Does the RTX series create an openning for AMD?

[Dribble]

A 7nm card from AMD now will not obsolete anything, it will only be a high-end product selling at 600+ USD leaving the rest of the line intact, much like NVDIA did with RTX cards.

It would cost more due to the cost of 7nm and them being only able to make a few. Even if they could do 7nm that cheaply no one would buy it - no one was buying the 290 when it was competing with the 970 (18% market share for AMD then), and that was a lot more competitive. Now all AMD has is the same GCN which is limited to 64 rops/4096 cores and doesn't have ray tracing or AI. That's forces it to be mid range at best, so it's no $600 competitor, it's a $300 part at the most. The numbers don't add up.

AMD are not stupid - they aren't going to waste money fighting a loosing battle with Nvidia when they could make much more money making CPU's. That's got to stay the focus. The gpu team can make the next console chips - another pretty assured source of income. The PC market is basically Nvidia's for now.

 

[Muhammed]

7nm AMD consumer GPUs will come AFTER Ryzen 7nm.

It will be like this: Vega 20 for HPC (late 2018) > Epyc Rome (2019) > Ryzen 3000 > Navi.

So late 2019, early 2020.

Q: And the absolute last question on the graphics side, 7 nanometer Vega coming to the data center side of it, you've talk about that before at the end of this year. When should we expect 7 nanometer to occur on the more traditional gaming…

A: We haven’t missed that piece. I think, if you look at it from what we have stated, we have 7 nanometer data center GPU launching later this year; we are sampling the 700 CPU this second half of ’18 and then launching in 2019; after that, we'll have the client piece of it; we haven’t been specific about the timing; and graphics will be coming out later than these products.

https://www.reddit.com/r/Amd/comments/9frjoe/amd_cfo_devinder_kumar_on_7nm_products_schedule/

 

[AtenRa]

It would cost more due to the cost of 7nm and them being only able to make a few. Even if they could do 7nm that cheaply no one would buy it - no one was buying the 290 when it was competing with the 970 (18% market share for AMD then), and that was a lot more competitive. Now all AMD has is the same GCN which is limited to 64 rops/4096 cores and doesn't have ray tracing or AI. That's forces it to be mid range at best, so it's no $600 competitor, it's a $300 part at the most. The numbers don't add up.

Sorry but 14nm VEGA 10 can compete up to GTX1080 even today, a 7nm VEGA will be more than capable to compete with GTX1080Ti and up. The numbers do add up, even just porting VEGA 10 to 7nm will create a product with sub 200W TDP and performance close to GTX1080Ti. Such a product could easily be sold at 600-700+ USD MSRP leaving current VEGA 64 at its current MSRP of $499.

AMD are not stupid - they aren't going to waste money fighting a loosing battle with Nvidia when they could make much more money making CPU's. That's got to stay the focus. The gpu team can make the next console chips - another pretty assured source of income. The PC market is basically Nvidia's for now.

Im the one that already have said that im expecting AMD to focus heavily on CPUs at 7nm in 2019, but IF they would release a VEGA 20 for the consumer in H1 2019 it wouldnt directly compete with the existing Polaris and VEGA 10 cards. I wasn't implying they will release a consumer VEGA 20, I was just explaining that such a card will be in a different higher priced segment than current cards and will not directly compete with them.

 

[SpaceBeer]

I don't think is't production cost, but the TSMC capacity that might be the issue. Even if they get only 40 good Vega 20 dies per wafer, they could easily fit card cost under $400, which would be enough to sell it for $600+ if it is GTX 1080Ti / RTX 1080 competitor

But why would they do that when they can use that limited capacity and produce 7nm Epyc with the $200 cost and sales price of $1500+

 

[JustMe21]

If AMD could do a better 14nm or lower process and price the Vega 64 at RX580 level, plenty of people wouldn't care about the high end. There's many more people buying video cards around the mid range level than at the high end. The new generation of Nvidia cards aren't so appealing because of their cost and the near future release of 7nm cards and hardware Ray Tracing has to evolve a few generations.

 

[Kenmitch]

I don't think is't production cost, but the TSMC capacity that might be the issue. Even if they get only 40 good Vega 20 dies per wafer, they could easily fit card cost under $400, which would be enough to sell it for $600+ if it is GTX 1080Ti / RTX 1080 competitor

But why would they do that when they can use that limited capacity and produce 7nm Epyc with the $200 cost and sales price of $1500+

The current focus at AMD is nothing more than producing products they know they can sell. When timing is right for AMD they'll offer a compelling products and then we'll see all the too little too late, can barely beat a XXXX, etc that is used to justify buying the nvidia product anyways. Wanting AMD to respond just for some pricing pressure on nvidia is somewhat silly and selfish. It really sucks for those who actually buy AMD's offerings....I feel for ya!

This go around the middle finger looks to be on Lisa Su's hand. She's no dummy and has most likely ran the numbers.

 

[krumme]

Seems like the majority just wants AMD to make Nvidias prices lower. If you wouldn't buy the products then why bitch they haven't made them? I've read too many comments around the net to think otherwise.

Thats what i observe too. There is too types of buyers. Those that buy buy nv or amd and those that only buy nv. Happens for all big brands and why building a brand is so darn important and valuable investment strategy.

 

[AtenRa]

Im sure the majority of buyers would buy an AMD card IF it was faster than NV, it happened before (HD5870) and it could happen again.

 

[Geegeeoh]

I don't think AMD sold more than nVidia… ever. Even in the best days.

 

[AtenRa]

I don't think AMD sold more than nVidia… ever. Even in the best days.

In HD5870/5850 time, AMD sold more high-end ($300+) GPUs than NVIDIA.

 

[Kenmitch]

In HD5870/5850 time, AMD sold more high-end ($300+) GPUs than NVIDIA.

I don't remember nvidia having a cult like following way back then. You've been around long enough to see the transformation.

When the 290x slayed the mighty TItan what happened? Too hot, too loud, too power hungry, etc....The new standard was performance per watt, temps, noise,etc. Do you remember how fraps was used to show AMD's inferiority and then dropped once AMD's offerings were better?

If you view the comments on the multiple articles around the web about nvidias latest offerings it doesn't take long to come to the conclusion that the end user wants to use AMD as a bargaining chip to get lower prices on their offerings. There are some end users that would purchase a AMD gpu based on it's performance and value. The majority just follow the sheep or use some silly little metric that's in their mind justifies the purchase.

I think LIsa Su decided that the time isn't right to jump back into the game just yet. I don't think it's they can't make it, but more like they just choose not too at this time.

 

[PeterScott]

In HD5870/5850 time, AMD sold more high-end ($300+) GPUs than NVIDIA.

AMD never had more market share than NVidia. ATI last had more market share in the time of the X800-X850:

 

[Malogeek]

AMD never had more market share than NVidia. ATI last had more market share in the time of the X800-X850:

The chart doesn't disprove his statement, in fact it shows it's likely true. He said it sold more, not that they had more market share.

 

[PeterScott]

The chart doesn't disprove his statement, in fact it shows it's likely true. He said it sold more, not that they had more market share.

The chart is market share of sales, not the accumulation of what everyone in the past bought.

So in this case, sold more == more market share.

 

[krumme]

Nv wiped off 3dfx with a weaker and worse tnt2. Got good reviews even back then. Their PR and grip on reviewers was superior back then even when their brand was weak and their marketing organization small. They are that good.
Now their brand is huge. They got the best cards. It wouldnt matter as much today if amd got something like 5850 again.
Every new stupid argument is invented. This year its die area. Back tnt2 days it was 32 bit color was needed.
Each year its like that. And as mentioned what is important constantly changes. Its what brands do.

 

[Kenmitch]

Nv wiped off 3dfx with a weaker and worse tnt2. Got good reviews even back then. Their PR and grip on reviewers was superior back then even when their brand was weak and their marketing organization small. They are that good.
Now their brand is huge. They got the best cards. It wouldnt matter as much today if amd got something like 5850 again.
Every new stupid argument is invented. This year its die area. Back tnt2 days it was 32 bit color was needed.
Each year its like that. And as mentioned what is important constantly changes. Its what brands do.

If it came down to both AMD and Nvidia having the same exact performance, watts, temps, features, etc and Nvidia was priced higher the new metric would be....If it costs more then it has to be better. lol

 

[PeterScott]

Nv wiped off 3dfx with a weaker and worse tnt2. Got good reviews even back then. Their PR and grip on reviewers was superior back then even when their brand was weak and their marketing organization small. They are that good.

Or, OTOH, most good reviewers were unbiased as they have generally been, and your personal bias made you see bias in review that didn't agree with your biased worldview.
Anands TNT2 reivew:
https://www.anandtech.com/show/288/32

TNT2 had superior image quality, and was faster in Direct3D. 3DFX was faster in OGL/Glide. So it wasn't like 3DFX was better across the board.

BTW. I owned a Voodoo 3 back then (AFAIK I still have it somewhere).

 

[Malogeek]

The chart is market share of sales, not the accumulation of what everyone in the past bought.

So in this case, sold more == more market share.

I stand corrected then, thanks.

 

[AtenRa]

AMD never had more market share than NVidia. ATI last had more market share in the time of the X800-X850:

You have to read again, I said AMD sold more high-end GPUs ($300+) at the time of HD5870 I never said they had more market share.
AMD released HD5870/5850 in September 2009 and HD5970 (dual Cypress) in November 2009, at the time NVIDIA was struggling to sell GTX285 and GTX275 for 9 months until they would release the FERMI in March 2010. If i remember correctly in Q1 2010 AMD sold more High-End GPUs at $300 and up than NVIDIA, simple because Cypress was in every way the better product than NVIDIA offerings.
NVIDIA was still getting higher overall market share at the time simple because they could move more capacity at the entry level.

 

[AtenRa]

I don't remember nvidia having a cult like following way back then. You've been around long enough to see the transformation.

When the 290x slayed the mighty TItan what happened? Too hot, too loud, too power hungry, etc....The new standard was performance per watt, temps, noise,etc. Do you remember how fraps was used to show AMD's inferiority and then dropped once AMD's offerings were better?

If you view the comments on the multiple articles around the web about nvidias latest offerings it doesn't take long to come to the conclusion that the end user wants to use AMD as a bargaining chip to get lower prices on their offerings. There are some end users that would purchase a AMD gpu based on it's performance and value. The majority just follow the sheep or use some silly little metric that's in their mind justifies the purchase.

I think LIsa Su decided that the time isn't right to jump back into the game just yet. I don't think it's they can't make it, but more like they just choose not too at this time.

I agree in most of what you are saying, but im sure that IF AMD had the fastest card, the majority of the High-End TAM would buy. There would still be people that would buy NVIDIA but most of the High-End consumers only care about performance.

 

[krumme]

Or, OTOH, most good reviewers were unbiased as they have generally been, and your personal bias made you see bias in review that didn't agree with your biased worldview.
Anands TNT2 reivew:
https://www.anandtech.com/show/288/32

TNT2 had superior image quality, and was faster in Direct3D. 3DFX was faster in OGL/Glide. So it wasn't like 3DFX was better across the board.

BTW. I owned a Voodoo 3 back then (AFAIK I still have it somewhere).

Since when did you pay for the reviews directly?

Its totally naive to think reviewers are unbiased. You fund technical marketing for a reason. And you wouldnt invest if you couldnt change reviewers mind. Same goes for investment in brand and marketing.
I dont know what makes you blind of that fact? Its pretty obvious thats how it works. I dont understand your reaction. Do you have some indirect or direct interest in nv?

Go look at thg original tnt2 review. Its was as baltantly biased as their current nv RT ramblings.

 

[krumme]

Another example of different standard is this talk from the same users about vega beeing uncompettitive because of big die and hbm. Makes it a bad gpu. For consumers that is.
But when it comes to nv then the high prices is explained by the huge die. So nv can get away with selling gpu at absurd prices.
Whaever is needed to fit the narrative.

 

[PeterScott]

Since when did you pay for the reviews directly?

Its totally naive to think reviewers are unbiased. You fund technical marketing for a reason. And you wouldnt invest if you couldnt change reviewers mind. Same goes for investment in brand and marketing.
I dont know what makes you blind of that fact? Its pretty obvious thats how it works. I dont understand your reaction. Do you have some indirect or direct interest in nv?

Go look at thg original tnt2 review. Its was as baltantly biased as their current nv RT ramblings.

I said good reviewers were unbiased. What does THG have to do with Good reviewers?

Also I linked decent TNT2 review from that time, AnandTechs. Anandtech has always seemed professional and unbiased. What is your problem with that review?

 

[krumme]

I said good reviewers were unbiased. What does THG have to do with Good reviewers?

Also I linked decent TNT2 review from that time, AnandTechs. Anandtech has always seemed professional and unbiased. What is your problem with that review?

At tnt2 time thg was actually a respected site. Thats years back.
Technical marketing serve a purpose for respected sites also. Its what makes the company seem professional from the reviewers perspective. They feed them information and support. In return they get better reviews. Thats how it works. Goes for amd as well as nv. Nv just does a better job here. I dont have anything against that specific at review.
Still you dont answer my question. Do you have any inderect or direct interest in nv?

 

[PeterScott]

Another example of different standard is this talk from the same users about vega beeing uncompettitive because of big die and hbm. Makes it a bad gpu. For consumers that is.
But when it comes to nv then the high prices is explained by the huge die. So nv can get away with selling gpu at absurd prices.
Whaever is needed to fit the narrative.

Nonsense.

Big dies present exactly the same problem for both NVidia and AMD. It increases production costs, and usually some/all of that gets passed onto the consumer in the form of increased retail price.

Right now Vega 64 is going for almost $100 more than GTX 1080, despite both companies reportedly having excess stock to clear. That bigger die on Vega is a large part of the reason Vega 64 is trying to sell for a much higher price.

It's also a large factor in why the new RTX card are much more expensive.

That is reality. Big dies increases production costs and selling price for both NVidia and AMD. That reality doesn't have a bias.

The only issues are from people who want to pretend that reality doesn't exists, and that massive dies don't drive up production costs.

So they pretend that those big dies were never a problem for AMD's competitiveness, and that the price increases from NVidia have nothing to do with increased die size.

If you need a bigger die than your competitor to deliver the same performance, then it will hamper how competitive you can be. The bigger the delta, the bigger the problem, because you can't really compete on price without killing your margins.

That applies equally to NVidia and AMD.

It's just that until RTX, AMD was suffering the Big die disadvantage.

Now that RTX has massive die, it could turn the tables. Which, BTW, is exactly the point of this thread.

Can AMD take advantage of NVidias big die problem? The RTX tax on die size is the exactly the opening we are discussing.