gtx 1180 leaks and infofmation from wccftech

[dizmo]

3 hours ago, Dionyz said:

July??? They are not going to release it anytime soon. You have to understand business. If they are selling out of their current cards then why would they release a new better product? It would be retarded for Nvidia to release a product anytime soon.

You need to understand release cycles.

They're not going to hold back releasing new cards. Why else would the companies be ramping up GDDR6 production?

It's the same thing with the automotive industry; they don't really need to release new models every year, but they do.

[thorhammerz]

1 hour ago, dizmo said:

It's the same thing with the automotive industry; they don't really need to release new models every year, but they do.

There's a key difference between the GPU and automotive industries though: the automotive industry makes a LOT of money on supplementary services such as servicing/maintenance, and financing.

 

All of which benefit substantially from consumers buying new cars (e.g. the "latest and greatest" models with all the fancy heated seats, etc) as often as they can be fooled into doing so - accomplished in part by making something marginally shinier each and every year.
 

Companies (NVIDIA is no exception) will seek to maximize their bottom line - if holding off a product release because they believe sitting on their current stack of products will make them more money, then they will not hesitate to do so.

[WereCat]

6 hours ago, thorhammerz said:

 

Companies (NVIDIA is no exception) will seek to maximize their bottom line - if holding off a product release because they believe sitting on their current stack of products will make them more money, then they will not hesitate to do so.

All the money that went into R&D to make a new architecture, manufacturing cost, marketing, etc... That's just talking from NVIDIA standpoint, makes no sense to hold off the product, otherwise these money, time and resources would be wasted.

 

Also, if Hynix makes GDDR6 and the GPU manufacturers decide to hold off and wont buy the DRAM, they will screw both themselves and Hynix or other DRAM manufacturer in the same position.

 

These are products that have to fly off the shelves to sustain the business going, holding off the product is possible but very costly for each side.

[michaelocarroll007]

19 hours ago, Coaxialgamer said:

7nm and 12nm are just technology nodes , roughly telling of how small the individual transistors in the chip will be  , and thus how many transistors/performance you can fit in a given amount of space.

The chip die size tells you how large the actual chip size is , in the case roughly 400mm² , which is fairly typical for a gpu . In theory , the true cost of a chip is determined with the die size , and stays the same regardless of the technology node ; ie the cost per mm² is the same .In practice , the cost of area has been increasing in the last few years , and a 7nm , 400mm² chip should cost more than an equivalent 12nm one . But that should be offset by the increased integration provided by the more advanced node .

also , it's worth noting that TSMC's 12nm is in production now . It's not even their most advanced tech , as they are also producing 10nm chips and are working on 7nm . There are currently no production-ready 7nm process nodes , and they won't be available until well within H2 , at the earliest . 7NM will be at the very bleeding edge , and as such presents a much riskier proposition to AMD , as well as a greater reward

TFLOPS basically mean nothing . There is a reason it isn't compared across architectures , and it's because the actual calculation doesn't take it into account . I can tell you exactly how many flops a 3ghz nvidia turingator gpu with 9600 cuda cores will put out ( 57.6TFLOPS ) , and that's because the formula is :

clocks * ALU_count(cores) * 2 = peak floating point throughput

You are correct its a bad example But nvidia Only Compairsons so far last few years have been pretty Decent scales. 

[thorhammerz]

2 hours ago, WereCat said:

All the money that went into R&D to make a new architecture, manufacturing cost, marketing, etc... That's just talking from NVIDIA standpoint, makes no sense to hold off the product, otherwise these money, time and resources would be wasted.

Why would a company release a painfully expensive new products right now, when there is absolutely no competitive need to do so: selling more of last generation's products would continue to make you the same or more money (compared to releasing a new product stack)? 

 

Why would a company not hold off their latest and greatest when releasing them would not increase profits (read: Pascal cards sell out anyways, there is no competition (cost included, Vega 56/64 doesn't really function as 'competition' when they are substantially pricier) worthy of the name from Intel/AMD above the GTX 1060).

 

If consumers continue to buy current products, then devoid of other market factors (materials shortages, or whatever), then it is better for business to simply keep something waiting in the wings until the consumers begin to buy less of their current products (in which case you pull an Intel and do a marginally-better performance refresh), or until the competition comes up with something remotely useful (and available to purchase at a competitive price point).

 

As a side note, Nvidia is not spending all of that R&D on consumer gaming products - AI and data-center related products (of which no doubt some elements will eventually seep back into the consumer space) is where the profit margins are at (and where the release of the "latest and greatest" products occur - such as the Volta cards).

[Coaxialgamer]

2 hours ago, michaelocarroll007 said:

You are correct its a bad example But nvidia Only Compairsons so far last few years have been pretty Decent scales. 

that's because they've been using the same base architecture since 2014 ?

[Razor01]

2 hours ago, thorhammerz said:

Why would a company release a painfully expensive new products right now, when there is absolutely no competitive need to do so: selling more of last generation's products would continue to make you the same or more money (compared to releasing a new product stack)? 

 

Why would a company not hold off their latest and greatest when releasing them would not increase profits (read: Pascal cards sell out anyways, there is no competition (cost included, Vega 56/64 doesn't really function as 'competition' when they are substantially pricier) worthy of the name from Intel/AMD above the GTX 1060).

 

If consumers continue to buy current products, then devoid of other market factors (materials shortages, or whatever), then it is better for business to simply keep something waiting in the wings until the consumers begin to buy less of their current products (in which case you pull an Intel and do a marginally-better performance refresh), or until the competition comes up with something remotely useful (and available to purchase at a competitive price point).

 

As a side note, Nvidia is not spending all of that R&D on consumer gaming products - AI and data-center related products (of which no doubt some elements will eventually seep back into the consumer space) is where the profit margins are at (and where the release of the "latest and greatest" products occur - such as the Volta cards).

There is no revolving door in GPU's in the OEM segment where this is the lions share of GPU sales, unlike in CPU and OEM business segment where most IT handbooks of major corporations have laid out to buy new computers ever 5 years or so.  By holding things back too long after actually creating the product is a waste of money.  Too long I mean, if creating the new chip on a new process is more costly due to the cost of a wafer being higher.  So waiting till that drops down to what current wafer costs are on current nodes or close to it, is a smart choice, anything after that though, they will be losing potential money.  Its a balance of scales and economics at the end that will push the product out the door. 

 

Fixed costs like R&D are factors that have to be looked into when bringing a chip on to the market.  Holding on to something too long on the back burner doesn't make much sense because it gives time for the competition to catch up.  This is a mature and saturated market, that is why we don't see increased GPU sales per year, the same people are buying a certain amount of products.  Without that revolving door Pascal sales will diminish.  Also the gaming market is by far where nV makes most of its money, so AI and HPC markets till those pick up more, even though they are high margin markets as you stated, don't recover enough of the money to keep pushing R&D.

 

This has very little do to competition because even with Pascal there is little competition, with next gen it will be no competition.

[Mira Yurizaki]

1 hour ago, thorhammerz said:

Why would a company release a painfully expensive new products right now, when there is absolutely no competitive need to do so: selling more of last generation's products would continue to make you the same or more money (compared to releasing a new product stack)?

Because not everyone can afford $500 (MSRP) video cards and if you continue releasing the same product over and over without apparent improvements, then you'll eventually saturate the market and sales will cave in, only picking up somewhat when people start buying replacements. But the cycle of buying replacements is typically longer than the cycle of upgrading so you'll never reach the revenue stream you once had. Then investors who were all for you start to see you're stagnating in the market and so they'll start to care less and less.

 

So yeah, even if you're the king, it's not really a good idea to sit on your ass in the tech sector.

[Dionyz]

23 hours ago, dizmo said:

You need to understand release cycles.

They're not going to hold back releasing new cards. Why else would the companies be ramping up GDDR6 production?

It's the same thing with the automotive industry; they don't really need to release new models every year, but they do.

For automotive yes they do. Look up product life cycle. Automotive has a trend which a lot of people like new stuff, so they purchase the newest year model or rent the newest year one. If there were no sales with newer models then they would stop this trend.

 

As a marketer for Nvidia to release a new card when their current card is out of stock constantly would be stupid on their part to release a new product that's better... They are going against the PLC. 

[dizmo]

3 minutes ago, Dionyz said:

For automotive yes they do. Look up product life cycle. Automotive has a trend which a lot of people like new stuff, so they purchase the newest year model or rent the newest year one. If there were no sales with newer models then they would stop this trend.

 

As a marketer for Nvidia to release a new card when their current card is out of stock constantly would be stupid on their part to release a new product that's better... They are going against the PLC. 

Ok, so you just proved my point. The new card is better. More performance, akin to the extra options you get with an automobile.

Are you actually trying to imply that PC enthusiasts don't like getting new stuff every year?
How are you not seeing this?

 

Not only that, but the cards are out of stock due in part to component shortages, as well as an impending release; you don't manufacture tons of last gen cards when you have a new gen about to launch.

[Dionyz]

8 minutes ago, dizmo said:

Ok, so you just proved my point. The new card is better. More performance, akin to the extra options you get with an automobile.

Are you actually trying to imply that PC enthusiasts don't like getting new stuff every year?
How are you not seeing this?

 

Not only that, but the cards are out of stock due in part to component shortages, as well as an impending release; you don't manufacture tons of last gen cards when you have a new gen about to launch.

I am not arguing with the new card not selling out. I am sure it will sell out, but the problem with releasing a new card right now is that there is no point. They cannot maintain markets demand for the older card, so what makes them capable with fulfilling the demand for the newer card? You don't get to manufacture more graphics chips out of nowhere. You need manufacturing facilities, and those facilities are right now bottlenecking the amount they can produce, since again they cannot meet demand. So Nvidia will have to tell the manufacturing to slow down production for the older card, so they can produce the newer one... It's extremely inefficient to do that.

 

Automotive has the issue with their cars not selling out anymore, so they have to make a newer model. These two cases are different. Nvidia is selling out of their cards, but automotive is not.  

[Canada EH]

All I care about is when the GTX 1180 comes out.

[Humbug]

23 hours ago, thorhammerz said:

Companies (NVIDIA is no exception) will seek to maximize their bottom line - if holding off a product release because they believe sitting on their current stack of products will make them more money, then they will not hesitate to do so.

To a limited extent. But one thing you have to respect about Nvidia is that they keep innovating and getting faster with new generations regardless of how fast the competing AMD architecture is at that moment. When AMD is faster than them it does have an impact on Nvidia prices, but for the most part even when AMD is slower Nvidia keeps marching forward albeit at a price premium.

 

This is unlike Intel in the CPU market who seems to need AMD to light a fire underneath them in order to really get moving.

[laminutederire]

3 hours ago, Humbug said:

To a limited extent. But one thing you have to respect about Nvidia is that they keep innovating and getting faster with new generations regardless of how fast the competing AMD architecture is at that moment. When AMD is faster than them it does have an impact on Nvidia prices, but for the most part even when AMD is slower Nvidia keeps marching forward albeit at a price premium.

 

This is unlike Intel in the CPU market who seems to need AMD to light a fire underneath them in order to really get moving.

They don't really have a choice. AMD had compelling competition up to 1070-1070ti range,  and they have an alternative at 1080 level. If they don't push higher tech and catches up, since they are just one tier below performance wise. I'd say it's exactly like Ryzen now. They concede the absolute top end (against the 18c parts of Intel for instance in productivity or the 8700K in gaming), and then they give good alternatives for every parts below.

[Taf the Ghost]

13 minutes ago, laminutederire said:

They don't really have a choice. AMD had compelling competition up to 1070-1070ti range,  and they have an alternative at 1080 level. If they don't push higher tech and catches up, since they are just one tier below performance wise. I'd say it's exactly like Ryzen now. They concede the absolute top end (against the 18c parts of Intel for instance in productivity or the 8700K in gaming), and then they give good alternatives for every parts below.

X299 was in 2016 planned to be a 10c top-end line. The i9-7900X was the top SKU. Late in 2016, they added a 12c model to it, so the i9-7920X. After Threadripper got confirmed, they rushed to add the 14c, 16c & 18c parts. They were never going to be on X299 until about March 2017, only a few months before the release. (This is why a bunch of early X299 motherboards had VRM issues under load, as they were designed for a different thermal amount.)

 

Nvidia is also iterating because events like how badly AMD caught Intel off guard with Threadripper happen in Tech. Nvidia can't allow that to happen in the core markets, as that'd be disastrous for them. Gaming is still Nvidia's biggest single sales category and over half of their total revenue. The big change, for Nvidia, has been in Data Center, which is massively profitable.

 

This is mostly why Nvidia is going to launch the 11x0 series on the TSMC 12FF node with GDDR6 going into the H2 2018. Their big time for selling Gaming GPUs is H2, which should help encourage the replacement of 960s & 970s by putting them 2 generations behind current. (And clearing out previous inventory.) They'll be out in front of AMD in a per-price category.

 

AMD is rushing towards 7nm in both CPU & GPU as fast as they can. AMD still has a pretty big issue with the scaling of the GCN architecture, but if they can drive down Nvidia's margins, they'll be able to stay within the market well enough. And it's not like AMD & Nvidia won't both sell out their entire year of production, if they don't increase the production. (Both are DRAM limited, realistically.)

[laminutederire]

12 minutes ago, Taf the Ghost said:

 

My point with Intel is just a comparison of current lineups. I'm not saying that Intel innovated as much as Nvidia at all ( which in many ways wouldn't be true, even if Intel may have one of the best processes of the industry. Ryzen on Intel's process would probably be skylake- kaby lake clocks with coffee lake core counts).

I think Nvidia isn't scared about and coming back on top. They know it'll happen. It has happened many times. They try to encrust themselves to get sales even with inferior products (with GPP, or other mind share and partnerships they build when they are on top).

In the per price category well have to see. It does seem like this line up will be the current line up with names rounded down a tier and prices between current tier. 1180 is a 1080ti with a price in the middle of 1080 and 1080ti. Like it would seem a bit artificial progress. Bit it helps mind share and they want to cash in on people fed up with mining prices inflation who will tho k it's a good deal only because prices are inflated currently.

At those prices by the way, they have to release it before Navi if rumours are to be true (1170 would be big Navi  (1080 perf) but the 1170 would clearly not be at 250$ at all given current trends.

Yeah they do seem to be often limited by manufacturing processes clockwise (Like the 4Ghz wall for Ryzen (which has been moved up to 4.3Ghz now if I remember right)

[cj09beira]

25 minutes ago, laminutederire said:

Yeah they do seem to be often limited by manufacturing processes clockwise (Like the 4Ghz wall for Ryzen (which has been moved up to 4.3Ghz now if I remember right)

7nm should be a great equalizer for the whole industry, it will be very fun 

[Taf the Ghost]

19 minutes ago, laminutederire said:

My point with Intel is just a comparison of current lineups. I'm not saying that Intel innovated as much as Nvidia at all ( which in many ways wouldn't be true, even if Intel may have one of the best processes of the industry. Ryzen on Intel's process would probably be skylake- kaby lake clocks with coffee lake core counts).

I think Nvidia isn't scared about and coming back on top. They know it'll happen. It has happened many times. They try to encrust themselves to get sales even with inferior products (with GPP, or other mind share and partnerships they build when they are on top).

In the per price category well have to see. It does seem like this line up will be the current line up with names rounded down a tier and prices between current tier. 1180 is a 1080ti with a price in the middle of 1080 and 1080ti. Like it would seem a bit artificial progress. Bit it helps mind share and they want to cash in on people fed up with mining prices inflation who will tho k it's a good deal only because prices are inflated currently.

At those prices by the way, they have to release it before Navi if rumours are to be true (1170 would be big Navi  (1080 perf) but the 1170 would clearly not be at 250$ at all given current trends.

Yeah they do seem to be often limited by manufacturing processes clockwise (Like the 4Ghz wall for Ryzen (which has been moved up to 4.3Ghz now if I remember right)

My guess is Nvidia is going to shift their entire stack up about 50USD in MSRP, or more, given the persistent GPU prices (Cryptos sell off in the beginning of the year then rise in the end of the year, for reasons I don't know). By launching into their big selling season at relatively good prices, they should sell out their expected sales stock. AMD's Navi won't effect them until it actually lands.

 

Given that 14nm -> 7nm generations are about 1.5 Nodes worth of previous shrink level, we're really unsure what AMD could be dropping on 7nm with Navi. We can assume it's going to be a fairly sizable generational improvement (node shrinks do that), but that's all where looking at right now. (I do hope it's a 48 CU Navi and then cut-down versions of that die, but we'll see.) 

[Taf the Ghost]

Just now, cj09beira said:

7nm should be a great equalizer for the whole industry, it will be very fun 

Yup. AMD has clearly aligned their focus to get on the node as fast as possible, while Nvidia could very easily do a 1 year generation on 12FF then move to TSMC's 7nm process by late 2019, if Navi comes out really good.

 

7nm "generation" is also where almost everyone catches up to Intel, which is pretty much the first time that's happened since about 1999 or something like that.

[cj09beira]

7 minutes ago, Taf the Ghost said:

Yup. AMD has clearly aligned their focus to get on the node as fast as possible, while Nvidia could very easily do a 1 year generation on 12FF then move to TSMC's 7nm process by late 2019, if Navi comes out really good.

 

7nm "generation" is also where almost everyone catches up to Intel, which is pretty much the first time that's happened since about 1999 or something like that.

it should give them a good head start, but even for the midrange it would be best if they could make the gpu pipeline more Thicc, that scalable thing has me really dying to know what it means, it will also show what raja was able to muster

Quote

Given that 14nm -> 7nm generations are about 1.5 Nodes worth of previous shrink level, we're really unsure what AMD could be dropping on 7nm with Navi. We can assume it's going to be a fairly sizable generational improvement (node shrinks do that), but that's all where looking at right now. (I do hope it's a 48 CU Navi and then cut-down versions of that die, but we'll see.) 

i think they need to go all out or risk getting destroyed when nvidea arrives on 7nm as they aren't limited by how far they can scale their gpu, 

even a full sized vega die would be small enough 

[Taf the Ghost]

1 minute ago, cj09beira said:

it should give them a good head start, but even for the midrange it would be best if they could make the gpu pipeline more Thicc, that scalable thing has me really dying to know what it means, it will also show what raja was able to muster

Well, some discussion by Looncraz has me thinking they might have found a way to deal with some of the GCN design limitations, as it's really the problem with GCN. Adding more CUs has quick diminishing returns, so it would make sense if they found some ways to mitigate that.

 

Though I really want to see MCM come to GPUs, just because of the possibilities that it brings. However, someone at AMD would have needed to come up with some really clever tricks to pull it off, even if it's some really small logic controller you have to place between the GPUs.

[cj09beira]

6 minutes ago, Taf the Ghost said:

Well, some discussion by Looncraz has me thinking they might have found a way to deal with some of the GCN design limitations, as it's really the problem with GCN. Adding more CUs has quick diminishing returns, so it would make sense if they found some ways to mitigate that.

 

Though I really want to see MCM come to GPUs, just because of the possibilities that it brings. However, someone at AMD would have needed to come up with some really clever tricks to pull it off, even if it's some really small logic controller you have to place between the GPUs.

i think they need to do both anyways because even if they go mcm they will endup at the 64rop/cu limit in no time flat (good chance it could happen in the second gen),which one will be first i have no clue, but vega already having IF gluing everything together could point to a design mcm in the works

[Cheezdoodlez]

Are we back to calling it 1180? I thought the rumors were 2080 and if I recall correctly there was also a rumor with a more confusing naming scheme similar to intels naming schemes? 

[Taf the Ghost]

4 minutes ago, cj09beira said:

i think they need to do both anyways because even if they go mcm they will endup at the 64rop/cu limit in no time flat (good chance it could happen in the second gen),which one will be first i have no clue, but vega already having IF gluing everything together could point to a design mcm in the works

The IF in Vega might have more to do with future APUs, so they can skip using PCIe to the GPU. It also helps with the control layers.

 

The scalable part might just be changes with the way memory is handled, as AMD is in a weird place where both their CPUs & GPUs have memory bottlenecks in this generation. But, even if the CU/ROP limit hasn't been worked through, a 32CU x2 design could allow them to put it all on GDDR6 rather than having to use HBM2 in their Enthusiast card. Also would allow for a 48 or 56 CU part a lot easier.

[cj09beira]

1 minute ago, Taf the Ghost said:

The IF in Vega might have more to do with future APUs, so they can skip using PCIe to the GPU. It also helps with the control layers.

 

The scalable part might just be changes with the way memory is handled, as AMD is in a weird place where both their CPUs & GPUs have memory bottlenecks in this generation. But, even if the CU/ROP limit hasn't been worked through, a 32CU x2 design could allow them to put it all on GDDR6 rather than having to use HBM2 in their Enthusiast card. Also would allow for a 48 or 56 CU part a lot easier.

having 2 dies that small could be good or bad, it makes even the 580 tier need more than one die but it could allow for having 64rops and 32 cus which if they can have 4 dies it could be really powerful, hmm there might be a reason to only have 32 cus after all