Radeon RX 480 "Polaris" Launched at $199

[karik666]

I dont think Amd are doing themselves any favors with having such a long Nda.

 

[h4rm0ny]

They could have simply done what nvidia did when they announced the 1080 and there was no completion - compare it to the existing cards in the lineup to show improvements.

Meanwhile in an alternate timeline, Alt-D.P. is calling AMD out for their marketing fail of not comparing their current generation to Nvidia's current generation.

But the whole AotS bench was a complete farce. Not only did AMD have no idea what they are talking about wrt 51% utilisation and had backtrack and correct themselves, but the whole driver fiasco is very suspicious.

Hardly backtracking. Somebody on Reddit asked for clarification on what the utilization number meant and AMD provided it. Did they contradict themselves? No. Did anything they said turn out to be false? No. Backtracking indeed! As to "fiasco", if that's the term you break out for a driver issue which in all probability was in Nvidia's favour anyway, and was a result of Nvidia sending out different drivers to the press than they release to customers, then I don't know that you'll have any suitable terms left for when something actually problematic happens. TL;DR: Not a "fiasco".

Turns out the driver AMD used was a pre-release review driver not for public circulation. The driver was only allowed to be used by reviewers under NDA.

I'm frankly not very impressed by the way you try to spin this situation into a criticism of AMD. So it turns out Nvidia send out special "press drivers" to reviewers different from those the customer is using, and that's a criticism of AMD? AMD use the press driver which in all probability makes Nvidia look better than if they'd used the release driver, and that's a sign of ill-intent by AMD? Who knows who provided the driver to them. Maybe Nvidia will sue them... But I suspect not.

AMD illegally used a driver that happened to have a bug. They could have, indeed should have, used the release driver which had the bug fixed. The issue is Oxide then make innaprriate statements about the driver buf affecting performance leading to all sorts of tin-foil conspiracy garbage about Nvida cheating. Oxide could have simply used the actual release driver and verified that with the bug fixed there was no performance change, but then AMD wouldn't have got their publicity stunt.

This is not a "publicity stunt" by AMD. I can see the boardroom now... "I've got a great idea, lets use a specially optimized driver by Nvidia that performs a little better than the current public one and then later on we can be called out on it by Nvidia even though that makes no sense. It will be a wonderful publicity stunt.".

Or not.

 

[D.P.]

An NDA that had expired and the very driver that NVIDIA wanted reviewers to use. Oh my, how shocking to then use it as a comparison base. As opposed to the drivers released shortly before the demo - cause setting thing s up at the last minute is a good idea. Also what law do you think was broken?

The problem is AMD had no legal access to that driver, It has nothing to do with NDA expiration. There were release day drivers that AMD should have used and avoided the entire debacle, but why would they when they can create all this free publicity. The entire benchmark was very last minute because it was released within days of the 1080 being made public, in fact it is questionable if they even used a retail card but a reviewer illegally provided a review card and the review driver.

None of that is that important and I doubt nvidia will do anything about it. The real issue is oxide making unsubstantiated comments about performance when they could have used the release driver and noted no performance discrepancy.

 

[Ferrari100]

The problem is AMD had no legal access to that driver, It has nothing to do with NDA expiration. There were release day drivers that AMD should have used and avoided the entire debacle, but why would they when they can create all this free publicity. The entire benchmark was very last minute because it was released within days of the 1080 being made public, in fact it is questionable if they even used a retail card but a reviewer illegally provided a review card and the review driver.

None of that is that important and I doubt nvidia will do anything about it. The real issue is oxide making unsubstantiated comments about performance when they could have used the release driver and noted no performance discrepancy.

If I saw Linus, for example, tomorrow and he gives me information on the RX480 without then subsequently I tell everyone about it, I am not the one who is breaking agreements/laws, Linus is.

Same as when a mole leaks NDA info to Nvidia about AMD - It's not Nvidia breaking the law but the mole who leaks it.

Problem is, you have to prove who the mole is.

 

[five8five]

No laws are broken except for by the person who handed the driver to AMD, if it was still under NDA.

Nvidia would happily use similar tactics if they had the chance to expose AMD for releasing 'press only' drivers to make their own products look better to unsuspecting consumers.

I thought the 'press only' drivers were essentially Beta drivers? It was silly of AMD to use them and publicly release the results. They should've just used fully released drivers.

 

[D.P.]

Meanwhile in an alternate timeline, Alt-D.P. is calling AMD out for their marketing fail of not comparing their current generation to Nvidia's current generation.

Are you just trolling, no where have I said that is what AMD should have done.

Hardly backtracking. Somebody on Reddit asked for clarification on what the utilization number meant and AMD provided it. Did they contradict themselves? No. Did anything they said turn out to be false? No. Backtracking indeed!

Complete rubbish, as I've come to expect from you. Yes they back Gracie,d yes they contacted themselves. They first claimed 51% GPU utilization, then they said 51% xfire scaling! then they corrected themselves again and stated 83% scaling and the 51% was the time when GPU bound under the lowest settings and the actual utilization is completely unknown.

That is the definition of a fiasco.

As to "fiasco", if that's the term you break out for a driver issue which in all probability was in Nvidia's favour anyway, and was a result of Nvidia sending out different drivers to the press than they release to customers, then I don't know that you'll have any suitable terms left for when something actually problematic happens. TL;DR: Not a "fiasco".

I'm frankly not very impressed by the way you try to spin this situation into a criticism of AMD. So it turns out Nvidia send out special "press drivers" to reviewers different from those the customer is using, and that's a criticism of AMD? AMD use the press driver which in all probability makes Nvidia look better than if they'd used the release driver, and that's a sign of ill-intent by AMD? Who knows who provided the driver to them. Maybe Nvidia will sue them... But I suspect not.

The release day drivers were not ready to send to reviewers. The release day drivers have bug fixes, as seen in AotS. Nvidia don't have a time machine, they can't go into the future to pick up. Future driver and take it back to the past to give to reviewers.


I'm not criticizing AMD, but the interaction between AMD, oxide and the public. AmD had no legal access to the review drivers but that isn't a big deal. But it does indicate they also had illegal access to a review 1080 sample, because the retail cards shipped with the release drivers with the big fixed. The problem is before making any far fetched accusations oxide could a be simply downloaded the latest drivers from Nvidia's website,mor you know, installed the drivers that cam with a retail card.

This is not a "publicity stunt" by AMD. I can see the boardroom now... "I've got a great idea, lets use a specially optimized driver by Nvidia that performs a little better than the current public one and then later on we can be called out on it by Nvidia even though that makes no sense. It will be a wonderful publicity stunt.".

Or not.

The release day driver actually performs better than the reviewer driver.
But no, I don't think this is planned stunt at all, it is just AMD jumping the gun and using drivers and quite possibly a card they shouldn't have had access to, and oxide making extremely unprofessional comments

 

[fs123]

The problem is AMD had no legal access to that driver, It has nothing to do with NDA expiration. There were release day drivers that AMD should have used and avoided the entire debacle, but why would they when they can create all this free publicity. The entire benchmark was very last minute because it was released within days of the 1080 being made public, in fact it is questionable if they even used a retail card but a reviewer illegally provided a review card and the review driver.

None of that is that important and I doubt nvidia will do anything about it. The real issue is oxide making unsubstantiated comments about performance when they could have used the release driver and noted no performance discrepancy.

You and Nvidia cannot do anything except whinge about AMD using drivers which were given to press only. Whoever gave them to AMD would have not signed any NDA otherwise they wouldn't have risked giving them out.

If Nvidia is using alternate drivers for the press then it speaks volumes. Surely they haven't tuned them for benchmarks ?

 

[Ferrari100]

AOTS benchmark clearly getting to people.

I cant wait to see June's update as it should absolutely be utilizing the latest tech on both the 480 and GTX1080. It will give us a good awareness of the direction both AMD and Nvidia are heading with there respective DX12 implementations.

 

[D.P.]

If I saw Linus, for example, tomorrow and he gives me information on the RX480 without then subsequently I tell everyone about it, I am not the one who is breaking agreements/laws, Linus is.

Same as when a mole leaks NDA info to Nvidia about AMD - It's not Nvidia breaking the law but the mole who leaks it.

Problem is, you have to prove who the mole is.

No, AMD are guilty here of using something which they had no legal right to use.


I really don't think that is a big issue, but that action is what resulted in all the following fiasco.

 

[Ferrari100]

No, AMD are guilty here of using something which they had no legal right to use.


I really don't think that is a big issue, but that action is what resulted in all the following fiasco.

How do they know they have no legal right to use it?

You now say it's not a big issue. OK THEN.

 

[D.P.]

How do they know they have no legal right to use it?

You now say it's not a big issue. OK THEN.

Because they accessed it through illegal means. The review driver was given to reviewers only, the retail cards had retail drivers with the big fixed.


There are 2 points I have made:
1) it is fairly dubious how AMD got their hands on a retail 1080 and used reviewer drivers and not retail drivers. I don't think this really matters, they just wanted a review as soon as possible, or wanted to use the same drivers as reviewers.
2) oxide made extremely unprofessional comments about performance due to a bug in the review drivers. Oxide, if they installed the retail drivers would have observed the bug to be fixed without a performance drop, actually a performance gain if anything. This is extremely dishonest.


Issue 2 never would have happened if it wasn't for issue 1 though.


To be clear, I am not saying that the result AMD presented were flawed, or that they purposely tried to confuse anyone, or planned any kind of anti-nvidia defamation. AMD almost certainly had no control over oxides statements, but neither oxide or AMD tried to make the record straight.

AMD just enjoyed the publicity. It was effective, we see right in this forum people claiming nvidia sent reviewer drivers with cheats to improve performance in AotS when the reality is a simple driver bug, a bug that doesn't effect performance to any significant degree.

 

[SiDeards73]

Dunno why you lot pander to him, just ignore him, he wont go away i know, but it might stem the amount of drivel he spews.

So the PC thing at E3 is on Monday and should be at a time we can all view it right? will be interesting to see if anything is revealed.

 

[agent00x]

No, AMD are guilty here of using something which they had no legal right to use.


I really don't think that is a big issue, but that action is what resulted in all the following fiasco.

no LEGAL right? Can you point me to what law they are breaking exactly?

 

[fs123]

no LEGAL right? Can you point me to what law they are breaking exactly?

The law of fanboy but*hurt I think. Totally a no no for someone to use a driver for a gfx card even though there was never an NDA for it.

 

[bru]

If I saw Linus, for example, tomorrow and he gives me information on the RX480 without then subsequently I tell everyone about it, I am not the one who is breaking agreements/laws, Linus is.

Same as when a mole leaks NDA info to Nvidia about AMD - It's not Nvidia breaking the law but the mole who leaks it.

Problem is, you have to prove who the mole is.

+1 this is it exactly.

So on that front can we stop blaming NVidia, as they had nothing to do with it, they just made the damn hardware and driver, they certainly didn't run it for AMD to use in their presentation.

 

[AlamoX]

Because they accessed it through illegal means. The review driver was given to reviewers only, the retail cards had retail drivers with the big fixed.


There are 2 points I have made:
1) it is fairly dubious how AMD got their hands on a retail 1080 and used reviewer drivers and not retail drivers. I don't think this really matters, they just wanted a review as soon as possible, or wanted to use the same drivers as reviewers.
2) oxide made extremely unprofessional comments about performance due to a bug in the review drivers. Oxide, if they installed the retail drivers would have observed the bug to be fixed without a performance drop, actually a performance gain if anything. This is extremely dishonest.


Issue 2 never would have happened if it wasn't for issue 1 though.


To be clear, I am not saying that the result AMD presented were flawed, or that they purposely tried to confuse anyone, or planned any kind of anti-nvidia defamation. AMD almost certainly had no control over oxides statements, but neither oxide or AMD tried to make the record straight.

AMD just enjoyed the publicity. It was effective, we see right in this forum people claiming nvidia sent reviewer drivers with cheats to improve performance in AotS when the reality is a simple driver bug, a bug that doesn't effect performance to any significant degree.

deflecting wont change much, eveyone knows corp spying is part of the business, claiming Nvidia doesn't spy on AMD would be hypocritical and stupid, and the other way around is true also, trying to blow it out of proportion wont help take focus away from the danger of having reviewers only driver with renderer issues that never makes it to the public.
and this is not about blaming Nvidia, i would rather see this practice banned from both AMD and Nvidia, ignoring it or deflecting it wont help solve it.
without brining me the ready or not yet ready, D.P do you agree for reviewers to have their own drivers that never makes it to the public ?

 

[D.P.]

deflecting wont change much, eveyone knows corp spying is part of the business, claiming Nvidia doesn't spy on AMD would be hypocritical and stupid, and the other way around is true also, trying to blow it out of proportion wont help take focus away from the danger of having reviewers only driver with renderer issues that never makes it to the public.
and this is not about blaming Nvidia, i would rather see this practice banned from both AMD and Nvidia, ignoring it or deflecting it wont help solve it.
without brining me the ready or not yet ready, D.P do you agree for reviewers to have their own drivers that never makes it to the public ?

You mistake long seem to think there is something special about the reviewer drivers. The reviewer drivers are just beta drivers, so have bugs and gave not been fully tested.

As I said above, Nvida don't have a the machine. They have ability to go in to the future and get release drivers and bring them back in time to give to reviewers. The review drivers are not released to the public because they contains details under NDA, like different models and clock speed info etc.

Reviewers are free to download the release drivers to check performance anomalies. Most likely they will find improved performance and reduced bugs, that is what happens with newer drivers.

There is nothing explicitly against nvidia releasing the review drivers but they are kind of useless since they have been surpassed with superior drivers with less bugs.



I would absolute be against nvidia or AMD providing special reviewer drivers that gave anomalous performance but that isn't the case. Providing reindeers with a non-public driver under NDA to prevent sensitive I formation leaking to the public is absolutely within their rights.

 

[Lokken86]

We are moving away from using Nvidia as a comparison to just an Nvidia thread again.

 

[RynceWynd]

RX480, Do we think they will do an ITX version?

 

[D.P.]

RX480, Do we think they will do an ITX version?

Good chance I would think.