Initial experience going 295x2 quadfire to 980Ti SLI at 4K

[RaistlinZ]

I wouldn't put anything past Nvidia. If they think they can get away with something, I'm sure they'll try. "Fool me once..." as the saying goes.

I wish there was a 3rd competitor too.

 

[Silverforce11]

Although GTA5 while using more memory with the 980Ti's seems to lack the middle distance graphical fidelity on other cars and pedestrians that the 295x2s gave me. I don't 'swoon' when playing like I did with the 295x2s because everything not in the immediate foreground doesn't look as polished -much more like watching a PS4 version. Will have to fiddle with settings some more as GTA 5 definitely doesn't look as nice on exact same game settings as if middle distance textures have been downgraded.

Here's the same run through video (same capture quality & Ultra settings):

https://www.youtube.com/watch?v=FBijzEQ6x6g

http://forums.overclockers.co.uk/showpost.php?p=28258333&postcount=2391

 

[Silverforce11]

Considering the performance cost of Anisotropic Filtering in modern GPUs is basically 0, I can't imagine Nvidia would intentionally degrade their filtering, as it's not going to result in any measurable gain in FPS. It's more likely a driver issue or a problem specific to filtering within a single game. Regardless, a jpg from a compressed YouTube video is probably the least effective way to measure such discrepancies as there are so many other quality issues that can be introduced along that path. Let's see some uncompressed screenshots from the source.

It's not just AF lacking in that example. It's the LOD cut-off distance, very short compared to AMD which extends all to the end of the view range on Ultra settings (which is specific to Ultra, it cuts off earlier on High).

Then there's the lack of AO, shadows are very bland without depth.

So what we have on AMD is clearer, better lighting & better shadows, with much longer LOD viewrange. In BF4, that's basically the only difference between High & Ultra.

If it's a bug, it needs to be fixed because people expect Ultra = Ultra and not High. If it isn't a bug and a "feature", then LOL @ cheating scum.

 

[Silverforce11]

Same thing in Project Cars:
https://www.youtube.com/watch?v=wyYeT9Wvy1A

Poor lod distance and poor texture filtering & lighting.

One game is a bug.

 

[Udgnim]

do review sites compare render quality anymore?

 

[Hitman928]

do review sites compare render quality anymore?

They use to all the time, it's pretty rare now though.

 

[Silverforce11]

do review sites compare render quality anymore?

Guess they don't do anymore so whatever driver hacks are done to IQ, they just getting away with it.

 

[Hitman928]

Guess they don't do anymore so whatever driver hacks are done to IQ, they just getting away with it.

We're still not sure if it is a bug or not though. If there's no performance gain, then it's something I'm sure Nvidia will want to correct.

 

[amenx]

It's not just AF lacking in that example. It's the LOD cut-off distance, very short compared to AMD which extends all to the end of the view range on Ultra settings (which is specific to Ultra, it cuts off earlier on High).

Then there's the lack of AO, shadows are very bland without depth.

So what we have on AMD is clearer, better lighting & better shadows, with much longer LOD viewrange. In BF4, that's basically the only difference between High & Ultra.

If it's a bug, it needs to be fixed because people expect Ultra = Ultra and not High. If it isn't a bug and a "feature", then LOL @ cheating scum.

This is what I believe may happen in a few games where anistropic filtering doesnt work properly when set to 'application controlled' with Nvidia:

Solution AF = x16. Almost nil impact on performance. Texture filtering though does have a slight impact on performance though when going from default 'quality' to 'high quality' (lower scores in 3dmark runs).

 

[Tapoer]

From Gregster video is clear that on Nvidia, AF isn't working and textures have lower quality, this have to effect performance.
And it seems that this happens not just in BF4.

LOD is also the reason that textures don't shimmer on Nvidia because they have a reduced LOD, while on AMD they use the LOD that the aplication requests.
It's like Nvidia have the AMD's AI option enabled, but there is no way to disable it.

So over this last years the GPU reviews are all wrong and deceiving.
Probably no tech site will pick this up and make a article, they will be afraid of Nvidia hammer. Money talks and we will not get the true.

 

[vissarix]

From Gregster video is clear that on Nvidia, AF isn't working and textures have lower quality, this have to effect performance.
And it seems that this happens not just in BF4.

LOD is also the reason that textures don't shimmer on Nvidia because they have a reduced LOD, while on AMD they use the LOD that the aplication requests.
It's like Nvidia have the AMD's AI option enabled, but there is no way to disable it.

So over this last years the GPU reviews are all wrong and deceiving.
Probably no tech site will pick this up and make a article, they will be afraid of Nvidia hammer. Money talks and we will not get the true.

Admirable the effort you guys put on for so many years :biggrin:

i mean its not easy to find excuses and believing blindly on your conspiracy theories for that long...you should all get a reward.

 

[greatnoob]

Mobile game devs usually fallback to 16bit/short values when programming shaders since it provides a good boost in performance at the cost of lower IQ. I sure hope Nvidia's driver updates don't consist of these sorts of hacks but then again 16 bit is too limited while 32 bit is pretty much always overkill and values are rounded or clamped in a lot of instances so who knows what's going on - they certainly won't be able to use shorts everywhere so I doubt they would do something like this unless they have some ingenious compression techniques or estimation tables to which everyone in the dev industry would applaud Nvidia.

 

[Tapoer]

Admirable the effort you guys put on for so many years :biggrin:

i mean its not easy to find excuses and believing blindly on your conspiracy theories for that long...you should all get a reward.

So everyone who is reporting a lower IQ when changing from AMD to Nvidia are lying, blind, or is all on their heads?

 

[Flapdrol1337]

So everyone who is reporting a lower IQ when changing from AMD to Nvidia are lying, blind, or is all on their heads?

No, it's because you said:

So over this last years the GPU reviews are all wrong and deceiving.
Probably no tech site will pick this up and make a article, they will be afraid of Nvidia hammer. Money talks and we will not get the true.

which is silly.

 

[Tapoer]

No, it's because you said:

which is silly.

Yes, I was rush with that conclusion, I will be glad to be proved wrong, but only time will tell.

 

[beginner99]

From Gregster video is clear that on Nvidia, AF isn't working and textures have lower quality, this have to effect performance.
And it seems that this happens not just in BF4.

It'a at least interesting because Fury x loses big in BF4 and in general NV fares much better in BF4.

 

[ShintaiDK]

It'a at least interesting because Fury x loses big in BF4 and in general NV fares much better in BF4.

I think something is wrong with the bencher. Because looking on the FPS the Titan X is 30-50% faster than Fury X in BF4. While professional reviews is 10-20%.

Thats why we trust professional reviews with people knowing what they do.

 

[Aristotelian]

Admirable the effort you guys put on for so many years :biggrin:

i mean its not easy to find excuses and believing blindly on your conspiracy theories for that long...you should all get a reward.

Objective differences in image quality (I watched the videos - it's clearly perceptible even with my eyes, which will soon be upgraded to 20/20) are "conspiracy theories"?

Do you have 295x2 quadfire and 980Ti SLI to benchmark and record? Or are you calling the OP a liar as well for his/her referred perception?

What's your explanation?

 

[Ire123]

I think something is wrong with the bencher. Because looking on the FPS the Titan X is 30-50% faster than Fury X in BF4. While professional reviews is 10-20%.

Thats why we trust professional reviews with people knowing what they do.

Agreed. Also the fact that he hasn't posted still screenshots yet, as many people have requested but is more then willing to do these horrible quality vids. Seems like click bait. Almost 2k views on his vid already. He hasn't posted shots of his CC or NV panel either. For all we know he could have set it to override application settings - 0x af for the TX.

 

[railven]

So everyone who is reporting a lower IQ when changing from AMD to Nvidia are lying, blind, or is all on their heads?

I can't respond for other guys, but in the games I play the most, I saw zero change or improvements (WoW specifically looks better on Nvidia hardware. There was a shadow issue on Radeons that I reported to AMD, think it was corrected but I haven't had a Radeon in 2 years, should have checked when I had those 290Xs).

These aren't small little blemishes, which is why I'm sure a reviewer *SHOULD* have caught this by now IF it is an ongoing issue and not say a driver bug or something else related to game files.

Time to see if reviewers pick up on it and investigate. Nothing like getting some good ol egg on yo face to make you try harder!

 

[5150Joker]

I think something is wrong with the bencher. Because looking on the FPS the Titan X is 30-50% faster than Fury X in BF4. While professional reviews is 10-20%.

Thats why we trust professional reviews with people knowing what they do.

Pretty much this but you know, certain people will latch on to this because there's nothing else left for them after the fury disaster.

 

[railven]

So I'm reading that threaded linked, interest new info (like he used the wrong setup for the Titan X capture). Anyways, one thing that keeps coming up is that the Fury X has better color, well...the poster who captured said he tweaked the color for Fury X not Titan X (wonder what it looked like before) but most importantly:

Yeah, that's black crush. Objectiveness on the settings - black crush is notorious for creating "detail" where there is none. Examples of Black Crush:

A lot of people will swear the right side of that image looks "better." Going as far as saying it has more detail. It's sharped out of it's mind which makes edges look better, but then of course you loose a lot of detail.

Anyways, still reading the thread but that image from the BF4 run screamed black crush at me. You can't even see the lines on the column on the Fury X side.

EDIT: Someone else got a better screen shot, yeah dude the Fury X side is getting some serious black level crushes. It probably has to do with the guy who tweaked the color settings.

The PS3/PS4 got slammed until the consoler peasants learned what black crush was and how it adversely affected visual fidelity.

 

[railven]

Okay, so this is why you can't use this video to make a comparison ESPECIALLY after the original creator openly said he tweaked the video for one side.

Look at this image:

The things noted.
EDIT: Man, you can see the black crush right in this image too. Look between the two biggest purple circles. On the Titan X there are whitish markings on the curb right between the two purple circles. THAT IS GONE on the Fury X side, completely. Not "blurred" or "faded" completely gone. And yet whoever made this image didn't notice that? At all?

This image:

Because the image is so crushed on the Radeon side, the white lines on the second image look more prominent. Ie, "faded" on both images for the Titan X - or as the circle tries to imply "blurred"

The circle implies the textures are "underneath" the line, they are more likely on top of the line, and the crushed side the line is so "bright" it blends the textures together. Look at the second image, on the left side the black is so dark it obscures the details in the building. Black crush you don't just suffer lost of dark levels you also suffer loss of white levels. So something white and something off-white both appear the same shade of white.

Yeah the green circle of the top image, again crushing the blacks bolds white, and makes the "white" image appear darker or more detailed/prominent.

Woof, this guy BUTCHERED this video. The fact that people posting this here DIDNT include the fact that HE ALTERED THE FURY X VIDEO is very disingenuous, especially when you're going to try to claim IQ differences. You can't compare IQ if one side is altered.

And it seems the same issues persist through his other videos. He needs to present unaltered videos.

 

[railven]

Wow, barely anyone in that thread makes note the videos were altered thus not 1:1 comparison. I guess these guys also think LCD screens > Plasma. Woof. I forgot not all gamers actually really look at color levels too closely. If they think the Titan X side is washed out based on the Fury X having crushed blacks they should have NO SAY in IQ comparatives. Woof. Why does this feel like a console wars thread all of a sudden? :awe:

And in case you guys missed it, since people reporting here from that thread sure didn't bother telling you:
http://forums.overclockers.co.uk/showpost.php?p=28258333&postcount=2391

@ Gregster:

If Titan X and Fury X were at same setting in your video, why is there detail missing from Titan X version? Plus draw distance looks reduced..

http://forums.overclockers.co.uk/showpost.php?p=28258345&postcount=2394

No idea. I set the colours up for the Fury X but didn't for the TX as I was just doing a quick run to compare frames. Both had the exact same settings and no shenanigans at all.

EDIT:

Wow, thanks for this Boomstick/Greg. The AMD FURYX has a much higher quality image - glad I didn't go for a 980ti now

Hopefully the big review websites will start testing this and expose NVIDIA's cheating once more.

HAHAHAHA.

 

[railven]

Even with this glaring info, people still don't realize the issue.

They even made a thread for this issue and not a single person mentions the crushed blacks or altered video. I got half the mid to start an account there and post about it.

Woof. This is now turning into a "find something" hunt. Haha. I'll need to keep this thread watched because the magician who made the first video that created this cluster-funk is gonna capture a NEW video with SCREEN SHOTS. I hope he alters them too or he'll end up with so much egg on his face.
http://forums.overclockers.co.uk/showthread.php?t=18679713

<popcorn.gif>

Ayeee dios mio.

If I had to give a reason, it would be the new fangled Nvidia colour compression algorithms they're using to get round the relative weakness of Maxwell's memory bandwidth.

Possibly...maybe ????

AMD have the same tech and use it.

I suppose it can cause washed-out colour and less sharp textures but not to the extent apparent in the side by side Youtube, youtube is not ideal i know.

Delta Colour compression and Texture compression is something that is a part of Cryengine, the Delta Colour difference between say 2x and 8x can hardly be seen with the naked eye.

Edit; it wouldn't explain missing details, missing objects...

Please, someone here who posts there post in that thread that Gregster altered the Fury X video, not realizing he crushed the black levels and created a console war topic for them. Now they're just looking for a cause when the cause was the original creator.