[AMD_Robert] Concerning the AOTS image quality controversy

[William Gaatjes]

You're just looking in the wrong spots.

http://forums.anandtech.com/showthread.php?t=2475437

At my usual supplier, they are all out of stock. I do not know what the initial stock was, but i can see that the 1080 is sold as soon as it is available.

 

[zinfamous]

I don't get the question about an even playing field. If NVIDIA were using an old AMD driver to perform a public demonstration you could bet this forum would turn into hell on earth, Reddit servers would break down and twitter would shut down to contain the rage of the tinfoil hatters that see NVIDIA conspiracies everywhere. This thread is the perfect example of that.

Is this a troll post?

 

[Yakk]

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?

Nasty one that car emissions bug... why do these bugs happen to the best companies?

 

[boozzer]

Is this a troll post?

ahah :thumbsup: you got good eyes.

 

[zinfamous]

RX480 launch date will be June 29th. The Geforce 1080 has been launched on May 27th. Now lets check availability on nVidias website:
http://www.geforce.com/hardware/10series/geforce-gtx-1080
Out of Stock.
So the gtx 1080 is available only on paper - now THAT is a paper launch.

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.

 

[antihelten]

Is this a troll post?

Basically it is OK for him to speculate about AMD cheating:

They lied about competition not supporting async compute or shader intrinsics so them setting up an unfair comparison wouldn't really surprise me that much.

But as soon as anyone speculates about Nvidia cheating then they're basically brainwashed cultists:

once again we have seen how it's so easy for people to scream NVIDIA cheated without a shred of evidence. Moreover, as you can see in this thread, they won't stop claiming NVIDIA cheated. I'll assure you they are driven by the same mindset of flat earthers and global warming deniers, nothing will change their minds.

So yeah, either a troll post, or the worst case of cognitive dissonance I've seen in a long time.

 

[thilanliyan]

Nasty one that car emissions bug... why do these bugs happen to the best companies?

Because the "best" companies cheat?

 

[Yakk]

Because the "best" companies cheat?

 

[zinfamous]

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?

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.

 

[vissarix]

I remember the same threads on bf4 image quality...back then amd gpus were rendering better images according to few folks...here is the same thing...and knowing how it ended up that case i think same will haoen with this case.

I dont blame these guys tho...i mean nvidia just release 2 gpus which beat amd old gpus on every metric and amd has nothing to fight back...just an upcoming 720p capable gpu that will be on par with 2 year old maxwells on efficiency.

 

[sm625]

...just an upcoming 720p capable gpu that will be on par with 2 year old maxwells on efficiency.

720p. lol. Why stop there. Why not just label it 480p?

 

[crisium]

I'd be shocked if it could handle 16-bit color. Might have to dial-back to 256 colors for Polaris.

 

[Elixer]

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

The 480 is supposed to launch on the 29th of this month, so you can't say it is a paper launch until the 29th--if they have no cards available.

 

[poofyhairguy]

720p. lol. Why stop there. Why not just label it 480p?

Because 720p is half of 1080. Duh.

 

[Erenhardt]

This is how you lock unwanted threads. 10/10

 

[Layer8]

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

That depends on the definition. Lets see...

According to Wikipedia, both don't really qualify as paper launches.

In the Urban Dictionary the definition starts with

A release of a product, especially a computer component, in extremely limited quantities, making it very difficult for consumers to get their hands on.

We don't know for AMD yet (release is at 29th), but 1080 is a paper launch. Especially according to your post.


Anyway. Back to topic:
I don't think that NVIDIA did it on purpose. I don't think it really had a big influence on the performance.
But at least NVIDIA should have come out and said something. I mean other than blaming AMD on using the same drivers the press tests their cards on.

 

[Adul]

The 480 is supposed to launch on the 29th of this month, so you can't say it is a paper launch until the 29th--if they have no cards available.

Typically a paper launch coincided with reviews of the card being released. There are no reviews out there yet for the rx480, they only talked about the card that is comming. Both companies have talked about future products, but they where not launched. Nvidia had reviews for the 1080 and 1070 out before the card was available at retail. That is a paper launch. We have to see if that is the same situation for AMD and the rx480. If reviews are out on the 29th and cards are not in retails stores for purchase, then it will be also a paper launch. Even if it is a few days later.

 

[Adul]

I remember the same threads on bf4 image quality...back then amd gpus were rendering better images according to few folks...here is the same thing...and knowing how it ended up that case i think same will haoen with this case.

I dont blame these guys tho...i mean nvidia just release 2 gpus which beat amd old gpus on every metric and amd has nothing to fight back...just an upcoming 720p capable gpu that will be on par with 2 year old maxwells on efficiency.

Different market segments still. Its a smaller die as well and does not lose performance with Directx 12. Really a great upgrade for users in that price bracket. I know what I am buying for my brother when his birthday comes around in July.

And really Nvidia has nothing to compete with polaris for a much larger market either.

 

[antihelten]

That depends on the definition. Lets see...

According to Wikipedia, both don't really qualify as paper launches.

Actually, if one goes by the wikipedia definition, then both were clearly paper launched:

A paper launch is the situation in which a product is compared or tested against other products of the same kind, despite the fact that it is not available to the public at the time.

Both the 1080 and RX 480 were compared to other products of the same kind (i.e. other GPUs) during their presentations, and of course neither one was available to the public at the time.

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.

 

[zinfamous]

The 480 is supposed to launch on the 29th of this month, so you can't say it is a paper launch until the 29th--if they have no cards available.

well, I also included that with the "products that are delayed and do not release on their intended release day."

Again--I'm simply pointing out that this was a very clear, nonambiguous definition not too long ago, which merely seems to have morphed recently based on current industry practices.

I don't get it.

I'm no fanboy for either company, but it's clear to me that the original definition of "paper launch" remains the best one:

"Today, AMD announces their brand new Hyperreal HD Ultra SuperFX Virgin-destroyer X969000XtX XT GPU capable of crushing all modern games at 260 elite FPS at 16k resolution, heating a small fiefdom for the winter season, and terraforming unclaimed planetoids! Available 3 weeks from now!"

--part announced, unavailable for anyone to purchase on day of announcement = paper launch. Now, I do recall there was a time where these announcements would be made without any real release date mentioned. I would say that is even more of a paper launch. At least AMD has 480s out there/on their way for reviewers. I do think there is a difference there.

"Today, NVIDIA announces Unreal 102950 FXTRXTRHDRFX UltraHighEND SoulCrusher with 9000billion CUDA cores outputting 16trillion, trillion TF and BeyondReal UltraHD 32K processing Platinum! Available now at all major retailers!"

--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.

The latter example is the only one I'm really addressing: claiming something is a paper launch simply because demand is too high for them to get one at that moment is incredibly silly.

In the first example, I think the distinction between the two situations is real (announcing product with future release date vs announcing product with no release date), and more worthy of, well "a discussion," such as it is.

I guess the difference between Sony/Microsoft/other major hardware vendor announcing their next big toy with no real product date, then announcing future release date some time down the road.

 

[Magee_MC]

I don't get the question about an even playing field. If NVIDIA were using an old AMD driver to perform a public demonstration you could bet this forum would turn into hell on earth, Reddit servers would break down and twitter would shut down to contain the rage of the tinfoil hatters that see NVIDIA conspiracies everywhere. This thread is the perfect example of that.

An even playing field for the comparison. The 1080 was benched using the press driver, and other pre-release drivers as well. Since those were the predominant benchmarks available to the public, it made sense to bench the 480 against those numbers. With the release of the 1080 and the new driver, what should happen, but never seems to is that the 1080 pre-release drivers would then be benched against the new driver.

Doing it that way would make it possible for a potential customer to have a accurate representation of the two cards and to see any improvement from the newer driver.

 

[zinfamous]

Typically a paper launch coincided with reviews of the card being released. There are no reviews out there yet for the rx480, they only talked about the card that is comming. Both companies have talked about future products, but they where not launched. Nvidia had reviews for the 1080 and 1070 out before the card was available at retail. That is a paper launch. We have to see if that is the same situation for AMD and the rx480. If reviews are out on the 29th and cards are not in retails stores for purchase, then it will be also a paper launch. Even if it is a few days later.

This is how I see it.

Also, I didn't realize that the 1080 was released that way. I thought it was available the day it was announced (I don't normally pay attention to these cycles.)

 

[Piroko]

But when reading about the 3 different queues in directx 12, that just seem to match the hardware of gcn. DMA engines for copying, graphic command processor for the graphics and 8 asynchronous compute engines (ACE) for the compute tasks.

That's not a coincidence

If i understand it correctly, all these compute threads, would be run by the 8 ACE asynchronously from each other. How much of these async compute threads (I hope that is the correct word for it) can run in parallel on GCN ? 8 ?
And how much can run on Pascal or Maxwell(2) ?

I believe Tahiti from GCN1 and everything from GCN2 onwards has 8 independent ACEs with 8 queues each:
https://commons.wikimedia.org/wiki/File:GCN_command_processing.svg
Not entirely sure if each queue within an ACE can be scheduled asynchronously, I assume they can.

Maxwell also has at least 8 asynchronous compute queues in its GigaEngine I believe, there was an article on AT that went into detail but I can't find it.

I think i understand, they mention Queues and not Queue for each described queue. So there can be for example multiple compute queues that can run in parallel. In a given queue a synchronous command list. But multiple compute queues can all run asynchronously giving asynchronously running command lists with compute tasks Yes ?

You have any number of the queues from that Wikipedia picture running in parallel, Graphics queue 0, DMA queues 0 and 1 as well as Compute queues 0...x. Well, at least in GCN and Pascal I assume. An individual command list should be synchronous within itself.

 

[Adul]

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.

Reviewers need time to bench and write up the reviews. So if a new driver shows up the day or two before the review is released, it may not necessarily make it into the review. As for why AMD used that driver, we can speculate that all day.

But really, if you are busy with setting up a presentation, dealing with other work items, dealing with the trip itself. You are going to miss that a new driver was available for only a few days. There is a lot of work that goes into setting up these presentations, meetings, etc.

 

[Piroko]

"Today, AMD announces their brand new Hyperreal HD Ultra SuperFX Virgin-destroyer X969000XtX XT GPU capable of crushing all modern games at 260 elite FPS at 16k resolution, heating a small fiefdom for the winter season, and terraforming unclaimed planetoids! Available 3 weeks from now!"

--part announced, unavailable for anyone to purchase on day of announcement = paper launch.

This is confusing to me. You mix the terms announcement and release and treat them as the same when they are not.

If I were to take this definition into other fields then:
Every car ever was a paper launch, because they got announced at a motor show with a set release date a couple months later.

This line of thinking is much more logical:
First date: Product roadmap unveil
Second date: Product announcement
Third date: Review embargo (also called NDA) is lifted
Fourth date: Product is released for everyone to buy, also called "launch day".

If the product was officially launched on the fourth date but is sold out within minutes or simply never turned up in stores, then it's a paper launch (probably better to only call it a paper launch if next to no new stock hits the stores within a reasonable time, say, ten workdays. That would be an indicator of low supply). If the fourth date gets pushed into the future time and time again (your Duke Nukem Forever example) then this is called vaporware.
Also, some dates might fall on the same day, that's free for any company to do.