logic defiant arguments, ignoring the facts they don't like and creating new ones out of thin air to make some more baseless accusations.

Presentation like the one Raja gave are planned and rehearsed well in advance. Why would they risk updating the driver days before the presentation? When the graphics and everything else is in place. Those computers weren't even connected to the internet. There are also days lost in travel/packing and unpacking. Tapei is a long flight. In fact AMD_Robert mentioned they were offline, in order to explain that the "leaked" AotS benchmark database links circling were not authentic because they were offline. So yes, an opinion formed from incomplete information is indeed a conjecture.

Also why is the method in which AMD obtained the driver version relevant to the discussion at hand? Does the method change anything about the benchmark results?

It might be an interesting side gossip discussion, but that piece of information has nothing to do with the issue at hand. It's simply an attempt to discredit AMD.

 

[Sweepr]

Presentation like the one Raja gave are planned and rehearsed well in advance. Why would they risk updating the driver days before the presentation?

Can you explain where AMD got that press-only driver? Makes us wonder where that Geforce GTX 1080 came from, if a hardware website 'close' to them provided a review sample.

 

[renderstate]

It's a bit embarrassing because you don't seem to know what the word conjecture means.

 

[sirmo]

Can you explain where AMD got that press-only driver? Makes us wonder where that Geforce GTX 1080 came from, if a hardware website 'close' to them provided a review sample.

No, how should I know?

 

[dogen1]

Do you need another example from that post, because I can give you another example from that post.



Presentation like the one Raja gave are planned and rehearsed well in advance. Why would they risk updating the driver days before the presentation? When the graphics and everything else is in place. Those computers weren't even connected to the internet. There are also days lost in travel/packing and unpacking. Tapei is a long flight. In fact AMD_Robert mentioned they were offline, in order to explain that the "leaked" AotS benchmark database links circling were not authentic because they were offline. So yes, an opinion formed from incomplete information is indeed a conjecture.

https://www.reddit.com/r/Amd/comments/4m692q/concerning_the_aots_image_quality_controversy/

As a parting note, I will mention we ran this test 10x prior to going on-stage to confirm the performance delta was accurate.

Sounds to me like they were finishing it shortly before the presentation.

It might be an interesting side gossip discussion, but that piece of information has nothing to do with the issue at hand.

Agreed.