Are you buying a GTX 1070/1080?

[railven]

Likewise, I feel the same about Async Compute. It's only the start, as devs get better at it, by 2018, expect 20-30% performance gains routine. Guess what? Volta will arrive with fully capable hardware Async Compute. Just in time to a market ripe for the picking.

This is pretty much how I feel about AC. By the time it is used heavily in more games, not just sponsored games, Volta will be out.

NV is playing the market like a fiddle. Give the audience what they want, let AMD pioneer new tech, jump on it when it is profitable, make AMD look ridiculous, ride into the sunset.

They've been doing this since AMD took over ATI. I don't expect things to change.

 

[renderstate]

Both companies are pioneering new tech. Both support cool features the other one does not support yet. Async compute is a great feature but it's hardly the only new cool thing out there.

 

[JamesDax3]

I'm on a budget so I'll be getting the R9 480.

 

[boozzer]

This is pretty much how I feel about AC. By the time it is used heavily in more games, not just sponsored games, Volta will be out.

NV is playing the market like a fiddle. Give the audience what they want, let AMD pioneer new tech, jump on it when it is profitable, make AMD look ridiculous, ride into the sunset.

They've been doing this since AMD took over ATI. I don't expect things to change.

If true, this reads like amd got super, super stupid upper management in both R&D and the business side. Business 101, timing, they can't even get that right.

If true.

 

[Arachnotronic]

This is pretty much how I feel about AC. By the time it is used heavily in more games, not just sponsored games, Volta will be out.

NV is playing the market like a fiddle. Give the audience what they want, let AMD pioneer new tech, jump on it when it is profitable, make AMD look ridiculous, ride into the sunset.

They've been doing this since AMD took over ATI. I don't expect things to change.

Why waste die area on features that aren't going to make a difference for what people are going to play now and over the next couple of years? It's sort of like how Intel wastes all the die area/effort adding crap like AVX2 to client processors when what we really need is faster legacy performance.

 

[tviceman]

Well, let's see, maybe I wasn't buying hardware back then so I couldn't have? My join date is 2013 after all. Possible.

Weird, I had no idea that the date someone joins anandtech forums is the date that they start buying computer hardware on.

And I'm not in the market for smaller Vega, but yes, I would criticize it the same if it turns out like that. However, since Vega isn't out, and we know little to nothing about it, I don't know what relevance it has to the discussion.

If you don't want to talk about Vega then don't bring it up.

This is the thing that I see from NV/Intel fans all the time, in pretty much any discussion, "AMD would do the same thing in this situation", but see the thing is, 90% of the time, they aren't, and they don't. It's a complete red herring.

AMD bent everyone over the bed with Tahiti prices. They went from charging $379 with the 389mm2 Cayman on 40nm to $550 for the 365mm2 Tahiti. Both the price delta and percentage increase for Tahiti over Cayman was way more than GM204 over GK104 as well as the MSRP price difference for the upcoming GP104 non-founder cards over GM204. So AMD did do it and they will do it again if they ever get the opportunity. Anyways, have at it. Continue to be wrong about AMD and ignore the fact that GTX 1070 is going to be both bigger in die size, come with 4x the amount of memory, and have a cheaper non-founder's MSRP than the GTX 670.

 

[tential]

If true, this reads like amd got super, super stupid upper management in both R&D and the business side. Business 101, timing, they can't even get that right.

If true.

Do you have arguments against any of that?

 

[Silverforce11]

You already decided it's useless without even knowing how it works..wow Care to elaborate some more and tell us why it won't be used?

Just like how you often declare a lot of AMD stuff to be useless before fully understanding it, amirite? Discard Accelerator anyone? lol

Here's why it won't be used. It's Pascal only. Not even NV only. Pascal only.

The recent time NV presented about marketshare, they noted, only 30% of it's user base has moved onto Maxwell. 70% is still on Kepler and Fermi.

With Pascal being launched now, even a year from now, it's still the same story, those on the latest are a minority.

Thus as a VR market base, those that support Pascal SMP is a small %. Devs will not bother to add these tech to their VR games. Not unless NV sponsors it and pays them to do so. Look at the vast majority of VR games, what's the ratio that has NV's sponsorship? Close to zero.

If you think this is illogical, note it and we will revisit this in a year and two years later. I will be correct.

 

[imaheadcase]

Why waste die area on features that aren't going to make a difference for what people are going to play now and over the next couple of years? It's sort of like how Intel wastes all the die area/effort adding crap like AVX2 to client processors when what we really need is faster legacy performance.

Which is strange that nvidia is adding stuff that amd did already that most users don't care about. Like the "trueaudio" clone in pascal..i just don't see why it was even included if not for just saying "we can do this to amd".

 

[RussianSensation]

OFF-TOPIC Warning but needs to be addressed.

AMD bent everyone over the bed with Tahiti prices. They went from charging $379 with the 389mm2 Cayman on 40nm to $550 for the 365mm2 Tahiti. Both the price delta and percentage increase for Tahiti over Cayman was way more than GM204 over GK104 as well as the MSRP price difference for the upcoming GP104 non-founder cards over GM204. So AMD did do it and they will do it again if they ever get the opportunity. Anyways, have at it. Continue to be wrong about AMD and ignore the fact that GTX 1070 is going to be both bigger in die size, come with 4x the amount of memory, and have a cheaper non-founder's MSRP than the GTX 670.

You are making some critical mistakes in your comparison.

1. We should compare next gen cards against the fastest cards of that generation (whether they come from AMD/NV). Why are you comparing Tahiti against HD6970 and ignoring how Tahiti OC mopped the floor with a $450 GTX580 by 48-80%, while doubling its VRAM? Using this comparison, 1080 OC would need to beat Titan X/980Ti by 48-80% OC vs. 980Ti/Titan X OC and double their VRAM (so at least 12GB of GDDR5X at least).

Otherwise, I can just as easily make the statement that AMD almost doubled the prices from HD6950 $299 @ unlocked into a 6970 when they released the 7970. What difference does it make though since NV fans never bought the 6950 nor the 6970. It's 100% irrelevant that you are bringing up 6970's price as justification because AMD did everything they could to try and convert NV users with low prices of HD4000-6000 series and it didn't work. Had NV known this in hindsight, not a chance would they have even bothered. If AMD knew this strategy would fail 100%, they would have priced HD4870 at $499 and HD6970 at $499-549. This is WHY they raised the price of 7970 to $549 -- not because they decided to take advantage of the market but because they gave up seeing that it's hopeless to try to convert NV loyalists.

2. In hindsight, 7970 OC outperformed 680 OC, 770 OC and is now often trading blows with GTX780/OG Titan. Let me know when 1080 OC can take take 2 next gen AMD cards and then in its 3rd and 4th years since launch it can trade blows with AMD's top-of-the-line $650 2017 card.

In fact, I am going to say it right now. 7970 is the best videocard EVER made, better than 9700Pro and 8800GTX Ultra. No videocard in the world before 7970 destroyed the competition so badly over the long-term useful life of the graphics card, while offering both graphics + compute performance (FP64, mining) and managed to have a perfect amount of VRAM, provided 30-60 fps @ 1080p for 4 years straight, while also paying for itself in the process. 8800GTX could never play games at high settings at 1080p for as long as 7970 managed.

No matter how good Vega or 1080 will be, neither of those videocard will ever live up to the glory of 7970 OC, not a chance.

Do you realize just HOW good 7970 really was/is?

A September 2014 GTX970 is only 32% faster than 280X (7970Ghz is actually faster than 280X!) at 1440p.

http://tpucdn.com/reviews/ASUS/GTX_950/images/perfrel_2560_1440.png

You literally decided to mount an offense against what will go down in history as either the best videocard ever or easily one of the best videocards ever. Not a single NV card from Fermi, Kepler or Maxwell generations will make this list.

In case you forgot, 7970 is literally wiping the floor with GTX680 in modern games, the type of beating we haven't seen in a decade.

In Quantum Break, 680 is unplayable while 7970/280X is up there with a 970.

When 1080 is outperforming AMD's 14nm competitor by 30-50% in modern AAA games of 2019-2020, then we can talk because that's exactly how much faster HD7970/280X is today against 680/770 2GB - you know its competition.

Why waste die area on features that aren't going to make a difference for what people are going to play now and over the next couple of years? It's sort of like how Intel wastes all the die area/effort adding crap like AVX2 to client processors when what we really need is faster legacy performance.

Because per NV's own stats, 70% of their own customers are using pre-Maxwell GPUs. That means only 30% of their customer base keep GPus for 1.5 years or less. That means a large fraction of gamers would have benefited had Fermi and Kepler aged better long-term given their pricing premiums over that generation.

For me a 7970 made thousands of dollars, while smashing NV's competing 680/770 cards in games since November 2014, without suffering VRAM gimping along the way. I guess it sucks for AMD that they made such a great card that I didn't even need to upgrade until 2016, having saved my $ by not wasting it on a $650 780, $699 780Ti, $1000 OG Titan, $550 980, $550 290X. That's a lot of $ saved/can be used on other hobbies/more games, CPU platform upgrade, etc. Then again since I am a consumer, I should want my graphics card to last as long as possible and perform as best as possible beyond the first 1.5 years of ownership -- that's logical for people who aren't shareholders of the GPU brand they are emotionally invested in.

It is telling though how SO many people who are eyeing 1070/1080 upgrades are still on outdated Sandy Bridge/Ivy Bridge i5s and 1080p 60Hz screens. That should tell us that money is a factor since if they could easily afford it, they would have already upgraded to Haswell/Skylake i7s and you know a real monitor worth of a 1070/1080 level of performance. Instead, they overspent on NV's poorly aged and overpriced past gen GPU hardware; and today have outdated CPU platform and an outdated monitor. And you were saying that who cares if a GPU lasts long? Ya, I guess if you are in the top 5% of guys who buys GPUs every 12 months it doesn't matter but using Steam and NV's own stats, at least 70% of NV's userbase were affected.

By far the more interesting poll would have been how many of those who clicked I'd purchase a 1080 would have even considered an AMD Vega card if it were launching in June. My guess is it would be a small fraction to begin with.

 

[renderstate]

Just like how you often declare a lot of AMD stuff to be useless before fully understanding it, amirite? Discard Accelerator anyone? lol

Please quote the message where I said it's useless. All I said is that it cannot be the magical thing you think it is. Too bad for you I do understand how realtime 3D graphics works and I explained very well what it can't or can do. IIRC you simply disappeared from that discussion. I guess I know why

Here's why it won't be used. It's Pascal only. Not even NV only. Pascal only.

Some features require massive undertaking to be adopted, some are easy to use. I am willing to bet single pass stereo rendering falls in the second case since VR apps already do stereo rendering. Supporting it on Pascal won't be that challenging.

 

[Silverforce11]

Please quote the message where I said it's useless. All I said is that it cannot be the magical thing you think it is. Too bad for you I do understand how realtime 3D graphics works and I explained very well what it can't or can do. IIRC you simply disappeared from that discussion. I guess I know why

Some features require massive undertaking to be adopted, some are easy to use. I am willing to bet single pass stereo rendering falls in the second case since VR apps already do stereo rendering. Supporting it on Pascal won't be that challenging.

I couldn't continue because a mod banned me for mocking JHH.

As for your prediction, we will see, let's revisit this in a year's time and count how many VR games have SMP support.

 

[tviceman]

OFF-TOPIC Warning but needs to be addressed.

You are making some critical mistakes in your comparison.

1. We should compare next gen cards against the fastest cards of that generation (whether they come from AMD/NV). Why are you comparing Tahiti against HD6970 and ignoring how Tahiti OC mopped the floor with a $450 GTX580

I wasn't ignoring anything. The cry fest is that everyone thinks die size is the end be all the pricing. I was comparing launch prices and die sizes from same company, nothing else. I never compared AMD pricing and/or performance to Nvidia. Look through my post history and you'll consistently see that when I talk of such subjects, I compare prices of cards/chips of generations from the same company. I never said Tahiti wasn't a good chip etc. etc. Therefore my points all stand. AMD went from charging $379 on 40nm to $550 on 28nm with a slightly smaller die size. Everyone can complain all they want about Nvidia raping their kid's college funds (and I'm right there with the complaints), but AMD did it too and will happily do it again if they get the opportunity.

 

[Silverforce11]

I wasn't ignoring anything. I was comparing launch prices and die sizes from same company, nothing else. Look through my post history and you'll consistently see that when I talk of such subjects, I compare prices of cards/chips from generations of the same company. I never said Tahiti wasn't a good chip, still holds it's own today, etc. etc.

My points all stand. AMD went from charging $379 on 40nm to $550 on 28nm with a slightly smaller die size. Everyone can complain all they want about Nvidia raping their kid's college funds (and I'm right there with the complaints), but AMD did it too and will happily do it again if they get the opportunity.

AMD did jack up the prices, I was very critical of them asking $550 for Tahiti. You can find several times where I said they are overpriced in the launch threads.

These are companies and if they can get away with jacked up prices, they will. Companies love profits. We gamers love value.

Everybody wants to be a winner, but only one can. When there's no competition, companies win, it's really that simple.

 

[tviceman]

Everybody wants to be a winner, but only one can. When there's no competition, companies win, it's really that simple.

Yep. Had Tahiti been released with the better drivers it had a year after launch, no way on earth GK104 would have launched at $499 and no way we would have saw a $1,000 Titan at the time. The pricing we face today is not AMD's fault, Nvidia is free to price as they choose and potential customers are free to buy as they see fit. I simply view the circumstances surrounding Tahiti's release (not a huge die, crap drivers, massive price increase vs. previous gen AMD) as the open door that allowed Nvidia to permanently raise the price of their ~300-350 mm2 chip.

Unfortunately, as has been predicted many times over, Nvidia's larger R&D budget focused on less products is allowing them to iterate quicker and more often than AMD now. AMD's new node lead went from 6 months, to 10 weeks, to likely falling behind Nvidia. Nvidia will have new 4 chips on the market (one possibly being professional only for the time being) before summer's over while AMD will only have 2. This is the GPU world we live in now. AMD needs a home run with Polaris and Vega to keep from falling further behind.

 

[IllogicalGlory]

AMD bent everyone over the bed with Tahiti prices. They went from charging $379 with the 389mm2 Cayman on 40nm to $550 for the 365mm2 Tahiti. Both the price delta and percentage increase for Tahiti over Cayman was way more than GM204 over GK104 as well as the MSRP price difference for the upcoming GP104 non-founder cards over GM204. So AMD did do it and they will do it again if they ever get the opportunity. Anyways, have at it. Continue to be wrong about AMD and ignore the fact that GTX 1070 is going to be both bigger in die size, come with 4x the amount of memory, and have a cheaper non-founder's MSRP than the GTX 670.

Unsurprisingly, you're completely silent about Hawaii, so I'll take that as an admission that you don't have anything to say about it. No, you've got to go back two generations to back up your point.

Here's the other thing. AMD sold Tahiti for a crap price 4 years ago. NV is selling GP104 for crap prices right now. It's what's happening right now, and frankly, any discussion about Tahiti is completely irrelevant as literally no one in the entire world is being affected by Tahiti launch prices right now. It's another red herring just like your talk about Vega.

And wow, two new architectures, a die shrink and we got more hardware for the same price. That's new. AMD is selling a 438mm^2 die with 8GB for $250 right now. That's amazing value, right? 8GB of RAM. Are you saying that it would have been acceptable to outfit the 1070 with 2GB of RAM? Seriously, what's that about?

And with the 1070, well you get what you pay for. Seriously cut down and 256-bit GDDR5, and more expensive than its larger predecessor. Plus, pricing is not known yet. The MSRP may be $380, but the 970 was around $350 at launch, despite its MSRP and who knows how the founder's edition will affect pricing.

You can continue to support a company that charges $700 for a midrange chip if you like, but I'm going to vote with my wallet and not reward them for that.

NV can jack up prices all they want; that's their prerogative as a company. My prerogative as a consumer is to not buy it. It's not even about NV vs AMD at this point. I'm just not going to let NV sell me an *04 chip for this price, and I'd encourage others to do the same, but go ahead and buy it if you want. I'm certainly not crying either. I wasn't holding out much hope anyway, and I've decided what I'm going to do about it.

 

[RussianSensation]

I wasn't ignoring anything. The cry fest is that everyone thinks die size is the end be all the pricing.

No, this isn't right. If GP100/102 Pascal was 300mm2 and GP104 was 150mm2 and they outperformed 600mm2 Vega 10, we would still consider GP104 mid-range. The die size and pricing are all in relation to what the entire next gen stack looks like. This along HBM2 vs. GDDR5X and other topics were covered. The reason 1080 is mid-range is not only because of die size.

I was comparing launch prices and die sizes from same company, nothing else.

While ignoring that 7970 has demolished 680/770 in games starting late 2014 until now. Once 1080 can do the same against AMD's $600-700 card of this generation, then you'll have a point. I already said 7970Ghz outperforms 680/770 by 30-70% in today's AAA games. If 1080 can do this to Vega in 2019-2020, I will make sure to tell you how amazing 1080 ended up.

Fact is NV raised the price of $499 680 to $599-699 1080. Leave AMD out of it completely.

Did NV raise or not raise prices significantly since 680? Yes or no?

What are you trying to say now that if NV raised 1080 to $599-699, then if AMD's Vega is even faster, they should raise prices from Fury X's $649 to $749-799 because that's kinda what you are implying here -- as long as GPUs get faster, higher pricing is justified.

AMD went from charging $379 on 40nm to $550 on 28nm with a slightly smaller die size. Everyone can complain all they want about Nvidia raping their kid's college funds (and I'm right there with the complaints), but AMD did it too and will happily do it again if they get the opportunity.

Ya and your comparison is flawed for 2 reasons.

1. Most people on this forum who buy AMD/are brand agnostic, skipped 6970 and bought 6950. If you are going to make a point, say that AMD raised prices from $299 6950 @ 6970 to $549 7970. That actually makes your case stronger and I won't disagree. I was pissed when AMD raised the price of 7970 to $549 and voted with my wallet by not buying it until it dropped at least $100.

2. Because HD6950/6970 were so close in performance to a 580 (today just a 9% difference separates them), might as well make the claim that 6950/6970 was underpriced/undervalued.

But what's that have anything to do with 1080's pricing? Are you saying because AMD raised prices to $549 with 7970, it allowed NV to raise prices from 560Ti's $249 to $499 with the 680? Ok, so why is it AMD priced 4870 at $299 when GTX280 was $649? You are just making excuses for NV that because AMD decided to raise prices on us that NV couldn't have introduced 680 at a much more reasonable price. What about thereafter, such as $280-300 280X and $650 R9 295X2 getting slaughtered by $550 980? You think that has no impact on NV decided to keep raising prices with next gen? Don't be naive. When you purchased your 980 for $550 USD, you literally voted with your wallet by telling NV you'll pay $250-300 extra for 25% more performance. They heard you loud and clear as well as the voices of millions of other 980 owners. Every person buying GTX1080 FE for $699 is sending a message to NV that they will pay $700 USD for a mid-range reference card. Why in the world would NV NOT release Volta GV104 for $699 or even $749 when their own customers just voted during Pascal generation: "Ya, it's perfectly fine with us, go ahead, keep raising prices."

Later, don't come back and tell everyone who AMD was noncompetitive so you had no choice but to buy a $600-700 "mid-range Pascal" since Vega wasn't around. You know if AMD dropped $650 Vega 10 GDDR5X in June and GP104 was 6 months behind, many people who are criticizing 1080 now would slam Vega just the same.

Besides, even if we agree that 680's $499 price was justified, how is $599-699 for 1080 justified? The 7970/680 generation is long behind us.

Also, way to ignore the $50 price increase on the 980 as well, and a price increase on the $329 970 to $379-449 1070.

I guess NV is forced to raise prices because they are doing so poorly, you know projecting 57% gross margins for 1st fiscal quarter 2017...poor company, barely making $ because TSMC is charging them so much for those 16nm FF wafers. :sneaky:

Your GTX570 cost you $350 and you are going to be paying $600-700 US for a 1080, which is a GTX560Ti family. Leave AMD out of it since they have no $ at all. No one is putting a gun to your head or anyone's to pay $600-700 for a mid-range GP104 but you are defending it using "Well AMD raised prices too so NV is forgiven" tactic. Really?!

What makes you think if AMD prices Vega 10 at $699 and if it barely matches a 1080 that people won't slam that card as well?

The big picture is GPU prices have skyrocketed and both AMD/NV are responsible. No need to point fingers at one of them. However, many of us slammed 7970's launch price, or did you forget that already?

I simply view the circumstances surrounding Tahiti's release (not a huge die, crap drivers, massive price increase vs. previous gen AMD) as the open door that allowed Nvidia to permanently raise the price of their ~300-350 mm2 chip.

You still aren't seeing it. No magic allowed NV to permanently raise the price. It was consumers who permanently raised the price by buying the product. Do you realize that when AMD raised the price on 7970 how many people waited to buy it until it dropped in price, or alternatively they waited for NV's product offering to see what's up? When AMD released 290X, many couldn't care less and bought a 290 or the cheaper 780 instead. When AMD introduced the 30% faster Fury X against 290X, do you know how many "AMD users" couldn't care less? They didn't say "Oh man, NV raised prices of 780Ti to $699 and 980 to $550, and 980Ti to $649. That means by that account even if Fury X's 30% increase is not good enough over the 290X, I'll still bend over and pay the $649, why not, it's not AMD's fault that TSMC couldn't get 20/22nm node right." Instead, many of these gamers just looked at all of these cards as Card A, Card B, Card C, Card D (who cares the manufacturer), and said those increases are just not good enough.

Are you getting it yet? When AMD launches a new card 20-25% faster for $550-700, there is no mad rush by AMD users/objective PC gamers towards it. If AMD released 3 consecutive cards, with each 20-25% faster for $550-700, a lot of people wouldn't buy them consecutively because it's just not good enough of an incremental upgrade. On the NV side, 20-25%, $600-700? Please, where do I sign....then when GP100/102 drops, upgrade again, then mid-range GV104 for $700, my body is ready. For that reason, AMD will never win over NV. If Vega is only 20% faster than Fury X, a lot of 290X/390X users will probably skip that too since most games now are just console ports and among that group, many are optimized turds.

On the NV side, you will have people who bought a $160-200 960 2-4GB and then will spend another $200 on a 1050/1060 card that barely matches/beats a $200 after-market 290 ==> You know spend $350-400 to have $200 R9 290 performance almost 2 years after. Who does that?!

 

[tviceman]

You're more or less agreeing with me but you're getting mad at me about it. I'm confused.

Also, I'm not buying GP104, did you forget?

 

[badb0y]

I'm still wondering how third party custom cards are going to cost less than reference cards. I mean, has this ever happened in the history of GPUs where a custom AIB card cost less than a reference card? Am I a potato for asking this question? Because it seems like none of the journalists bothered to ask nVidia this question in the press conference.

 

[boozzer]

Do you have arguments against any of that?

arguments against speculation/observation? why? there is no right or wrong. It's like having a serious what if discussion.

what AMD needs, and what it lacks is marketing. It has zero problems with hardware or innovation, but what it needs more than anything is marketing.

 

[exar333]

I'm still wondering how third party custom cards are going to cost less than reference cards. I mean, has this ever happened in the history of GPUs where a custom AIB card cost less than a reference card? Am I a potato for asking this question? Because it seems like none of the journalists bothered to ask nVidia this question in the press conference.

Nvidia reference models have commanded a price premium for some time. Nothing new. This is better for AIB and will spur both affordable and higher end options.

 

[IllogicalGlory]

arguments against speculation/observation? why? there is no right or wrong. It's like having a serious what if discussion.

what AMD needs, and what it lacks is marketing. It has zero problems with hardware or innovation, but what it needs more than anything is marketing.

I may have to disagree with that. As much as I'd like to defend and recommend AMD hardware, the fact is that Fiji sucked and isn't worth buying, they haven't brought us much new since then, overclocking margins are slim and the poor DX11/launch performance is a real thing.

Of course, AMD had some really great products in the past. Hawaii is the go to example, but they managed to completely ruin it with that horrible reference 7970 cooler. When people thought of Hawaii, especially at launch, they thought "hot, loud, power-guzzling, throttling". They didn't think "Titan/780 performance and 4GB of RAM for $400." If AMD had shipped Hawaii either with a better reference cooler, or simply said AIB designs only, day 1, they would have sold a lot more (although mining had something to do with that too). Even today, with so many great aftermarket cards, the hot and loud reputation persists. Hawaii will forever be hot and loud, though it has seen some success in the 390.

It's really a combination of both.

 

[badb0y]

Nvidia reference models have commanded a price premium for some time. Nothing new. This is better for AIB and will spur both affordable and higher end options.

What? I am pretty sure I bought my 980 Ti for MSRP. The custom cards always cost more from my experience.

 

[x3sphere]

I'm curious what the premium AIB cards will come in at, like Gigabyte G1 and MSI Gaming. I don't think they will necessarily be cheaper than the Founder's Edition. What if AIBs don't want to be seen as the 'cheaper' option?

The existence of this FE card is like saying, hey it's ok to charge +$100 for your premium design. It gives more validation to higher prices if anything. I hope I'm wrong but that is the way I see it.

 

[tajoh111]

No, this isn't right. If GP100/102 Pascal was 300mm2 and GP104 was 150mm2 and they outperformed 600mm2 Vega 10, we would still consider GP104 mid-range. The die size and pricing are all in relation to what the entire next gen stack looks like. This along HBM2 vs. GDDR5X and other topics were covered. The reason 1080 is mid-range is not only because of die size.



While ignoring that 7970 has demolished 680/770 in games starting late 2014 until now. Once 1080 can do the same against AMD's $600-700 card of this generation, then you'll have a point. I already said 7970Ghz outperforms 680/770 by 30-70% in today's AAA games. If 1080 can do this to Vega in 2019-2020, I will make sure to tell you how amazing 1080 ended up.

Fact is NV raised the price of $499 680 to $599-699 1080. Leave AMD out of it completely.

Did NV raise or not raise prices significantly since 680? Yes or no?

What are you trying to say now that if NV raised 1080 to $599-699, then if AMD's Vega is even faster, they should raise prices from Fury X's $649 to $749-799 because that's kinda what you are implying here -- as long as GPUs get faster, higher pricing is justified.



Ya and your comparison is flawed for 2 reasons.

1. Most people on this forum who buy AMD/are brand agnostic, skipped 6970 and bought 6950. If you are going to make a point, say that AMD raised prices from $299 6950 @ 6970 to $549 7970. That actually makes your case stronger and I won't disagree. I was pissed when AMD raised the price of 7970 to $549 and voted with my wallet by not buying it until it dropped at least $100.

2. Because HD6950/6970 were so close in performance to a 580 (today just a 9% difference separates them), might as well make the claim that 6950/6970 was underpriced/undervalued.

But what's that have anything to do with 1080's pricing? Are you saying because AMD raised prices to $549 with 7970, it allowed NV to raise prices from 560Ti's $249 to $499 with the 680? Ok, so why is it AMD priced 4870 at $299 when GTX280 was $649? You are just making excuses for NV that because AMD decided to raise prices on us that NV couldn't have introduced 680 at a much more reasonable price. What about what ensured thereafter, such as $280-300 280X and $650 R9 295X2 getting slaughtered by $550 980? You think that has no impact on NV decided to keep raising prices with next gen? Dont' be naive. When you purchased your 980 for $550 USD, you literally voted with your wallet by telling NV you'll pay $250-300 extra for 25% more performance. They heard you loud and clear as well as the voices of millions of other 980 owners. Every person buying GTX1070 FE for $699 is sending a message to NV that they will pay $700 USD for a mid-range reference card. Why in the world would NV NOT release Volta GV104 for $699 or even $749 when their own customers just voted during Pascal generation: "Ya, it's perfectly fine with us, go ahead, keep raising prices."

Later, don't come back and tell everyone who AMD was noncompetitive so you had no choice but to buy a $600-700 "mid-range Pascal" since Vega wasn't around. You know if AMD dropped $650 Vega 10 GDDR5X in June and GP104 was 6 months behind, many people who are criticizing 1080 now would slam Vega just the same.

Besides, even if we agree that 680's $499 price was justified, how is $599-699 for 1080 justified? The 7970/680 generation is long behind us.

Also, way to ignore the $50 price increase on the 980 as well, and a price increase on the $329 970 to $379-449 1070.

I guess NV is forced to raise prices because they are doing so poorly, you know projecting 57% gross margins for 1st fiscal quarter 2017...poor company, barely making $ because TSMC is charging them so much for those 16nm FF wafers. :sneaky:

Your GTX570 cost you $350 and you are going to be paying $600-700 US for a 1080, which is a GTX560Ti family. Leave AMD out of it since they have no $ at all. No one is putting a gun to your head or anyone's to pay $600-700 for a mid-range GP104 but you are defending it using "Well AMD raised prices too so NV is forgiven" tactic. Really?!

What makes you think if AMD prices Vega 10 at $699 and if it barely matches a 1080 that people won't slam that card as well?

The big picture is GPU prices have skyrocketed and both AMD/NV are responsible. No need to point fingers at one of them. However, many of us slammed 7970's launch price, or did you forget that already?



You still aren't seeing it. No magic allowed NV to permanently raise the price. It was consumers who permanently raised the price by buying the product. Do you realize that when AMD raised the price on 7970 how many people waited to buy it until it dropped in price, or alternatively they waited for NV's product offering to see what's up? When AMD released 290X, many couldn't care less and bought a 290 or the cheaper 780 instead. When AMD introduced the 30% faster Fury X against 290X, do you know how many "AMD users" couldn't care less? They didn't say "Oh man, NV raised prices of 780Ti to $699 and 980 to $550, and 980Ti to $649. That means by that account even if Fury X's 30% increase is not good enough over the 290X, I'll still bend over and pay the $649, why not, it's not AMD's fault that TSMC couldn't get 20/22nm node right." Instead, many of these gamers just looked at all of these cards as Card A, Card B, Card C, Card D (who cares the manufacturer), and said those increases are just not good enough.

Are you getting it yet? When AMD launches a new card 20-25% faster for $550-700, there is no mad rush by AMD users/objective PC gamers towards it. If AMD released 3 consecutive cards, with each 20-25% faster for $550-700, a lot of people wouldn't buy them consecutively because it's just not good enough of an incremental upgrade. On the NV side, 20-25%, $600-700? Please, where do I sign....then when GP100/102 drops, upgrade again, then mid-range GV104 for $700, my body is ready. For that reason, AMD will never win over NV. If Vega is only 20% faster than Fury X, a lot of 290X/390X users will probably skip that too since most games now are just console ports and among that group, many are optimized turds.

On the NV side, you will have people who bought a $160-200 960 2-4GB and then will spend another $200 on a 1050/1060 card that barely matches/beats a $200 after-market 290 ==> You know spend $350-400 to have $200 R9 290 performance almost 2 years after. Who does that?!

You need to look at this from an economic and business perspective.

You base launch pricing on initial performance, not future performance and upon launch the 7970 only beat the gtx 580 by 15%. Basing it on future performance requires hindsight so pricing a card based on future performance is just plain stupid.

Additionally, AMD is the value brand and with this comes certain expectation with launches from consumers, or they will wait and not buy your product. What is universally expected for the value brand is they offer superior price to performance. When the brand recognized for their value discards this characteristic and launches their product at a high price. What do you think is likely to happen when the brand with more prestige launches a product with better performance and generally all around better characteristics, particularly in a duopoly?

And this is why much of the price increases are atleast half AMD fault. Launching their product at 550 when it was only performing 15% better than the competition while being the value brand was an incredibly dumb move on their part. There was little possibility, that Nvidia was not going to do better than that at launch. So when this came true, it was not unexpected for Nvidia to start charging more for their midrange. Thus following AMD leads and increasing their price for their midrange is what is expected and is simply business. This is because Nvidia is the more prestigious brand. The brand with the higher good will with the consumer is allowed and is supposed to charge more for their product.

If the prestige brand didn't, the second brand marketshare and margins would get smaller and smaller until they went bankrupt. We got a preview of this effect at 28nm. This is what kind of happened when the gtx 670 launched with similar performance to a 7970 and a 150 dollar lower price. And what happened when the gtx 970 launched vs the 290x with a 220 dollar lower price point. AMD got the crap smacked out of them as far as profit and marketshare go and if this pattern continues, we know what will happen.

In a duopoly, it's up to the value brand to keep the prestige brand price in check. If they try to raise pricing, this give every right for the prestige brand to increase pricing because the floor pricing for a certain level of pricing is determined by what the value brand is willing to sell it for. This is because at equal pricing/performance, the prestige brand will vastly outsell the value brand. This is the right of the prestige brand. Hence, the concession for the prestige brand is they raise their prices to have higher margins, which allows the value brand to continue to exist or make money at lower margins. Thus when AMD raised pricing, Nvidia did as well and they put pricing on one of their products lower than AMD to get the second effect as well.

Lets take AMD brand out of the equation(since this might be distorting your view of this situation) and put up a similar example. Lets pretend Vizio and Samsung were the only two tv makers. Lets say Vizio next year, decided to raise the pricing of their products to a price above samsung because it had slightly better picture quality than the samsung. Samsung releases a tv with even better picture quality, so what do you expect pricing to be like for this tv considering there is no alternative brands out there?

Prestige brands can get away with certain behavior the value brand cannot. Thus AMD raising the price with their entire initial 28nm line is what doomed pricing for the consumer. It doesn't mean AMD is stuck as the value brand.

The only way AMD can raise pricing, is if they launch an absolutely superior product that the competition has no chance of beating within a reasonable time frame. Do a enough of these in a row and a few marketing trick and they can become the premium brand.