AMD Navi 23 ‘NVIDIA Killer’ GPU Rumored to Support Hardware Ray Tracing, Coming Next Year

[melmac]

Where the #### are you getting 80% from? on 7nm the 2080TI is still way over 500mm^2 large, to get 80% more performance out of it you would have to double the shader count, at least, Nvidia are known for making super large GPU's but good grief man how big do you think this thing is going to be? Not even Nvidia are going to make GPU's so huge they can only get 15 out of a 300mm $12,000 wafer...

Be realistic.

I am not saying it's going to be 80% faster than the 2080Ti. You are.

That doesn't seem like a huge leap... but i would like to see Nvidia go 180% or more of the 2080TI to make it a meaningful difference from Big Navi.

I asked you the following question then, because I misread your post thinking that you meant Big Navi been 80% faster than the 2080Ti.

You want AMD to produce a GPU that's 80% faster than the 2080Ti? You really think that's possible?

And Lephuronn replied with:

Why do you think it's not possible?

And I answered why I didn't think it was possible, so direct your post at Lephuronn, not me.

Secondly, your calculations are still wrong. The 2080Ti is not way over 500mm2 on 7nm.

 

[humbug]

Yes Navi 23 we already know is 505mm^2, Navi 10 (5700XT) is 251mm^2.

The 2080TI uses Samsung 12nm?

I couldn't find it, i think that was a node specifically for Nvidia.

What i did find was Samsung 10nm at 58.1 density
TSMC 7nm HPC at 66.7 density.

TSMC 7nm HPC is about 15% more dense than Samsung's 10nm. 0.85.

 

[melmac]

So where are the 56 and 64 CU models? Why was Vega allowed to (potentially) undermine 5700 series sales for so long? Why was Radeon 7 EOLed without a replacement, completely removing AMD's presence at the top end? Why did Navi 14 take so long to show up? What about the full SKU list that was leaked that just never materialised? Why the needless name change to 5700?

You seem to not understand the concept of a theory. If I had empirical evidence to support my theory it wouldn't be a damn theory, now would it. So as is the nature of discussion of rumour, I present arguments to support a theory of which I've never claimed to be fact.


So cite your sources and provide evidence to support your theory. You can't harp on at me for baseless speculation yet do exactly the same thing.

How would it be a waste of money developing a high-end card? AMD were perfectly content to lose money repurposing Instinct MI50 dies for Radeon 7 so they could claim the victory of the first 7nm gaming GPU, and offer a competitor to Nvidia's top-end for the first time in many years. A PR stunt? Yes. Boost mindshare? Of course. But suddenly there was no intention for Navi to punch at the high end if it could?

Cite my sources? Look around you, there is no Navi first generation high end cards. Navi 10 is moving down a tier according to the latest news posted a few pages back and Navi 2x card are going to be filling in the tiers above that.

So now your turn, you keep talking about a range of of cards 56 cu and 64 cu models that were been made and were cancelled. But, there is no evidence of AMD cancelling cards, it would have been big news.

Why Was Vega allowed to potentially undermine sales for so long? Because they were in stock, that's why. They were EOL, discontinued, even the the Radeon 7 was discontinued. It happens time and time again in the GPU world. Why was the RX 480 allowed to disrupt sales of the RX 580 for so long. Why was the 1080Ti allowed to disrupt the sales of the 2080 for so long. Cards in stock continue to be sold, usually with deals to move them on.

You think repurposing a card is more expensive than designing a card from ground up? And you answered your own question, they did it purely to say that they had the first 7nm gaming GPU. It was a very limited run of cards, just a publicity stunt.

You keep acting like AMD haven't done this before. Release a range of GPUs aimed at a section of the market and release another range of GPUs over a year later at the higher end of the market.

Why the name change? You don't know what happened at AMD, you don't know what that E3 Render meant. Nobody does apart from AMD.

 

[melmac]

Yes Navi 23 we already know is 505mm^2, Navi 10 (5700XT) is 251mm^2.

The 2080TI uses Samsung 12nm?

I couldn't find it, i think that was a node specifically for Nvidia.

What i did find was Samsung 10nm at 58.1 density
TSMC 7nm HPC at 66.7 density.

TSMC 7nm HPC is about 15% more dense than Samsung's 10nm. 0.85.

Ok thanks, I didn't realise that was fact, just rumour.

And you will find that the 2080Ti would be well under 500mm2 if it was on the same 7nm process as Ampere.

 

[humbug]

Ok thanks, I didn't realise that was fact, just rumour.

And you will find that the 2080Ti would be well under 500mm2 if it was on the same 7nm process as Ampere.

Can't see it but i'll agree to disagree before i start going round in circles with it.

What i will agree to agree is Radeon VII was a very short run, a fat bloated workstation card quickly rebranded to put "something, anything...." on the market to draw some attention away from Turing, yes a marketing stunt. It was quickly replaced by the much smaller more efficient 5700XT and yes they are near identical in performance.

 

[LePhuronn]

You know what? I just spend about 15 minutes typing a reply but I'm just not going to bother. I will say though that if you're so eager to have evidence to support every aspect of discussion, I strongly recommend you just leave this thread because there is no evidence to support anything posted here. it is, after all, a rumour thread.

Oh and by the way, the absence of larger RDNA 1 cards on the market does not confirm your theory about zero intention any more than it confirms my theory of limitations.

 

[bru]

Yes Navi 23 we already know is 505mm^2, Navi 10 (5700XT) is 251mm^2.

The 2080TI uses Samsung 12nm?

I couldn't find it, i think that was a node specifically for Nvidia.

What i did find was Samsung 10nm at 58.1 density
TSMC 7nm HPC at 66.7 density.

TSMC 7nm HPC is about 15% more dense than Samsung's 10nm. 0.85.

The 2080ti is on tsmc 12nm according to the tech power up gpu database.

 

[humbug]

The 2080ti is on tsmc 12nm according to the tech power up gpu database.

Ah... well then to be fair TSMC 12nm is 33.8, this vs 66.7 on TSMC 7nm HPC, which equates to about 0.5.
Thank's

 

[gpuerrilla]

Yeah, I experienced this when I bought my current car. My father-in-law at the time thought he knew everything because he had been driving for the last 32 years. According to him, I didn't have to do any research at all, just buy what he would tell me to buy. I ended up with a car he hated , not because he hated it( that was just a solid bonus) but because the pros outweighed the cons after a lot of research. Silly silly man.

I know, its really sad, but you get older and realise the makes/models go through cycles and you just have to buy the right time and you will still get a good product - rarely are any perfect especially with build quality these days.

 

[EastCoastHandle]

Well, there is your 5nm mobile Lisa Su was talking about. It won't be long now.
Question is what's next CPU, APU, GPU? It might be CPU 1st followed by APU and GPUs. We will see.

 

[Gerard]

Wonder if HBM will ever make a comeback for amd gpu's. It never seemed to really get much traction because of cost and yields were apparently very low for fury\fury x and somewhat better for Vega. For the amount of work they put into it they didn't take too long ditching it for navi.

 

[Rroff]

Wonder if HBM will ever make a comeback for amd gpu's. It never seemed to really get much traction because of cost and yields were apparently very low for fury\fury x and somewhat better for Vega. For the amount of work they put into it they didn't take too long ditching it for navi.

Despite the hype it just isn't well suited to the needs of game data versus stuff like general compute. Possible a future evolution of it might change that but despite the claims otherwise by some GDDR has managed to keep up with gaming requirements.

 

[Gerard]

Despite the hype it just isn't well suited to the needs of game data versus stuff like general compute. Possible a future evolution of it might change that but despite the claims otherwise by some GDDR has managed to keep up with gaming requirements.

HBCC always seemed interesting to me, though i'm unsure if it ever really got use in the vega cards. Interesting concept though if it worked just like the cards vram and was indistinguishable in gameplay.

 

[EastCoastHandle]

 

[humbug]

Well, there is your 5nm mobile Lisa Su was talking about. It won't be long now.
Question is what's next CPU, APU, GPU? It might be CPU 1st followed by APU and GPUs. We will see.

Wait... this is an ARM based AMD APU, this is a phone chip and i don't think its the one that Samsung are using.

 

[EastCoastHandle]

Wait... this is an ARM based AMD APU, this is a phone chip and i don't think its the one that Samsung are using.

Yeah like I said Lisa considers certain APU based skus as mobile. And, this is at 5nm.
Still awaiting a formal announcement from AMD though.
Be that as it may, it's not clear the number of projects AMD has going on at 5nm. This is just one I found so far.

 

[humbug]

Yeah like I said Lisa considers APU based skus as mobile.

Right but this is not for Laptops

 

[EastCoastHandle]

Right but this is not for Laptops

Oh no, this is for cell phones. I see what you are saying now (SoC)
AMD doesn't have a "modem" for cell phones yet. That's expected to change at some point though. Not sure how much that matters here though.

 

[humbug]

I think its interesting they are branching out into their own ARM designs.

 

[EastCoastHandle]

I think its interesting they are branching out into their own ARM designs.

Well, they have to ARM is picking up traction. AMD must content with it before it's completely saturated in the market.