NVIDIA to launch next-gen Volta-based GeForce in Q3 2017

[Prysin]

6 hours ago, MageTank said:

You tell me, I'm not the one trying to do it, lol. 

i still don't get why you think i am trying to fill a slot. If you think i am not "like this" normally, you're dead wrong. I just dont act "like this" until i know people and who's who.

[dizmo]

18 hours ago, Humbug said:

depends on Vega. AMD indicated recently that they are targeting upto titan xp performance range with vega. That is motivation for Nvidia to launch Volta.

 

Below from an interview that Tom's hardware conducted

QUESTION : I know you won't be able to say much, but how does Vega compare to the GTX 1080ti and the Titan XP?

ANSWER : Vega performance compared to the Geforce GTX 1080 Ti and the Titan Xp looks really nice.

From AMD's DON WOLIGROSKI.

http://www.tomshardware.com/reviews/amd-ryzen-ama,5018-11.html

If AMD falls just short of their performance targets and Titan XP is still a bit faster then Nvidia will be comfortable. But if it matches or slightly beats Nvidia then they have motivation to launch Volta faster.

I don't think that really says they're going to compete with it, just that it looks nice. For all we know that could just mean that when compared to those, it's price looks good. Say, 60% of the performance but costs 40% of the price. Still wouldn't be bad, but it doesn't mean that it will be on the same level as them. You're reading far too much into that comment.

14 hours ago, laminutederire said:

Because they're seriously afraid of AMD?

I saw someone pointing out that the CPU market is so much more lucrative, and therefore Ryzen success gives an edge to amd in terms of r&d budget on the long term, which hasn't happened for a while (explaining why nvidia could be consistently ahead for a few years). Therefore nvidia would be trying to milk customers as much as they can before amd comes back in a big way and takes the lead. That could explain the new release of the titan Xp, and could be a motivation to cash on a new architecture even it's unfinished or tweak able to be better. 

How would that explain why Nvidia is so far ahead? They don't produce CPUs. Nvidia makes its money from the professional market, where AMD makes next to nothing and Nvidia is king. The margins are much, much better in that sector.

13 hours ago, MageTank said:

-snip-

Where I truly expect AMD to dominate, is APU's. Assuming they get their memory subsystem under control on Ryzen, their APU's are bound to cause an extreme stir in the low/mid segments of the market. With a lot of the markets switching to mobile, APU's are going to be a big deal. Nvidia likely knows this, but there is nothing they can do about it. Tegra will have to wait until ARM takes over, which might be an extremely long time from now. 

I hope so, but that rumored new Intel chip with a more powerful built in GPU might put a damper on that.

I'm interested to see how this batch of APUs perform. So far they haven't been anything too special.

13 hours ago, laminutederire said:

They played somewhat safe because they didn't knew how ryzen would sell. Now that they know it sells fairly well, they know they am have extra budget for both r&d dept and therefore can go the extra mile. The cpu market is ten times larger than the gpu market, that secured them financially. It isn't impossible for them to go back and dethrone nvidia. They can push extra money on the gpu dept and gradually come back. They won't inverse the tendance right away, but two generations of superior product in a row should suffice to make up their brand image.

-snip-

I have a feeling Nvidia is like Canon; they're so far ahead, they don't really need to put out the best of the best right now.

When they see appropriate competition, they'll put the hammer down.

AMD is going to take more than a couple generations of products to really make a comeback. They've been hurting for a long, long time.

12 hours ago, samcool55 said:

Nvidia also invests in self-driving cars and the Tegra which is now used in the switch. AMD doesn't.

And AMD has shown that they can do nice things with a much smaller R&D budget. If you don't R&D as much as your competition you don't need as much money. Makes sense?

If you don't invest as much into R&D, you don't see as much return. Nvidia invested insane amounts of money into Maxwell (or perhaps Pascal). They got excellent sales as a result. While yes, Nvidia has those, you're forgetting that AMD has the console market. That's not exactly a small chunk of change.

[Technomancer__]

On 4/18/2017 at 8:46 AM, Festive said:

- snip -

your profile pic fits you

 

[leadeater]

16 minutes ago, dizmo said:

If you don't invest as much into R&D, you don't see as much return. Nvidia invested insane amounts of money into Maxwell (or perhaps Pascal). They got excellent sales as a result. While yes, Nvidia has those, you're forgetting that AMD has the console market. That's not exactly a small chunk of change.

Still Nvidia makes more profit in a single quarter than AMD does over a few years, it's rather one sided. If AMD had the money they would spend more on R&D but they don't.

 

AMD:

http://ir.amd.com/phoenix.zhtml?c=74093&p=irol-newsArticle&ID=2241045

Nvidia:

FYI: Nvidia is weird with their reporting, minus 1 year off e.g. Q4FY17 = Q4 2016.

http://nvidianews.nvidia.com/news/nvidia-announces-financial-results-for-fourth-quarter-and-fiscal-2017

[Festive]

24 minutes ago, Ethocreeper said:

your profile pic fits you

 

and why is that?

[Technomancer__]

Just now, Festive said:

and why is that?

get it hype

[dizmo]

26 minutes ago, leadeater said:

Still Nvidia makes more profit in a single quarter than AMD does over a few years, it's rather one sided. If AMD had the money they would spend more on R&D but they don't.

 

AMD:

http://ir.amd.com/phoenix.zhtml?c=74093&p=irol-newsArticle&ID=2241045

Nvidia:

FYI: Nvidia is weird with their reporting, minus 1 year off e.g. Q4FY17 = Q4 2016.

http://nvidianews.nvidia.com/news/nvidia-announces-financial-results-for-fourth-quarter-and-fiscal-2017

Yup. It's not hard to beat a company operating on net loss year after year. That doesn't change what I said though, it's still true.  They'll likely never recover at this point, definitely not to a degree they'll beat Nvidia in terms of revenue. 

AMD made missteps and they paid for it. 

[Humbug]

1 hour ago, dizmo said:

 They'll likely never recover at this point, definitely not to a degree they'll beat Nvidia in terms of revenue. 

AMD made missteps and they paid for it. 

I don't know about ever beating NVidia but AMD will make a profit this year. Their losses have been reducing.

[Humbug]

1 hour ago, dizmo said:

I don't think that really says they're going to compete with it, just that it looks nice. For all we know that could just mean that when compared to those, it's price looks good. Say, 60% of the performance but costs 40% of the price.

maybe lol. he did say 'performance' though, not price/performance or price or anything else.

[Hunter115]

On 18.4.2017 at 7:50 AM, deXxterlab97 said:

What comes after 7? 9

What comes after 9? 10

What comes after 10? 20

there was a 8

[laminutederire]

4 hours ago, dizmo said:

 

I smell fan boy around here. Nvidia isn't so far away, they have competition from low to mid range as of yet, just zs they said Polaris would do and they have Vega for upper tier. While we don't know much about how it'll perform in games, we know that it has particularly smart features nvidia doesn't have at all, and those innovation are the one weighting in the balance in the end. Why? Because their cards last longer, and outperform their competitors through time. For smart consumers that means less or at least not forced update cycles and so on. That rewards the consumer, and this is why they have a better branding morally, and in the long run that counts as long as they keep nvidia at arms length, which they have done for years now, because that makes consumer accept to wait a bit longer etc.

As for the professional market, amd innovation are pretty relevant for that segment even more so than for gaming, and most importantly,  intel wants a chunk of that Market and they have so much more resources to do so that you see their fgpa competing quite significantly with nvidia raw horsepower offering, because ultimately sheer performance will always be shadowed by smart performance.

So I don't think so nvidia is so far away of touch. They have competition basically everywhere else than the high end gpus, which Vega will soon attempt to compete with, which shows they're not in a situation to be lazy and wait for real competition since it's already here.

 

2 hours ago, Humbug said:

maybe lol. he did say 'performance' though, not price/performance or price or anything else.

Since he clearly have a worship of nvidia, he won't believe that interpretation which is more literal anyway.

[Mr.Meerkat]

14 hours ago, Jito463 said:

You missed some, starting with Windows 386 (for a bit of hilarity, go to Youtube and search for that phrase, then prepare to watch the the most atrocious marketing video ever), then you missed ME between 98 and NT (or between NT and 2000, take your pick).

Hey, I wasn't the one who made this meme...go complain to the actual creator  

[Mira Yurizaki]

4 hours ago, laminutederire said:

I smell fan boy around here. Nvidia isn't so far away, they have competition from low to mid range as of yet, just zs they said Polaris would do and they have Vega for upper tier. While we don't know much about how it'll perform in games, we know that it has particularly smart features nvidia doesn't have at all, and those innovation are the one weighting in the balance in the end. Why? Because their cards last longer, and outperform their competitors through time.

By the time said hardware starts beating its competitors, it's too old to remain relevant. Sure, the first generation of GCN finally started to bare its fangs once DX12 and Vulkan came around, but by the time DX12 and Vulkan become mainstream, which it likely isn't going to be for a while, a $150 card from either side will smoke them (or rather, that's already happening: http://www.anandtech.com/bench/product/1722?vs=1869)

 

Similarly with the FX series of processors. Some tech websites have found that, when using modern games, the FX series starts closing the gap between it and Sandy Bridge unlike before when Sandy Bridge walloped all over it. But who the hell cares about the FX series now?

 

Buying hardware with the intention of it being an investment is silly at best. You buy what satisfies or exceeds your needs now.

Quote

As for the professional market, amd innovation are pretty relevant for that segment even more so than for gaming, and most importantly,  intel wants a chunk of that Market and they have so much more resources to do so that you see their fgpa competing quite significantly with nvidia raw horsepower offering, because ultimately sheer performance will always be shadowed by smart performance.

No, smart performance will always be better than raw performance, otherwise everyone would've been buying GCN instead of Keplar back in the day. But you know why nobody really bought GCN? Because it ran less efficiently. Performance be damned if you have to tack on an HVAC and electric bill that costs way more than any time lost to lesser performance.

[suits]

If Vega is more than expected than I could see Nvidia releasing it this year. I don't think Vega is going to live up to hype though

[Humbug]

10 minutes ago, M.Yurizaki said:

By the time said hardware starts beating its competitors, it's too old to remain relevant. Sure, the first generation of GCN finally started to bare its fangs once DX12 and Vulkan came around, but by the time DX12 and Vulkan become mainstream, which it likely isn't going to be for a while, a $150 card from either side will smoke them (or rather, that's already happening: http://www.anandtech.com/bench/product/1722?vs=1869

Hmmm I think it's not just Vulkan and DX12.

 

The main point is people who bought AMD cards are very happy with the long term driver support. Because even if you bought a 7970 years ago or just a midrange HD 7850 five years ago they still haven't forgotten you, and they keep releasing driver optimizations for the newest 2017 games.

 

I also bought my R9 290 years ago but today they still roll out new features to me like ReLive, frame rate target control etc rather than limiting them to new products apart, that's apart from the game specific optimizations.

[Jito463]

5 hours ago, Mr.Meerkat said:

Hey, I wasn't the one who made this meme...go complain to the actual creator  

I know, I just wanted an opportunity make a comment about Wa-wa-wa-Windows, Windows 386. 

[cozz]

Here's an update from TweakTown from literally 5 minutes ago, explaining GDDR6 will be coming to high end gpus in 2018

 

http://www.tweaktown.com/news/57250/gddr6-coming-high-end-gpus-early-2018/index.html

[Princess Luna]

11 minutes ago, cozz said:

Here's an update from TweakTown from literally 5 minutes ago, explaining GDDR6 will be coming to high end gpus in 2018

 

http://www.tweaktown.com/news/57250/gddr6-coming-high-end-gpus-early-2018/index.html

I thought the cards would use GDDR5x and HBM2... lets see what happens.

[Drak3]

On 4/18/2017 at 3:04 PM, MageTank said:

Still, if you are going to replace he who shall not be named, I expect more effort from you. Big shoes to fill.

If I exclaim that it's really a die shot of the GTX 210, can I be his replacement for the time being?

[leadeater]

On 4/20/2017 at 2:58 AM, M.Yurizaki said:

No, smart performance will always be better than raw performance, otherwise everyone would've been buying GCN instead of Keplar back in the day. But you know why nobody really bought GCN? Because it ran less efficiently. Performance be damned if you have to tack on an HVAC and electric bill that costs way more than any time lost to lesser performance.

Nvidia owes 90%, if not more, of it's market position in the professional space to CUDA and very little hardware wise. When it mattered CUDA was far superior to OpenCL so people went with Nvidia for that reason, hardware wise between the brands were so close neither was specifically better than the other.

 

If AMD had been able to really develop OpenCL to be competitive with CUDA in the first generation the market would be much different. Now AMD has to fight perception which is hard to change, CUDA is viewed as superior which as of now isn't true for all use cases or the difference is no longer significant and hardware gains now matters.

 

AMD also have exclusive implementations of technology in their hardware which is rather attractive to the people who know what it means. For example AMD has GPUs that support SR-IOV which has been very common on NICs and is widely used in the virtualization market, what SR-IOV allows us to do is at near hardware level create multiple virtual hardware copies of the GPU to share between virtual machines.  GPU sharing is not unique, both vendors can do it but Nvidia does it at the driver and software level and is in a lot of ways less efficient at it, you can have more virtual copies of an AMD GPU for example. Also SR-IOV is an industry standard protocol so implementation and support for upstream vendors such as VMware is very easy.

 

In the professional space it's never been about hardware performance or power draw for AMD it's always been about the inability to use the performance that is in the hardware. Solve that, which they have, means people can actually start to consider using them.

[LienusLateTips]

On 4/18/2017 at 8:36 AM, N1ghtshade said:

w h a t s  t h e  d e a l  w i t h  N v i d i a ?

7 -> 9

9 -> 10

10 -> 20

dingdingdingding simple math u dingus

Don't forget:

9000 -> 2

2 -> 4

 

And with Xbox:

Xbox (original) -> 360 -> 1 -> 1s -> Scorpio

[AntonChigurh]

13 minutes ago, JDE said:

Don't forget:

9000 -> 2

2 -> 4

 

And with Xbox:

Xbox (original) -> 360 -> 1 -> 1s -> Scorpio

Simple maff u dingus

 

Next in the Xbox lineup: The Xbox 420 dank edition. The next generation in MLG swag 

[Notional]

3 hours ago, cozz said:

Here's an update from TweakTown from literally 5 minutes ago, explaining GDDR6 will be coming to high end gpus in 2018

 

http://www.tweaktown.com/news/57250/gddr6-coming-high-end-gpus-early-2018/index.html

 

Yup, which proves a few things:

 

1. Volta will be out in 2018, as stated before. Maybe we will see a Volta paper launch in december or pro products. Geforce? Not a chance.
2. Volta consumer will use GDDR6, not HBM2.
3. Volta will be 16nm, not 10.

[leadeater]

26 minutes ago, Notional said:

Yup, which proves a few things:

 

1. Volta will be out in 2018, as stated before. Maybe we will see a Volta paper launch in december or pro products. Geforce? Not a chance.
2. Volta consumer will use GDDR6, not HBM2.
3. Volta will be 16nm, not 10.

Probably will get a Volta based Tesla or Quadro card with HBM if this year, pretty sure the professional line of Pascal came first but could be wrong since I wasn't that interested.

[Notional]

Just now, leadeater said:

Probably will get a Volta Tesla or Quadro card with HBM if this year, pretty sure the professional line of Pascal came first but could be wrong since I wasn't that interested.

Pascal with HBM2 exists yes, but they are pro cards, tesla's I believe. They are not relevant for Geforce anymore. They are entirely different chips today.