[Rumor, Tweaktown] AMD to launch next-gen Navi graphics cards at E3

[maddie]

Ryzen was only well priced because of how bad Intel was ripping us off over the years.

We were paying 350 dollars for dies about 114mm2 with skylake 4 cores with no board components, no memory(aside from cache), no cooler(in the case of K series pricessor), no board partner cut, additional milking on motherboard chipsets. We used to be able to get 300mm2 die for this price with lynnfield and this was pure CPU. Not 50mm2 of CPU and 64mm2 of integrated GPU.

There was a tonne of profit in processors because Intel used all the nodal improvements on die shrinkage which vastly increase yields per wafer.

Intel margins would be even more insane if they they did spend so much money on contra revenue trying to get into market where they take billion dollar losses. Arm mobility market, Mobility modem market(Intel stock jumped big yesterday after they announced they were leaving this market) are markets where Intel has spent billions, perhaps even tens of billion when you include the R and D which lower their margins.

The videocard market is different because your selling a relatively big die, where board partners in the MSRP still have to pay rest of the board components and generate a margin for themselves as well to be able to pay for RMA's and support which Videocards need more of.

There's a reason why Ryzen 8 core can drop from 499 and a better version only a year later drop to $299 without a sweat with the old 1800x selling for $199.

At it's original price of $499, the original price of ryzen 1800x still had a tonne of profit because it was only a 198mm2 die.

Videocards are different. AMD is continuous struggling to selling Polaris under 200 by themselves and they don't want to sell it below 150 because they start making no money. That is part of the reason they rebrand. Get the old SKU off the market and reset pricing back to their original levels. Right now they are forced to sell certain polaris cheap because of the over supply of Polaris chips and even now, they have been trying to resist prices changes by keeping retail RX pricing high by not doing any official price drops. This is because your getting alot of hardware, silicon, components with videocards.

In other words, they can't afford to undercut Nvidia by huge amounts because the margins are not there.

We cringe at prices of RTX 2080 ti and RTX titans, but AMD sells chips with such margins already and Intel does as well.

You don't think a $1000 dollar 2950x which has 2x 198mm2 dies does not have a crazy margin or a 32 core 2990x which consist of the above x2 doesn't have a crazy margin.

Chips like the RTX 2080ti and 2080 consist of huge monolithic chips, particularly the former which is a single chip at 751mm2. This die likely has terrible yields which is why even with it's high price, it is the only chip that has not been below MSRP even during sales because the quantity available is the low which is no surprise.

In comparison, the 9980xe which is a 18 core chip only has a die around 400mm2 but again a 2000+ price point.

If AMD has the performance, they will charge us more even for a 200mm2 die. Polaris pricing initially was the result of being 50% slower than a gtx 1070 which was supposedly priced at $399. When AMD had an upper hand even for a moment in recent history, they were selling similar die(7870) for 350 dollars which was only 213mm2, an even bigger rip off in pricing than we have today since it price to performance just matched the gtx 570 initially(techpowerup).

AMD will price Navi as high as it can initially because their products have long shelf life's and will need to start high to make subsequent rebrands look like better products than they are when the rebrands sell for lower prices. AMD is no savior when it comes to videocard pricing as they arguably accelerated/started it with 7970 and 7870 initial pricing which allowed Nvidia to increase the price of GK104 which had a cascade effect on all videocard pricing afterwards.

Forgive me, as a consumer, for not caring about Nvidia's or any company ever increasing margins, and simply wanting more for MY dollar.

The only goal for companies, outside of the minority of owner/CEOs who truly have a calling, is to transfer what is in your pocket to their pocket. The best we can hope for is getting the most value in the transfer.

 

[tajoh111]

Forgive me, as a consumer, for not caring about Nvidia's or any company ever increasing margins, and simply wanting more for MY dollar.

The only goal for companies, outside of the minority of owner/CEOs who truly have a calling, is to transfer what is in your pocket to their pocket. The best we can hope for is getting the most value in the transfer.

This is not so much about caring about margins but having realistic expectation on pricing.

I.e RTX 2080 performance at 250 dollars. That is a very poor margin product considering lack of competition. a 220mm2 die on 7nm manufacturing process + 8gb of relatively expensive DDR6 = worse margin than polaris despite the lack of no competition. Not smart business as AMD will sell out, run into supply issues, retailers buy and scalp card until supply meet demand at elevated pricing with AMD absorbing none of that profit and retailers absorbing it all. This on top of making a bad margin at 250 dollars. At 240 dollars, AMD admitted Polaris was on the lower end of margins for Polaris 10 initially. Inflation alone makes 250 dollar nearly equivalent to 240 3 years ago.

What this rumor is more likely is to raise expectations of Navi in terms of price to performance to stall RTX sales(a pattern that has been repeated like crazy since Fiji). Then release to disappointment (near the bottom of peoples expectations for the product) while drum up the war drums why we should buy AMD and not Nvidia even when reviews show the AMD being a worse card than the Nvidia equivalent, along with talks about undervolting, drivers improving performance or anything that can make up for the short comings of said AMD product.

 

[maddie]

This is not so much about caring about margins but having realistic expectation on pricing.

I.e RTX 2080 performance at 250 dollars. That is a very poor margin product considering lack of competition. a 220mm2 die on 7nm manufacturing process + 8gb of relatively expensive DDR6 = worse margin than polaris despite the lack of no competition. Not smart business as AMD will sell out, run into supply issues, retailers buy and scalp card until supply meet demand at elevated pricing with AMD absorbing none of that profit and retailers absorbing it all. This on top of making a bad margin at 250 dollars. At 240 dollars, AMD admitted Polaris was on the lower end of margins for Polaris 10 initially. Inflation alone makes 250 dollar nearly equivalent to 240 3 years ago.

What this rumor is more likely is to raise expectations of Navi in terms of price to performance to stall RTX sales(a pattern that has been repeated like crazy since Fiji). Then release to disappointment (near the bottom of peoples expectations for the product) while drum up the war drums why we should buy AMD and not Nvidia even when reviews show the AMD being a worse card than the Nvidia equivalent, along with talks about undervolting, drivers improving performance or anything that can make up for the short comings of said AMD product.

Unless you can give me an accurate breakdown on the real costs, then everything is narrative.

Yes, a bigger die, on the same process and the same volume by the same company at the same fab will cost more.
Yes, more memory on the card, in the same order volume, at the same specs, by the same supplier for the same manufacturer will cost more.

There are too many unknowns and most, if not all here, can't state a truly accurate value for the production and assembly costs of video cards. What I do know is that at least for Nvidia, desktop sales have been the major revenue earner for the company and their margins have increased a lot. No body is saying that prices have to be halved, but to argue that there isn't room for cuts is not very convincing.

By the way it's your money, spend away.

 

[sandorski]

If these rumours are accurate, I just might buy the Top Line card for the first time ever. Most likely get the one below it though. That said, I'll wait for the reviews.

 

[Head1985]

If these rumours are accurate, I just might buy the Top Line card for the first time ever. Most likely get the one below it though. That said, I'll wait for the reviews.

Its same 6months old rumor.But he add SP numbers to that.
navi10-3584SP
navi12-2560SP
How big will be polaris with 3584SP and 64rops at 7nm?

btw guys did you notice this?

i have new info and this video is last good news on navi

So probably price/perf will be same as turing

 

[DrMrLordX]

navi10-3584SP
navi12-2560SP

I thought Navi10 was going to be more Polaris-like with 2300 or less. 3584? Seems unlikely.

 

[ozzy702]

I thought Navi10 was going to be more Polaris-like with 2300 or less. 3584? Seems unlikely.

I'll believe it when I see it and take anything that Adored releases with a grain of salt.

 

[NostaSeronx]

I thought Navi10 was going to be more Polaris-like with 2300 or less. 3584? Seems unlikely.

I agree, AMD has burnt their very big die budget with Radeon 7. However, AMD also burnt out a successor to big Polaris with Polaris30 on 12LP. The only option available is an actual retail version of Vega-M of Intel and Vega 12 of Apple. Which would be an adequate successor to Baffins which didn't move to 12LP. While also being small enough to achieve high volume on the 7nm nodes.

 

[Bouowmx]

Polaris largest: 2304
Polaris smaller: 1024

Navi largest (?): 3584
Navi smaller: 2560

Doesn't seem like the thing AMD would do in the 7-nm era: design different dies with close capabilities.

 

[naukkis]

Doesn't seem like the thing AMD would do in the 7-nm era: design different dies with close capabilities.

At 7nm smallest possible chip build for 128bit memory interface won't be much smaller than 2560sp's.

 

[blckgrffn]

The idea that Navi won't achieve 2080 levels of Performance strikes me as being silly. Nvidia and AMD knew the approximate levels of Performance would exist at this time years ago when Navi's engineering began. Unless AMD completely cucked it up, their targets would have been at least that level of Performance. As someone else mentioned, AMD already reached that neighbourhood with Radeon VII on the generation prior to Navi. Not equaling the Radeon VII with Navi would be an epic fail. Could happen, I suppose, but I seriously doubt it.

That said, I don't know if AMD was expecting Ray Tracing to exist at this time, for Gaming anyways. If that's the case, I wouldn't be surprised if AMD can't match the 2080 for that task.

I completely agree on both points. So much of the 2080 transistor budget seems to be dedicated to a niche (small number of titles) feature, if it can be cheaper and do everything but ray trace as well (within some percentage error) then I'll see it as a competitive product and probably buy one. It's about damn time.

 

[darkswordsman17]

I completely agree on both points. So much of the 2080 transistor budget seems to be dedicated to a niche (small number of titles) feature, if it can be cheaper and do everything but ray trace as well (within some percentage error) then I'll see it as a competitive product and probably buy one. It's about damn time.

Some things to keep in mind though, we have no idea what the design goals of Navi were. At one point their roadmap listed Navi as mGPU and the rumors suggested that was full on GPU chiplet where it would be as transparent as possible to the software that it was two separate GPUs. So if it was designed for chiplets, it might mean they intended it to be smaller die size, with them scaling it up by adding more chiplets. I think that changed though, and probably more recently than people would expect. I think that's also why there's rumors saying Navi is not actually a new architecture (it was slated to be if I remember right), although I think that also has a lot to do with Sony being the lead target customer, as Sony likely preferred easier compatibility between PS5 and PS4, and a new GPU architecture might change that with not enough tangible benefits for the change. Meanwhile, I think there was enough areas that they could improve GCN that they could get significant gains while they give more lead time to an architecture change (which I think is smart now, as they've been evaluating mGPU/chiplet stuff, and now ray-tracing, and seems like they're re-evaluating their approach to compute in the GPU with Lisa Su saying that diverging GPU for graphics and compute "must be so" and so I kinda think their approach might be to do separate compute and GPU chiplets, with some base compute in the GPU - for stuff where it suits the GPU architecture well)

But in a similar vein, it comes down to intended markets and potentially die size. And if AMD somehow found a way to better utilize the hardware they had (like through improved software utilization, stuff like the NGG fast path that would help discard a lot of unseen geometry improving geometry throughput - something that Nvidia likely is already doing), they could go with a smaller die than they might otherwise have been able to, but that could restrict them from going after the over $500 Nvidia stuff. But AMD has not tended to do well in that market for years now, so it very probably was simply a business decision (hence why they've made it clear that Navi is a Polaris market level GPU). That's why I think Navi could compete with 2070, but I'm doubtful it'll hit well above that. Would love to be surprised though. But it doesn't need to in order to be a great product and help AMD tremendously.

Certainly RTX will eat up some of the die space of Turing that normally would have gone towards traditional raster and compute bits. But it did bring some improvements otherwise, and Nvidia already had an advantage in perf/W and perf/transistor. Which AMD needed 7nm to offer parity to Turing with Radeon VII (both had a lot of stuff that isn't really that gaming focused, especially traditional rasterization, so the argument that Nvidia was handicapped by that stuff ignores that Vega 20 has a lot that doesn't benefit their gaming performance). We'll see if Navi can improve on that. I think it will, and I think most of it will be due to big improvements to software utilization (which is where I think Nvidia actually has the biggest lead).

 

[tajoh111]

According to his video, AMD will make a separate part for 56 and 64 CU. Ridiculous and a waste of money and engineering. So that a big strike to the believably of his speculation.

On top of this, there will be a 30 percent difference in performance between said two parts, which means all the performance has to come from clocks since 56 and 64 Cu's is only 12.5% difference and if we look at vega 56/64 at the same clocks, there is only a 4% difference in performance(gamersnexus). This does not make sense if we look at a couple things we can analyze today.

Vega 56 is the more efficient part between vega 56 and 64 and there is only a 15% difference in performance.

Most of this efficiency is coming from reduced clock speed. However looking at the gap in performance, most of it has to be coming from clocks and as we have seen with GCN, increasing clocks by 20-25 percent will quickly ruin efficiency.

Somehow this gap in performance is supposed to double on top of a reversal of efficiency where the 64 CU part become more efficient(bigger increase in performance then the increase in power consumption). THis hype train doesn't even have rails based on the contradictions on previous patterns.

As someone said above all three parts have too similar of a configuration to make business sense. Look at the segmentation of AMD's previous products and you will see there is quite the spread in terms of Cu. 64, 36 and 16.

Here it's 64, 56 and 40. This does not make sense. The first two parts are ridiculous and the lowest end part is too large for the integrated market and low end market. 7nm is not a nod you want to be as efficient as possible when it comes to segmentation and not leaving wasted space.

Also look at the cost per wafer. AMD is not going to absorb the cost of the wafer when it is going up 65%. Especially after looking at RX 590 pricing.

These video's are becoming a testament and test on a person level of delusion. At a certain point, you have to to realize you have to use your own brain and not just blindly listen to whatever he says.

 

[AtenRa]

When NAVI will be announced in June, TSMC 7nm will be in full production for 12+ months already.

If NAVI 10 is targeting to replace Vega 10 (Vega 64/56) and compete against RTX2060 (cut down TU106) then a 200-230mm2 chip will be able to reach that goal.

VEGA 10 = 495mm2 - 12.5B Transistors at 14nm GloFo
TU106 = 445mm2 - 10.8B Transistors at 12nm TSMC

NAVI 10 at 230nm produced at 7nm TSMC will have close to 9B transistors. I believe it could easily reach GTX1080/RTX2060 performance at 150W TDP and at $275/300 MSRP will be a nice product for todays standards.

 

[ozzy702]

When NAVI will be announced in June, TSMC 7nm will be in full production for 12+ months already.

If NAVI 10 is targeting to replace Vega 10 (Vega 64/56) and compete against RTX2060 (cut down TU106) then a 200-230mm2 chip will be able to reach that goal.

VEGA 10 = 495mm2 - 12.5B Transistors at 14nm GloFo
TU106 = 445mm2 - 10.8B Transistors at 12nm TSMC

NAVI 10 at 230nm produced at 7nm TSMC will have close to 9B transistors. I believe it could easily reach GTX1080/RTX2060 performance at 150W TDP and at $275/300 MSRP will be a nice product for todays standards.

Possible if NAVI has no RT cores/hardware and AMD makes some serious efficiency improvements over Vega.

 

[AtenRa]

Possible if NAVI has no RT cores/hardware and AMD makes some serious efficiency improvements over Vega.

Im betting AMD will not install dedicated units only for RT, like NV did for Turing.
Also, I believe that NAVI will replace Polaris and not VEGA. Next Gen architecture in 2020-21 will replace Vega 20 for the enterprise segment.

 

[mohit9206]

Am quite excited for Navi based RX 570 successor at sub $200 price point.

 

[AtenRa]

Am quite excited for Navi based RX 570 successor at sub $200 price point.

With current NV prices, I highly doubt the RX 570 successor will be bellow $200.

 

[mohit9206]

With current NV prices, I highly doubt the RX 570 successor will be bellow $200.

If not then 1660 non ti is already a good product. If amd doesn't deliver then they will lose more sales.

 

[AtenRa]

If not then 1660 non ti is already a good product. If amd doesn't deliver then they will lose more sales.

If you mean a Navi competitor to the GTX1660 then yes we may see this at $199, but im expecting the RX 570 successor to be much faster than GTX 1660.

 

[Panino Manino]

Some time ago I heard comments about"VLIW2".
What it is and what it means? For what I believe I know this type of GPU is better for graphs than it is for compute like CCN supposedly is. But with all the talks about compute this and compute that in engine now seems like a bad move?

 

[maddie]

Some time ago I heard comments about"VLIW2".
What it is and what it means? For what I believe I know this type of GPU is better for graphs than it is for compute like CCN supposedly is. But with all the talks about compute this and compute that in engine now seems like a bad move?

AMD themselves said that GPUs are splitting into compute and gaming pathways.

 

[psolord]

AMD themselves said that GPUs are splitting into compute and gaming pathways.

I'm no expert, but something in my gut says that this is the correct way to go. I mean sure there will be extra cost for engineers that will have to follow these two paths, but I believe that the die size savings, power savings, the potential manufacturing cost savings and above all, the public's potential buying preference, will give AMD the returns it has planned.

 

[maddie]

I'm no expert, but something in my gut says that this is the correct way to go. I mean sure there will be extra cost for engineers that will have to follow these two paths, but I believe that the die size savings, power savings, the potential manufacturing cost savings and above all, the public's potential buying preference, will give AMD the returns it has planned.

Yep, abandoning the "Jack of all trades, master of none" approach is the only path to remain relevant in those markets.

 

[jpiniero]

Yep, abandoning the "Jack of all trades, master of none" approach is the only path to remain relevant in those markets.

At the same time the design costs at 5 nm are so insane, Navi better deliver or else it's going to be hard to justify continuing with a gaming only design. It'd just be easy to abandon and focus on compute only (and of course Zen and friends).