[AMD_Robert] Concerning the AOTS image quality controversy

Page 15 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

selni

Senior member
Oct 24, 2013
249
0
41
It is also possible that some German cars have a software bug that just shows on emission tests? Please think about what you are saying. The next bug will be a reduction of details and no AA regardless what is displayed on settings?

I suppose you missed the part where it made no performance difference? Assuming cheating is nonsensical - there's no motive when it didn't change benchmarks in any measurable way.
 

ElFenix

Elite Member
Super Moderator
Mar 20, 2000
102,393
8,552
126
I know words and phrases morph and change all the time....but why did it only take ~2 years for the meaning of this phrase to completely change?

Paper launch has nothing to do with availability due to demand.

Paper launches only ever referred to a company announcing hardware, then a further release date when that hardware would be available. Meaning--no product on the shelves, for anyone, in any way, on the day of launch. Oftentimes, those products would get delayed and further delayed, sometimes never appearing.

That is all a paper launch means, and all it ever means.

480X is a paper launch because no one can purchase one. Anyone can purchase a 1080--they just have to find it.

It never referred to limited supply issues or demand. Think: Duke Nukem Forever: Decade-long paper launch. :D
no, that's vaporware.

paper launch is something that's been reviewed/tested, but unavailable to buy. they're related concepts, often overlapping, but different.

i would include a trickle of shipments as a paper launch, but not vaporware.

that's also different from a demand tidal wave situation, like when new consoles launch.

480X isn't a paper launch because it hasn't been launched yet. it may be vaporware if it were to get pushed back several months.
 

zinfamous

No Lifer
Jul 12, 2006
111,700
31,059
146
Ah yes, vaporware. I had forgotten that expression--I only look into this sector every ~3-4 years.

I concede defeat in this corner of the discussion.

Now, everyone please continue to argue about nefarious business practices and 3rd-front maneuvers pulling the respective strings of your cherished silicon processing chip companies.
 

Ranulf

Platinum Member
Jul 18, 2001
2,825
2,425
136
Sorry what was I thinking, I should just read reviews and not play the actual games with my video card...

Good God man! How can you run that space heater cpu or gpu!? A gpu with a hair dryer loud cooler! Just play actual games man! And heat your house! Tee hee!

Perf/watt ftw! Raw perf ftw! Tastes great! Less filling! Save the earth, use less power! SLI FTW!
 

Bacon1

Diamond Member
Feb 14, 2016
3,430
1,018
91
I suppose you missed the part where it made no performance difference? Assuming cheating is nonsensical - there's no motive when it didn't change benchmarks in any measurable way.

Has anyone shown that its both actually fixed (IQ tested) and no performance hit?

I linked someone with a 1080 that said they didn't see it as fixed with the newer drivers.
 

selni

Senior member
Oct 24, 2013
249
0
41
Has anyone shown that its both actually fixed (IQ tested) and no performance hit?

I linked someone with a 1080 that said they didn't see it as fixed with the newer drivers.

Shrug, here's Ryan Smith (anandtech editor) retesting performance and finding no difference.

https://forum.beyond3d.com/threads/nvidia-pascal-reviews-1080-and-1070.57930/page-17#post-1919727

Here's Dan Baker (oxide) posting that it should be fixed in 368.25. Make what you will from should there, but presumably he wouldn't say that without some expectation something was fixed.

https://twitter.com/dankbaker/status/739880981612625920
 

Adul

Elite Member
Oct 9, 1999
32,999
44
91
danny.tangtam.com
Good God man! How can you run that space heater cpu or gpu!? A gpu with a hair dryer loud cooler! Just play actual games man! And heat your house! Tee hee!

Perf/watt ftw! Raw perf ftw! Tastes great! Less filling! Save the earth, use less power! SLI FTW!

I have solar, so I can use as much as I can produce :)
 

VulgarDisplay

Diamond Member
Apr 3, 2009
6,188
2
76
There is nothing going in here other than nvidia's inferior drivers.

In my opinion, amd has the better driver team these days. My gtx780 gave me so many problems in the games I wanted to play that I returned it for a r9 290.

Nvidia didn't do it on purpose they are just getting more and more incompetent. I mean look at how much better amd cards are than their direct competition at launch. It really highlights how hard it is to change perceptions once they have been accepted as fact. People still parrot that amd has bad drivers which hasn't been the case for 3 years, when in actuality nvidia's generation old cards fall off a performance cliff and they can't get games to render properly.

Sent from my XT1575 using Tapatalk
 

Bacon1

Diamond Member
Feb 14, 2016
3,430
1,018
91
Shrug, here's Ryan Smith (anandtech editor) retesting performance and finding no difference.

https://forum.beyond3d.com/threads/nvidia-pascal-reviews-1080-and-1070.57930/page-17#post-1919727

Here's Dan Baker (oxide) posting that it should be fixed in 368.25. Make what you will from should there, but presumably he wouldn't say that without some expectation something was fixed.

https://twitter.com/dankbaker/status/739880981612625920

Right, just saying it should be doesn't mean it is. Maybe Nvidia told him it was fixed. I'm asking for anyone to verify that it is actually fixed. Because so far, the only one that has tested it and looked at the IQ has said that it isn't fixed. So either its not working for him still, or it isn't actually fixed.

Also, why is he posting on those forums and not his own? Can you ask him to check the IQ?
 
Last edited:

Elixer

Lifer
May 7, 2002
10,371
762
126
Right, just saying it should be doesn't mean it is. Maybe Nvidia told him it was fixed. I'm asking for anyone to verify that it is actually fixed. Because so far, the only one that has tested it and looked at the IQ has said that it isn't fixed. So either its not working for him still, or it isn't actually fixed.

Also, why is he posting on those forums and not his own? Can you ask him to check the IQ?
There used to be IQ tests all the times back in the day, then they stopped doing that since they weren't seeing much differences. Now, seems they need to start doing them again to keep everyone in check.
That is one good thing about "late" reviews, you have a longer time to do a more thorough job--Ryan's reviews are still missing, so, we can assume he is doing more tests.

However, things still slip by like the 970 fiasco, or the "compute engines" chart that was never fixed (or...), so, seems PR "fluff" is the main winner here, tote the line or we won't send you review cards.

BTW, Ryan does sporadically post here, but the troll ratio is less over there.
 

sirmo

Golden Member
Oct 10, 2011
1,014
391
136
There needs to be an ACID test for GPUs. Because visually inspecting scenes isn't really viable.
 

Sonikku

Lifer
Jun 23, 2005
15,882
4,882
136
Ah yes, vaporware. I had forgotten that expression--I only look into this sector every ~3-4 years.

I concede defeat in this corner of the discussion.

Not good enough. Not BLOODY good enough!

Eat cat food for your insolence and your error!
 

3DVagabond

Lifer
Aug 10, 2009
11,951
204
106
Here we go. As I wrote multiple times people making accusations without proof should have waited for someone to check the bad driver was responsible for improved perf. This was a genuine driver bug.. not a surprise for whoever could use a little bit of logic and rational thinking.

So now we just have to find out why AMD would broadcast worldwide a comparison made with a driver they shouldn't even have (who gave it to them and why?) instead of using the most recent and up to date driver that had been available for 6 days before their public demo.

I am not saying they did it on purpose, mistakes do happen, but they should apply their GPUOpen philosophy to this event as well :)

And what about the people who assumed AMD was rendering at lower settings? That was totally not true. It was actually rendering correctly. Any of them say they were making something out of absolutely nothing?
 

3DVagabond

Lifer
Aug 10, 2009
11,951
204
106
Regardless of performance;I think the takeaway here is there should NOT be "press only" drivers which consumers do not have access to. It breeds mistrust and opens the door wide open to potentially shady activities.

If not this time, maybe the next?

^^^This.

It would be too easy to have drivers optimized differently for reviews returning better results than what consumers would have.
 

3DVagabond

Lifer
Aug 10, 2009
11,951
204
106
Yet another straw man. Can you walk in a store and buy a 1070? No.
Could AMD walk in a store buy a 1080 and therefore install the release driver at the time of the demo. Yes.

Reviewers use the best driver available at review time. AMD didn't use the best driver at the time of their demo, despite it being available for several days.

Each time you repeat this it's the same. Find a new point. You've beaten this one enough.
 

3DVagabond

Lifer
Aug 10, 2009
11,951
204
106
well, you have to ask yourself this: have you ever experienced a buggy driver in your x number of years using GPU products from various companies?

Ever? Is it entirely possible that brand new driver releases occur with specific bugs, some of which often go unnoticed and unaddressed?

I guess if that has never happened in the history of the industry, then it is reasonable to assume that NVIDIA released this evil performance-hiding driver explicitly to reviewers in order to hide their inferior performance....and their devil worship and of course their secret Illuminati backing.

...but that's only fair because AMD, obviously a puppet of the Lizard People, stole this driver under NDA from some crooked spy in an underhanded attempt to discredit the angels at NVIDIA and make their product out to be a low-rent toaster.

Just like we've had drivers with bugs we've also had drivers that cheated on image quality. Both are equally possible. Add to that this whole issue came to light because nVidia supporters accused AMD of cheating by running lower settings. Reality it was the nVidia demo that was rendering less. At that point it's perfectly understandable for turnabout on the issue.
 

Layer8

Member
May 3, 2016
43
0
6
BTW it's kind of funny looking at the edit history on wikipedia, and seeing how someone tried to rewrite the definition back on the 27th.
Seems like someone tried to push the Urban Dictionary's definition to Wikipedia.

--That is not a paper launch. Even if stock sells out for the next 5 months despite peak supply--high demand and all that--it is not a paper launch because product is out and available to purchase. Plenty of people have it, because it is so popular.
You got it the wrong way. When NVIDIA says their products will be available at 27th it is not yet a paper launch. They are just announcing the launch of their product at a certain date.
When, however, there are not enough products to satisfy demand at that date and miss that demand by large margins it is a paper launch. The contrary would be a hard launch.
When they have a lot of supply and still cannot meet demand, the retailers will increase the prices accordingly (see 6700k).
 

antihelten

Golden Member
Feb 2, 2012
1,764
274
126
You got it the wrong way. When NVIDIA says their products will be available at 27th it is not yet a paper launch. They are just announcing the launch of their product at a certain date.
When, however, there are not enough products to satisfy demand at that date and miss that demand by large margins it is a paper launch. The contrary would be a hard launch.
When they have a lot of supply and still cannot meet demand, the retailers will increase the prices accordingly (see 6700k).

Again that depends upon which definition you go with. The one you're describing here is the urban dictionary one.

Although I suppose one could say that both situations counts as paper launch, seeing as both have been described as such for the last 10 years or so (it's seems like it's only with the 1080 launch that people are trying to come up with a new third definition).

In other words a paper launch is either the situation where a company compares an upcoming product to other products before it's available to buy, or the situation where a product is launched to the public in extremely limited quantities just so the company can claim "first".

The two aren't mutually exclusive after all.
 

ElFenix

Elite Member
Super Moderator
Mar 20, 2000
102,393
8,552
126
Now, everyone please continue to argue about nefarious business practices and 3rd-front maneuvers pulling the respective strings of your cherished silicon processing chip companies.

:D
 

Bacon1

Diamond Member
Feb 14, 2016
3,430
1,018
91
So has anyone confirmed that the newest Nvidia drivers actually fix the issue?
 

swilli89

Golden Member
Mar 23, 2010
1,558
1,181
136
Sorru if I missed it but has anyone spotted new benchmarks with nVidia 368.25 drivers?