AMD: Navi Speculation, Rumours and Discussion [2017-2018]

[milk]

I think in AMDs position, they are doing the right thing strategically. They don't need to support modern DX12 features untill the next gen of consoles launches as that's when games will actually start using these features. Of course that's not ideal, but since they have to be mindful of their use of R&D resources, it's a good tradeoff. DX12 features that don't really get used aren't the most urgent matters now.

 

[Azhrei]

More Navi rumours, this time from AdoredTV. Jim talks a lot about Zen 2 first, but he has some fascinating figures about Navi graphics cards, with three cards rumoured to be announced at CES in January. Two Navi 12 cards, and one Navi 10 card, as follows -

 

[McHuj]

That would put top tier Navi 10 in 1080ti tier?

 

[Azhrei]

That would put top tier Navi 10 in 1080ti tier?

Top-tier Navi is meant to be Navi 20, coming either at the end of 2019 or (more likely, sadly, though who knows) 2020. Navi 10 would be around 1080/TI tier if true though, yes. Exciting stuff!

 

[kalelovil]

If true this would make the RX590 launch seem a waste of time. Who is going to want them after this? Unless it was just needed to fulfill obligations to Global Foundries.

 

[Azhrei]

Who knows when these are going to launch, though? Could be nearer the middle of the year rather than March as Jim proposes. But yes, that is a good point.

 

[sheepdogexpress]

These rumors are difficult to believe and remind me of the first batch of Polaris(faster than a gtx 980 ti), Fiji(gtx 980 ti + 10-15%) and Vega rumors(gtx 1080 ti + 15%).

And I believe they have a similar purpose. That is any rumors at this point help AMD out more than hurt it, even if they are hyped up ones.

That is the more they stall Nvidia cards sales, the more the potential marketshare available when the cards do release. They also help prop up the stock by creating hype for both their CPU and GPU release.

The problem with these rumors is that they lead to initial disappointment upon release and eventually people catch on. Reality dictates that the GCN architecture(and Navi will be the last card using it) has had most of its potential squeezed out and any performance increases at this point are mostly clocks and increasing tflops which are decreasing in terms of those flops turning into gaming performance.

With Vega running into a wall in terms of clocks due to power, I can't see Navi solving this issue on 7nm to the extent AMD needs to achieve that level of performance at 150watts. Navi will have a moderate amount of shaders because of the 256 bit bus and a moderate amount of ROPs to achieve mainstream pricing. That would means it would needs high clocks to make up the rest and this is where power becomes an issue. 7nm is not going to be the magic bullet considering where Polaris 30 currently is at and from what we have already seen with Vega 20 at 7nm.

From what we have seen already, Vega 20 7nm has performance approximately at Vega 64 + 15% considering the 1800mhz clocks and it does this at 300watts(and this is the main reason Vega 7nm is not coming to consumers as it performs along the lines of rtx 2070 but at much greater cost to AMD).

Now these rumors point to this level of performance without HBM2 at half the power? AMD viral marketing team can keep on tricking us with hype but they need to deliver at some point. Because past hawaii, the rumor about performance that has been the most true has been typically the most pessimistic and worse case scenario. This includes the r9 285 through Vega. That is, they are worse than the competition with significantly worse power draw. If you look at AdoredTV expectations of performance for AMD cards, the only one that was correct the one where Vega was a disappointment, which should have been obvious considering the silence from AMD officially while Pascal was cranking out record sales. The Shanghai GPU team has just been a disappointment so far compared to the Markham team.

AMD is barely able to catch Maxwell even today in terms of performance per watt.

Relative to the cards in the lineup, even the metrics of those cards don't make much sense at the moment.

The RX 3080 will not have linear pricing with an RX 3070 while having the same memory configuration because the above scenario shows the RX 3080 with 26% better performance for 25% more money vs the RX 3070. This is not how you price products going down the stack as it drives people to buy the supply limited high end SKU while leaving the more plentiful card less desirable and thus stalling sales. This is because demand for the top end outstrips supply while the lower end SKU does not sell because it is a worse value which causes these buyers to wait for the top SKU. This means lost profit at the top product can sell for more money which is shown from demand outstripping supply and secondly lost volume on sales on the lower end product as people wait for the top product.

Secondly, the pricing of these cards doesn't make sense considering what AMD just released the RX 590 at. AMD isn't going to release this card at a lower price than the RX 590 considering 7nm's is an even more expensive node than 14nm while having 8gb of gddr6 vs gddr5 is also an increase in pricing over RX 590.

What the last couple years have shown us is the success of AMD CPU division does not carry over to their GPU division. I see very little reason for this to change because AMD focus is still clearly more on their CPU as their road maps and presentations go and the Shanghai team has not proven itself yet.

 

[DavidGraham]

And what's with those ridiculous names? AMD is going to suddenly leave the 680 naming scheme and jump straight to 3080? further confusing customers when RTX 3080 comes out from NVIDIA?!

That would put top tier Navi 10 in 1080ti tier?

More like 2070 tier, or slightly above it.

 

[Kaotik]

Top-tier Navi is meant to be Navi 20, coming either at the end of 2019 or (more likely, sadly, though who knows) 2020. Navi 10 would be around 1080/TI tier if true though, yes. Exciting stuff!

Seriously? This "2020 Navi" still floats around? AMD has already confirmed next-gen for 2020.

 

[Bondrewd]

Seriously? This "2020 Navi" still floats around? AMD has already confirmed next-gen for 2020.

It can be early 2020 fat Navi and late 2020 HPC next-gen.

 

[3dilettante]

The rumored Navi products in those pictures have a notable amount of stretching of Navi 12 in terms of power and performance. It's stretching the same silicon 60% in power consumption and ~40% in performance, which is a decent gap in SKUs. I suppose the lower SKU would have less wiggle room in terms of typical board power, if there's no standard power connector to take up slack from the board slot. However, this stretch seemingly gives little room for a salvage SKU for the Navi 10 chip, which would find Navi 12 coming up from below--unless the Navi 10 product is partially deactivated.

 

[yuri]

...
With Vega running into a wall in terms of clocks due to power, I can't see Navi solving this issue on 7nm to the extent AMD needs to achieve that level of performance at 150watts. Navi will have a moderate amount of shaders because of the 256 bit bus and a moderate amount of ROPs to achieve mainstream pricing. That would means it would needs high clocks to make up the rest and this is where power becomes an issue. 7nm is not going to be the magic bullet considering where Polaris 30 currently is at and from what we have already seen with Vega 20 at 7nm.

From what we have seen already, Vega 20 7nm has performance approximately at Vega 64 + 15% considering the 1800mhz clocks and it does this at 300watts(and this is the main reason Vega 7nm is not coming to consumers as it performs along the lines of rtx 2070 but at much greater cost to AMD).

...Because past hawaii, the rumor about performance that has been the most true has been typically the most pessimistic and worse case scenario. This includes the r9 285 through Vega. That is, they are worse than the competition with significantly worse power draw. If you look at AdoredTV expectations of performance for AMD cards, the only one that was correct the one where Vega was a disappointment, which should have been obvious considering the silence from AMD officially while Pascal was cranking out record sales. The Shanghai GPU team has just been a disappointment so far compared to the Markham team.

AMD is barely able to catch Maxwell even today in terms of performance per watt.
...

Well written. Since the GFX10 cards have just started appearing in drivers (NAVI 10 and NAVI 10 Lite only), a January/CES launch is simply not happening.

It seems strange AMD would finally find a magic perf/watt sauce for GCN in its last gen. The same applies for >4096SPs rumors...

 

[del42sa]

More Navi rumours, this time from AdoredTV. Jim talks a lot about Zen 2 first, but he has some fascinating figures about Navi graphics cards, with three cards rumoured to be announced at CES in January.

I really dont know what to think about him seriously if he believes that. AMD has already proven their complete lack of competence in designing 7nm chips.

Vega10 MI25: 12.228gflops at 300W
Vega20 MI60: 14.746gflops only 20% faster at 300W
hypothetical Vega20 at 2x power efficiency: 16.507gflops at 202.5W

Vega10: 25.667.351 transistors/mm2 at 487mm2 die size
Vega20: 39.969.788 transistors/mm2 at 331mm2 die size => only 54.41% denser ( I realize it isnt simple shrink )
hypothetical Vega20: 51.334.702 transistors/mm2 at 258mm2 die size

his speculations were always based on something more concrete or demos,leaks or whatever.. but this time he pulls 50 mins speculation video based on ZEN2 and PS5 hype rumor.. and he literally ignores the damn Vega 7nm which literally is behemoth of 4096 cores at 1.8ghz...

 

[entity279]

It seems strange AMD would finally find a magic perf/watt sauce for GCN in its last gen.

Well, I'll say we'll only get to see if that is true once another arch is also launched on the same TSMC process, especially if it's from nVidia

 

[Love_In_Rio]

Well, I'll say we'll only get to see if that is true once another arch is also launched on the same TSMC process, especially if it's from nVidia

The suppossed 150 watts for a Navi 10 with Vega 64+15% performance, near the performance of Vega 20 at 7nm and 300 watts, only could be explained if Navi 10 wasnt indeed GCN.

 

[Kaotik]

It can be early 2020 fat Navi and late 2020 HPC next-gen.

Yeah, and pigs fly if you put them in a plane
Seriously though, the whole "Navi 2020" thing is based on what, one unreliable rumorsite claiming such "because Navi is made for PS5 and PS5 is coming in 2020" or something along those lines, which we know to be false to claim begin with.
If there would be new Navi's in 2020, they would be just "refreshes" like Polaris 20/30, if even that.
Also regarding the Adoredthingy, it's extremely unlikely that one Navi chip would cover about RX 580 and Vega 56 performance, that's a 40-50% difference between the two, and that there would be a second chip which would be just 30-35% faster than the other chip.

 

[Bondrewd]

AMD has already proven their complete lack of competence in designing 7nm chips.

*laffs in Rome*

 

[Kaotik]

I really dont know what to think about him seriously if he believes that. AMD has already proven their complete lack of competence in designing 7nm chips.

Vega10 MI25: 12.228gflops at 300W
Vega20 MI60: 14.746gflops only 20% faster at 300W
hypothetical Vega20 at 2x power efficiency: 16.507gflops at 202.5W

Vega10: 25.667.351 transistors/mm2 at 487mm2 die size
Vega20: 39.969.788 transistors/mm2 at 331mm2 die size => only 54.41% denser ( I realize it isnt simple shrink )
hypothetical Vega20: 51.334.702 transistors/mm2 at 258mm2 die size

his speculations were always based on something more concrete or demos,leaks or whatever.. but this time he pulls 50 mins speculation video based on ZEN2 and PS5 hype rumor.. and he literally ignores the damn Vega 7nm which literally is behemoth of 4096 cores at 1.8ghz...

Really? You're basing this on one chip that we don't actually have any real comparison point to. Vega20 has double the memory controllers and who knows what other changes AMD hasn't disclosed so far. Just because the shader counts etc are the same doesn't mean the chip is the same. As for density, everything doesn't scale identically, so before we know all the details on what changed on the inside you can't really make any judgements on how it is. Heck, do we even know how big the theoretical scaling would be from GloFo 14nm to TSMC 7nm?

 

[AlNom]

Heck, do we even know how big the theoretical scaling would be from GloFo 14nm to TSMC 7nm?

Wouldn't it just be roughly similar to TSMC's 16FF to 7 ? TSMC's website implies ~3x density based on 16->10 @ 2x density, 10 ->7 @1.6x density.

 

[Deleted member 13524]

AdoredTV just put out some rumors/leaks about what AMD will put out at Computex (June) 2019:

Three Navi GPUs:

1. RX 3060 75watts performance equal to RX 580 8Gb - target price $130.
   
   
2. RX 3070 120watts performance equal to Vega 56 8GB - target price $200.
   
   
3. RX 3080 150watts performance better by 15% than Vega 64 - target price $250.

Argh, the cringe on these names..