[wccftech]Nvidia Volta Allegedly Launching In 2017 On 12nm FinFET Technology

[Phynaz]

The point I was making when I brought up ancient history is that AMD has been operating in a different price bracket for years, at least when it comes to their single-GPU flagship vs NVIDIA's single-GPU flagship. Competition is much more fierce in the sub-250$ segment. This is not because AMD is forced to like it does with its current CPU lineup. With AMD cards you may not have the fastest product, but the alternative would be of much better value that it would make the fastest product from NVIDIA irrelevant.

If you had got the GTX 980 Ti, or the GTX 1070, knowing fully well that it would be adequate for your purposes for at least a couple of years, then that's a good thing. In fact, I'm a staunch believer that you should get the best card that suits you at the moment. However, with Maxwell and now with Pascal, there has been a tendency to equate best with costliest, and somehow this has stuck with the populace, and people now covet the priciest card which only indulges NVIDIA to jack up their prices even further.

Short version:
AMD has failed at marketing.

Response:
Yes, we know.

 

[Ajay]

Sigh. Do people post stuff in VC&G instead of the AMD or Nvidia sub forums JUST so that fight will break out...
I mean, this is clearly a Nvidia specific post.

 

[tential]

Sigh. Do people post stuff in VC&G instead of the AMD or Nvidia sub forums JUST so that fight will break out...
I mean, this is clearly a Nvidia specific post.

Then make one in the form. I enjoy talking about the cards in a pure vacuum Aka the subforums, but I also like here because we need to put it in context and it's fun to get opinions from everyone. I rarely go into the Nvidia forum, I'd rather talk about the Nvidia products in relationship to amd.

Actually I get jaded and think Nvidia is ultra terrible if I only look at them on a vacuum. Before fury x came out? I screamed how bad the 980ti was. After fury x? 980ti best card ever!!!

So ya, some times it's useful to have reference and competition points. Volta to me is an auto success due to Nvidia again having a midrange chip with 0 competition from amd long before amd will have a viable competitor in Navi.

 

[IllogicalGlory]

Huh, nothing you posted disputes the superiority of Nvidia at the high end. The 980ti is such a good value card it broke the perception of high end cards having poor performance to cost. Even the 1070 has done well on tech power up graphs. Even I had to look back and see the 1080 doing amazing... That's mind boggling.

It never was, it only looked good compared to the Fury X. Real value options like 290/970 were in an entirely different league on that front.

Techpowerup's data shows that at 1440p (at launch), the 980 Ti's value was the second worst on the market, beating only the Titan X at the time. A year later, with a custom version, it looks better, but certainly well into the premium category. Nothing perception-breaking at all. You have the rose-tinted glasses on. There's no doubt that the 980 Ti is a great high-end card, but it was far from a good value.

Meanwhile, you inexplicably snubbed the 290 earlier, when that's a card that actually might have broken such perceptions. Top-end performance with its value landing squarely in the middle.

And while we're recognizing the 1070, it only brought $650 performance down to $350-$380 (right now), while you can buy $300 performance (970/390) for $141 with a 4GB RX 480, when those $300 options were already on the value end of the spectrum. Compare that to, say, the 960, which brought less than $300 R9 280X performance at $170-200, there's no doubt that the 470/480 are offering something special, far and above anything NV has, at least as far as value goes. The best 1060 deal I can find is a 3GB for $169. Weak compared to a 4GB 480 at $141.

I don't know what you mean about the 1080, but mind-boggling doesn't seem like an apt way to describe it, as far as value. Second worst value on the market seems pretty standard for a premium card. If it is, the 290 is way more mind-boggling. Hell, even the 290X beat out seven other cards for value (in uber mode) at launch.

 

[RussianSensation]

I'm not talking about ancient history. I'm talking about now. People love to bring up some super old card when complaining about amd high end. I don't care. I'd say the vast majority of users shopping for a gpu now don't care, or remember what the HD4000 series. Amd hasn't had a viable competitive high end offering at launch in years that hurts. It hasn't even had something to truly compete against the Nvidia midrange "but I'm high end card because I'm the fastest on the market even though I'm clearly a midrange gpu" gpus.

Amd just hasn't had the grunt to compete at the high end.

As for the dollar amounts you're talking about... I regret not buying a 980ti at launch in hindsight right now. Why would I care about the 1070?

Also,what do you say to the tons of people who will pick up the amd competitor to the gtx 1070? Will you berate them for not getting a 1070 that had been out for months or ask them why they ignored a perfectly good gpu to wait a year to get the amd equivalent?

Except your strategy of buying flagships and not waiting isn't as clear cut as you made it sound. Instead of wasting my money on $1200 980 SLI, I just gamed longer on my 7970Ghz CF and got R9 295X2 for $600. 980 SLI was faster by what 20-25%? It's irrelevant since for 1440p gaming R9 295X2 was crushing it. I had $600 in my pocket towards the next upgrade [but 295X2 and later my 390 were mining $$$ mining too: double win].

You then hype up the $650 980Ti. I easily skipped that card since my R9 295X2 was still fast. I then got 1070 SLI for $652 (ya, I am patient and wait for good deals), but I could have spent $1300 on 980Ti SLI. That's another $600 saved. Would buying 980Ti SLI a year earlier made me more happy? No, not really since most AAA Pc games are either well optimized (thanks EA) or are console ports (turn down 2-3 settings, performance goes up 30-70% with 95% the same IQ)

Your posts make it sound like unless a gamer buys a flagship $700 card at launch, then it's pointless to hunt for deals or buy high-end cards later on the generation. For example, you flat out ignore how $280-325 R9 290X / $600 R9 295X2 were competing for at least 6 months during the $500-550 GTX980 generation. You ignore how after November 2014, $700 780Ti was already struggling to outperform a $400 AIB R9 290 in the most modern AAA games.

What makes you think the minute NV/AMD release a new card, high-end gamers just upgrade money be damned? I mean that's what you are saying by suggesting you want flagship card for just 1-2 years. That means you don't care at all about the resale value of your flagship card. That's fine, but then why aren't you willing to pay a GSync premium in the first place, get NV and never have to complain about AMD not being fast enough at launch? You seem to want cutting edge gaming experience but don't want to pay NV GSync and Ti/Titan premiums.

Now, you claim you want to spend $1500+ on Vega CF or 1080Ti SLI. Why? There are 0 next generation PC games and you don't have a 144Hz 4K monitor. It's your money and you are free to spend it. I am just saying, there is a difference between buying something you want and need as you said yourself. There are many gamers like me who upgrade when we need more performance. That's why 1070 and Volta matter far more to me than 1080Ti. Flagship $700 cards used to make sense to me when PC gaming was cutting edge; when it truly meant a $700 card provided a significantly better gaming experience than a $350 card. This is just not the case anymore. Take 95% of PC gamers, put them on 2 computers side-by-side and with minor IQ tweaks, it's going to be hard to justify why a $1200 TXP and $1700 6950X is worth that over a $350 1070 and $330 6700K.

Right now 1070 SLI destroys the vast majority of games at 1440p. Why should I want to "waste" $1500+ on 1080Ti SLI for worthless theoretical gains when it won't materially change my gaming experience? I'd rather wait for x70 Volta cards, save another $600 and buy 2 of those in SLI (or even just 1 since next gen PC games are MIA).

Since 980 generation, skipping 980 SLI, 980Ti SLI and 1080Ti SLI would have saved me $1800 over last 3 generations. This applies to single flagship cards too. Why pay $650-700 for a "future-proof" flagship when there are 0 next gen PC games that make the purchase satisfactory? You do agree that $600-700 flagship cards then have to come with $90-100 AAA game purchases like FO4/The Witcher 3 bought around launch? Let's not forget the software costs to truly feel like the flagship card is beneficial for the latest cutting edge games.

In the olden days, I'd fire up Crysis 1 on 7800GTX and get 20-25 FPS at 1280x720 with 2xMSAA on medium. I put in $700 8800GTX, I am at 40 FPS. HUGE difference. Today, 95% of games can be maxed out on a $350 GTX1070 at 1080p 60Hz and it still flies at 1440p with intelligently chosen adjustment of IQ settings in AAA console-ported games.

Flagship $700-1200 GPUs never made less sense in the history of PC gaming than now. The % of gamers who own 1440p 100-165Hz, 3440x1440, 4K or multiple monitors is tiny [look up Steam]. The amount of gamers on the market buying $650-1200 flagship cards is irrelevant, but yet they are the most vocal on forums....visit HardOCP forum and you'd think unless you have a 980Ti or better, you aren't a PC enthusiast.

So where does Volta come into play? I easily have the $ for 1080Ti SLI, but the question is what does that get me over 1070 SLI? I am going to need to get a $1200+ 3440x1440 100Hz or 1440p 144Hz or 4K monitor to benefit. That would be a killer rig, I don't disagree! My point is for 1440p 144Hz <-> 5K gaming, $1500 card setups make a lot of sense! But don't assume there are a lot of gamers who still want high-end performance 12 months later for 1/2 price (i.e., since they are working through a backlog of Steam/GOG/Origin, etc. games).

I would feel a lot better paying 1/2 for Volta 2070 SLI in 2018 when/if 1070 SLI chokes than paying 2X more for extra performance I'll never use in 2017, while 1080Ti SLI twirls thumbs waiting for next gen PC games for most of 2017.

You are also seemingly contradicting yourself. If you are gaming on R9 290(X) all this time, why would you need 1080Ti SLi/Vega SLi/Volta SLI? If you needed that much performance, you'd at least already have 290X CF by now to maintain FreeSync. If a single 290X is enough, maybe a single 1080Ti/Volta GV104/Vega would be enough.

The point tamz_msc is making is that the extra performance the top cards offer doesn't add much value to most PC gamers. Those gamers that see the value don't need to visit technical forums. They can just buy ToTL and move on with their life. I actually 100% agree and it goes for both AMD and NV. GPUs depreciate so fast and are replaced by next gen cards so quickly, that by the time the 980Ti/1080Ti level card is fully utilized, that level of performance costs 1/2 or less. PC software is lagging even more now since we are in the 2nd half of the console generation. When 980Ti's performance benefits in 2016-2017, we have $350-375 1070. When I see Horizon Zero Dawn, Forza Horizon 3, Uncharted 4/Lost Legacy, the Last of Us 2, Days Gone, Journey, etc. it makes me honestly sad. On own hand it's great that $130 RX470 and a $110 i3 deliver better experience than a PS4, but on the other hand a $2400 TXP and a $1700 6950X isn't worlds better. It's better but it's not THAT much nether to me. And then once we move to an i7 6700/7700K and GTX1070, the diminishing returns with $1200 cards and $1700 CPUs are just insane to me (imo).

One more point you didn't talk about. A lot of PC gamers have a giant backlog of games. Additionally, with $60 games and $30-40 DLC, many launch games cost $90-100 USD. Now let's take $600 savings every generation I mentioned by skipping Ti series in SLI, and then add $70-80 savings on NOT buying those AAA games at launch and buying them 12-18 months later for $20-30, multiply that by 10 AAA games a year, I get $600-800 savings on PC software.

Now, combine $600-800 annual savings on software + $600 annual savings on GPUs by skipping Ti cards entirely and buying x70 Kepler, Maxwell, Pascal and Volta cards, I am at $1200-1400 savings a year. Over 10 years, that's $12,000+. [with AMD it's MUCH more since every videocard since HD4890 days was free due to mining].

For that reason, x70 Volta card for $400-450 is miles more exciting than a $700-800 1080Ti imo. The irony too is that nowadays the best looking PC games tend to be the most optimized too (BF1/SW:BF/Doom, etc.).

It sounds like a lot of people on this forum are either making 6 figures, don't have kids, or don't have many hobbies outside of gaming. I sure like taking $1000+ savings on PC software+hardware a year and using it on other hobbies or building my stock portfolio so I can retire earlier. If PC graphics continue to stagnate, during Volta generation, if $250 2060 is 2X faster than Pascal 1060, I won't recommend anything above that card for 95% of gamers on forums. Buying $700 flagship cards for bragging rights and "fun" is perfectly fine but on a technical forum, I believe we shouldn't recommend people waste their $. They come here for good advice. What good would an audio forum be if everyone just recommend Focal Utopia for $4000? What good what a car forum be if everyone recommend a Panamera Turbo S for a family sedan? It takes little skill to recommend TOTL.

Otherwise what is the point in giving newcomers to PC gaming advice? I might as well tell them to buy a 6950X, Corsair H115i GTX Extreme, Titan XP SLI, Acer Predator 144Hz and come back in 3-5 years.

A PC enthusiast isn't someone who buys top-of-the-line PC parts. Anyone who makes $4-5K a month can do that with 2-3 months savings. That requires 0 brains, 0 knowledge. A PC enthusiast is someone who recommends a gamer the best system within specific limitations - case space constraints, budget, etc. It takes knowledge and research to know i5-6400 can hit 4.4-4.6Ghz on a specific Z170 board with a $15 Gamaxx 400 CPU cooler, when Z270 boards locked out BCLK overclocking and that i5-6600 don't overclock any better. It takes 0 skills and knowledge to buy a 7700K, Noctua NH-D15/NZXT 62/Cryorig A80, Asus Maximus board, SeaSonic 650W Titanium, and a 1080Ti. Wow, big deal! All that tells me is the person has saved up $, not acquired PC building skills. Buying TOTL is easy.

It takes A LOT more skills to recognize that 980 SLI was a waste of $ compared to R9 295X2, then it is to flat out buy 980 SLI because you can afford it. It takes knowledge to know a $40 Thermalright True Spirit 140 offers 95% of the value of the $90 Noctua NH-D15. It takes knowledge to know that Phanteks PH-1/2, Kryonaut Thermal Grizzly, or Thermalright ChillFactor III smash AS5. It takes research to know that Gigabyte Z170X-Gaming 5 has Burr-Brown DAC and a 600Ohms AMP that saves a gamer $100 on a separate DAC/AMP when hooking up Sennheiser HD650/6XX or AKG K7XX.

Someone who flashed HD6950s to 6970s, water-cooled them with a custom loop, and figured out a way to make $ mining, paying for PC hardware, while hitting 5Ghz on a self-delidded 2600K is way more of a PC enthusiast to me than someone who bought a 6950X and Titan XP SLI and goes online acting like they are a "PC enthusiast" (I am man enough to admit I am not skilled enough to delidd or build a custom WC loop - that's why I continue to learn and see how much more knowledgeable than me other PC users are. I look up to the Stilt or IDK, etc.). Same reason you probably know 100X more than me about GameCube/Wii/WiiU emulators. I might have 6950X and TXP but I don't know emulators, and I'd look for guidance. It's "easy" to just spend $10,000 on GameCube/Wii/WiiU hardware and games. See the correlation?

What fun/skill is there buying top-of-the-line parts every generation?! That's dictated by one's Bank account, not skill. Therefore, anticipating Volta and trying to save $ by not automatically spending it on 1080Ti SLI say by looking at 2017 PC titles and gauging that it might be better to wait for Volta and 100-144Hz 4K panels actually requires some research on next gen PC monitors/state of next gen PC games.

Of course it also takes skill, research and knowledge to anticipate buying 780Ti/980/1080/1080Ti and reselling them at an optimal time by reading rumours of impeding successors. I widely support this strategy too after NV went bi-furcating a GPU generation route. Buying a $650 980Ti and holding it for 2 years while it drops to $250-275 by the time 1080Ti comes out takes 0 skills. Anyone with a half-decent job in the US can do that....

But can you actually provide a reasonable argument why buying a 1080Ti is worth it in 2017 instead of waiting for 2070 Volta in 2018? If you say it's about want not need, then is it really that expensive to buy $700 cards every 12 months anyway? Apple just sold 78M iPhones in 1 quarter with ASP or $694. So clearly not many people want a $700 flagship card enough OR the majority of PC gamers just don't think a $700 card offers enough of a superior gaming experience over a $250-400 card.

PC gaming is absolutely huge with ~ 150M Steam users by now but only 1-2% of those them have a 1080. Why is that?
It's NOT really about the $600 price. Most gamers just don't think the card that costs $600 is worth it over a $250-350 card based on the marginal utility of the overall experience. It's only the top 1-3% of hardcore PC games who think it's worth it. The rest buy a card that's 90% good enough and upgrade again when that card ages. We just don't see tens of millions of gamers with $80-400 2012-2016 cards on forums since they are too busy gaming, not arguing how a $650 Fury X is trash since it lost to 980Ti by 9-10%.

I can already predict you'll call a $450 Volta GV104 2070 worthless since we could have had that performance in 2017 with 1080Ti (while ignoring its $650+ price) or how Vega will be crap if it only brings +10% performance over $700 1080 for $499 (while ignoring that during the first 2-3 months AIB 1080 cards cost $700-750, and yet you advocate buying flagship cards as soon as possible). If price isn't a factor, you get GSync, 980Ti SLI/Titan XP SLI/1080Ti SLI and just enjoy gaming. Then when Volta 2080 comes out, get 2 of those and enjoy again. Also, you can just buy NV stock and this way you'll get to have TOTL Pascal/Volta/GSync, while selling their shares for a profit. Don't be naive and think the hardcore pro-NV users on AT, HardOCP aren't doing this.***

***Just my opinion.***

The reason Volta is exciting is not only because a $450 2070 ~ 1080Ti, but because a $250 2060/Ti should be ridiculously fast! We should want more PC gamers to move away from 1080p 60Hz monitors but that won't happen if it takes $450-1200 Volta cards to do it. Most people don't want to spend that much every 12-24 months.

 

[tamz_msc]

And to drive home the excellent point that RussianSensation made in his well-argued post, let us consider this situation:

You are a gamer in mid-2015 who has a GTX 970 and a 1080p 60 Hz monitor who decides to go for the 1440p 60 Hz experience. Do you:

a) Sell that monitor and GTX 970 for 300$(say 200$ GPU and 100$ monitor) and get the 980Ti(700$) and the monitor(300$) for a total expense of 700$, or

b) Keep in mind what happened to the GTX 780Ti and sell that monitor for now; pay 200$(1440p monitor) in the short-term for gaming at reduced settings, then sell the GTX 970 for 150$ and pay the extra 250$ to buy a GTX 1070, for what is effectively the same performance, that saves you 250$, AND gives you an extra 2GB VRAM?

If you still insist on option (a), then, well, it's pointless to argue any further.

 

[96Firebird]

So in option B, you wait a year to get a new card? I don't understand why people do this. Also, you're taking hindsight into account, so your whole situation comparison is quite worthless.

How about option C, invest in NFLX back in 2008. If you don't pick option C, then there is no point in posting anymore.

 

[tamz_msc]

So in option B, you wait a year to get a new card?

Lowering a few settings for one year is unacceptable?

Also, you're taking hindsight into account, so your whole situation comparison is quite worthless.

That's called making a decision based on past trends.

How about option C, invest in NFLX back in 2008. If you don't pick option C, then there is no point in posting anymore.

That's called a non sequitur.

 

[96Firebird]

Lowering a few settings for one year is unacceptable?

Why only a year? Why not keep lower settings for 2 years... 3 years?

That's called making a decision based on past trends.

Then why are people waiting for Vega? Fury X showed us there is nothing worth waiting for.

That's called a non sequitur.

No, it isn't.

 

[caswow]

at this point i would not be quite condifent in keeping a nvidia card longer than 1.5-2 years-

 

[tamz_msc]

Why only a year? Why not keep lower settings for 2 years... 3 years?

Sure, why not? As long as performance doesn't suddenly falls off a cliff that is.

Then why are people waiting for Vega? Fury X showed us there is nothing worth waiting for.

Because if Vega is indeed a competitor to the GTX 1080, and actually does a decent job at that, then at least it will bring down the price of the GTX 1080.

No, it isn't.

In the context of this discussion, it is.

 

[Qwertilot]

The 1080's age has a good chance of bringing its price down all by itself, ditto the launch of P102 based stuff.

 

[tential]

So in option B, you wait a year to get a new card? I don't understand why people do this. Also, you're taking hindsight into account, so your whole situation comparison is quite worthless.

How about option C, invest in NFLX back in 2008. If you don't pick option C, then there is no point in posting anymore.

They're not even in the same conversation bracket as I am or others. They're mentioning things like "wait a year".

After waiting for Vega, I'll never wait again. I didn't PC game for the most part. I couldn't because I had to work around the games my gpu could handle...

I'd you're a midrange user and you want a great deal for old cards or whatever great. But again, when you're talking the true high end you only have 1 option. Amd doesn't even get to the table on time

 

[Elfear]

I'd you're a midrange user and you want a great deal for old cards or whatever great. But again, when you're talking the true high end you only have 1 option. Amd doesn't even get to the table on time

Depends on how you look at it I guess. The GTX 780 released 5-6 months ahead of the 290/290X. The 980Ti released ~3 weeks ahead of the Fury X. Depending on which Nvidia card Vega competes with will determine how late it is. If it competes with the 1080, it will be a solid year behind which is abysmally late. If it competes with GP102 though, it should only be a month or two late. Not the end of the world for high-end users.

 

[Jaydip]

I believe you should always buy the fastest card you can afford and which you may actually need, while Titan X was very intriguing I don't actually need it since I mostly play MMOs now a days. I am still waiting for 1080Ti though since there are quite a few single players games in my backlog which 980Ti can not max fully.

 

[richaron]

I have nothing in principal against climbing the "vanishing returns" ladder for hardware, but obviously most people who climb up the top don't need to. Sure it's your money, sure you can afford it, sure it brings you happiness (contrary to the saying), etc, etc. You can use any excuse you want, but unless you're actually making use of a ToTL GPU then you're wasting money. And when I say "making use" of a top of the line GPU I don't mean obsessively running all games at some arbitrary "max" settings. "Max" settings being designed to cripple performance for minuscule IQ gains and make those who wasted money on GPUs feel better about themselves.

When games come out which reasonably need more horsepower, or the user bumps up their resolution and/or refresh rates, that's a different story. And this is why I'm most interested in Volta, hopefully mainstream GPUs will be powerful enough to convince gamers to finally get over their 1080p@60Hz training pants (and/or get into some VR diapers).