AMD Readies Larger 7nm "Navi 12" Silicon to Power Radeon RX 5800 Series?

[Vayra86]

Well every tech reviewer on Youtube has one for testing. The 2080TI holds the 9 and 10 position on Amazon's top 10. It even outsells the 2080 according to that

Still doesn't say a single thing. Amazon's top 10 of what, sold GPUs? And what % does the spot 9/10 hold? Its less than what the first 8 spots hold...

Here's a more realistic take on market shares over time

UserBenchmark: Nvidia RTX 2080-Ti

 

[bug]

Has AMD said which nVidia card this one will compete with for performance?

They haven't, but I don't see Navi scaling much past 2080.

When you have two comparable alternatives, like RTX 2060 Super vs. RX 5700 XT, the majority of buyers will go for the one with the overall best deal. RTX 2060 Super costs the same, performs practically the same (especially if we consider partner cards), is more power efficient and much quieter, and the list goes on. For AMD to obtain market share, they need to offset their disadvantages with some equally attractive advantages; like significantly lower prices, 10-15% higher performance, or some useful "must-have" feature, etc. Otherwise RTX 2060 Super vs. RX 5700 XT will end up like GTX 1060 vs. RX 480/580 again; 9 out of 10 will go for the Nvidia option, not because they are fanboys, not because AMD's offer is terrible either, but simply because there is an obvious better offer.

Actually, 5700 beats the 2060 Super in perf/W (by a few percent, but still, it's in a place AMD hasn't been in years). But it doesn't do RTRT (arguably, neither the 2060 Super), DLSS or variable shading. It's still a coin toss in the end.

 

[eidairaman1]

Titan RTX has 24GB on 384bit, so 16GB on 256 should be possible. Someone correct me if I'm wrong.

Hexadecimal is key

 

[Prima.Vera]

Are you asking us?

 

[Divide Overflow]

Wake me when Navi 20 is ready for review.

 

[TheinsanegamerN]

Historically, AMD's biggest dies have had codenames in multiples of 10; Polaris 10, 20, 30, Vega 10, 20 are all the largest dies of those particular architectures/generations.

Polaris 11, 12, and 21 were smaller dies, cut-down from 2304SP of the full Polaris chip to 1024 and 640SP

I don't think Navi 12 will be a larger chip - The halo market is vanishingly small and expensive to operate in. AMD are far more likely to be chasing the mass-market where 90% of sales are with a lower-TDP, cheaper-to-make chip that serves the $200 and lower market, as well as being laptop-friendly. I'm only guessing, but I think that *if* AMD make a larger Navi die, it'll be called Navi 20, or given an new codename altogether.

Small GPUs in laptops are dead, the iGPU killed them. The only laptop GPUs that sell well are pro GPUs or high end gaming GPUs, both of which are dominated by nvidia. AMD needs a big laptop chip, not tiny ones.

The Halo market, OTOH, is one of only growing sectors of the PC market, along with gaming in general. Look at the cash Nvidia rakes in with gaming GPUs, and tell me that high end gaming chips are not printing money. Sure, they have lots of mid range GPUs sold, but their high end also sells respectably well, and prints cash. Now sure, its expensive to operate in, but any market with high margins will be. You have high risk to go with high reward.

Still doesn't say a single thing. Amazon's top 10 of what, sold GPUs? And what % does the spot 9/10 hold? Its less than what the first 8 spots hold...

Here's a more realistic take on market shares over time

View attachment 127975

UserBenchmark: Nvidia RTX 2080-Ti

Also doesnt say a single thing. All you are proving is that 336,466 users of UserBenchmark have 2080tis.

If you are insinuating that 1.8% isnt a good number of users, then by that measurement every single GPU AMD makes has a tiny number of users compared to Nvidia, as far as steam is concerned.

We know the 2080ti isnt the volume seller, but to say it isnt selling well and making money is just preposterous.

 

[ValenOne]

AMD needs to scale NAVI 10's dual shading engines (four prim units, 40 CU, 64 ROPS, 4MB L2 cache) into

Three shading engines (six prim units, 60 CU, 96 ROPS, 6MB L2 cache) with 384 bit bus

Four shading engines (eight prim units, 80 CU, 128 ROPS, 8MB L2 cache) with 512 bit bus or four stack HBM v2

 

[Manu_PT]

Why Manu, why are you selectively scared ?

You just made my point even easier to understand. As you can see the only card wich it can compete against, uses 226w, way far from 300w. Thanks for the graph!!

Fanboys are really funny.
Since when do you care for a 50w tdp difference when there is (maybe??) performance to back that up?
Let's all just buy 1030's then

Why would I get a RTX 2080 similar card, using 80w more? Makes no sense to me, unless you tell me it is way cheaper. Wich we all know it won´t be, considering their current 5700 line up prices.

 

[ToxicTaZ]

Once again AMD playing catch up.

Navi 10 is faster than TU106 Bravo!

TU104 is faster than Navi 10 (2070 Super)

Navi 12 would barely match unlocked TU104 (2080 Super)

As for TU102 AMD has Navi 20 November or later .... Nvidia probably release an unlocked TU102 (2080Ti Super)

Super is Superior to Navi
End of Story

 

[ValenOne]

He/she have a reason to worry.
And when it comes to power consumption, the main concern for heat, noise, throttling etc. is average consumption, not peak consumption.

RX 5700 XT already consumes ~225W for 40 CUs/2560 SPs, that's going to increase a lot if AMD wants larger dies, or they have to sacrifice a lot of clock speed.

That's AMD overclocking RX-5700 XT.

Far Cry 5 with RX 5700 consumes about 150 watts, hence AMD needs to scale from RX-5700 e.g.

RX-5700's dual shader engines (four prim, 36 CU, 64 ROPS, 4MB L2 cache) with 256 bit bus scales by 2, hence quad shader engines (eight prim, 72 CU, 128 ROPS, 8 MB L2 cache) with 512 bit bus config

RX-5700's dual shader engines (four prim, 36 CU, 64 ROPS, 4MB L2 cache) with 256 bit bus could scale into three shader engines (eight six, 72 CU, 96 ROPS, 6 MB L2 cache) with 384 bit bus config.

Scaling just CU count is not enough.

R9-290X's quad shader engines (four prim, 44 CU, 64 ROPS, 1MB L2 cache) with 512 bit bus is 2X scale Radeon HD 7870's dual shader engines (two prim, 20 CU, 64 ROPS, 512 KB L2 cache) with 256 bit bus

 

[eidairaman1]

Once again AMD playing catch up.

Navi 10 is faster than TU106 Bravo!

TU104 is faster than Navi 10 (2070 Super)

Navi 12 would barely match unlocked TU104 (2080 Super)

As for TU102 AMD has Navi 20 November or later .... Nvidia probably release an unlocked TU102 (2080Ti Super)

Super is Superior to Navi
End of Story

Not really, nice try.

 

[ratirt]

As I see it, AMD has a chance to get closer to 2080 super or maybe even TI. There's still headroom with Navi and the RDNA is not that bad really. We will have to wait and see but I do think AMD can pull this off. NV already released "Super" versions of the cards and I think that is the ace. Not sure if NV has anything more to add to this. Releasing cards within a half a year to jeopardize they previous cards isn't good for marketing and consumers get pissed off. I'd be pissed of if I buy 2080 or 2070 super and then it turns out, after half a year or less, that there are better cards with lower price.
Maybe the trick is to wait for AMD and see what will happen.

 

[nguyen]

As I see it, AMD has a chance to get closer to 2080 super or maybe even TI. There's still headroom with Navi and the RDNA is not that bad really. We will have to wait and see but I do think AMD can pull this off. NV already released "Super" versions of the cards and I think that is the ace. Not sure if NV has anything more to add to this. Releasing cards within a half a year to jeopardize they previous cards isn't good for marketing and consumers get pissed off. I'd be pissed of if I buy 2080 or 2070 super and then it turns out, after half a year or less, that there are better cards with lower price.
Maybe the trick is to wait for AMD and see what will happen.

GTX 1080 FE were selling for 700usd and eventually reduced to 500usd when 1080 Ti launched 9 months later, RTX line up launched 10 months ago. Hardware prices should naturally be falling, there is no point getting mad over something like that. Let just hope this time that the mining era is gone for good.

 

[ratirt]

GTX 1080 FE were selling for 700usd and eventually reduced to 500usd when 1080 Ti launched 9 months later, RTX line up launched 10 months ago. Hardware prices should naturally be falling, there is no point getting mad over something like that. Let just hope this time that the mining era is gone for good.

You can say I'm all but mad. it is not about getting mad but I dislike the fact that you buy crippled GPU and few months later you get a refresh for the same or lower price. I know it is about marketing and getting more cash from customers but it doesn't mean I need to be glad about it and take it as it is. Does it?

 

[ppn]

Better yet GTX 280 were selling for $650 and eventually reduced to $330 when GTX 285 launched 6 months later. HD 4870 was the spritual XT 5700 that made this possible. Now Nvidia will follow with a 12nm to 7nm shrink or just rename 2080 to 3070 with another $200 drop that remains to be seen. It is far from over, then we have 6nm and 5nm for sure. NAVI 12 is 170 sq.mm so no idea how ppl think it is 2080 territory.

 

[bug]

You can say I'm all but mad. it is not about getting mad but I dislike the fact that you buy crippled GPU and few months later you get a refresh for the same or lower price. I know it is about marketing and getting more cash from customers but it doesn't mean I need to be glad about it and take it as it is. Does it?

Do you understand yields? Would you feel better is, one yields went up, Nvidia kept selling the old GPUs?
Better products come out all the time, why rage about it now? Should AMD not release a 5800 in the future because that will screw 5700 buyers?

 

[kings]

Releasing cards within a half a year to jeopardize they previous cards isn't good for marketing and consumers get pissed off. I'd be pissed of if I buy 2080 or 2070 super and then it turns out, after half a year or less, that there are better cards with lower price.

Like AMD did with Radeon VII in just 5 months?

Things evolve, who doesn't want to be "outdated", better not buy anything!

 

[ShurikN]

NAVI 12 is 170 sq.mm so no idea how ppl think it is 2080 territory.

Navi 14 is the smaller one. Navi 12, like the tweet says, is a big die (in relation to Navi 10)

 

[ratirt]

Do you understand yields? Would you feel better is, one yields went up, Nvidia kept selling the old GPUs?
Better products come out all the time, why rage about it now? Should AMD not release a 5800 in the future because that will screw 5700 buyers?

What yields have to do with it? I'm talking about release a new GPU but Super isn't anything new. It is same as it was just a bump in cores. I'm not raging I dislike the idea cause it would seem they could have released it earlier. The difference is that AMD release only 2 of Navi 5700 and XT for one segment 5800XT is for higher end product. Don't you see the difference here? they didn't change the entire line-up.

Like AMD did with Radeon VII in just 5 months?

Things evolve, who doesn't want to be "outdated", better not buy anything!

True. Can you tell the difference between 5700 XT and RVii? Can you tell the difference between 2080 and 2080 super? Please share I'd like to read it.
And let us compare.

 

[bug]

What yields have to do with it? I'm talking about release a new GPU but Super isn't anything new. It is same as it was just a bump in cores. I'm not raging I dislike the idea cause it would seem they could have released it earlier. The difference is that AMD release only 2 of Navi 5700 and XT for one segment 5800XT is for higher end product. Don't you see the difference here? they didn't change the entire line-up.

Everything.
Turing is a big die. In the beginning, inevitably it has more defects and you have to disable some CUs. As manufacturing matures, the number of defects goes down and you don't have to disable as many CUs to get the same number of working dies from a waffer. So what do you do? Do you keep disabling the same number of CUs just because or do you disable just the number you need and sell a better product for the same $$$?

 

[ratirt]

Everything.
Turing is a big die. In the beginning, inevitably it has more defects and you have to disable some CUs. As manufacturing matures, the number of defects goes down and you don't have to disable as many CUs to get the same number of working dies from a waffer. So what do you do? Do you keep disabling the same number of CUs just because or do you disable just the number you need and sell a better product for the same $$$?

Sure that is one side of the medal. As you mentioned, the dies are huge, I dont think yields have to do something with it considering how big 2080 super is the yields aren't great and so 2070 are not awesome either. Defects can occur on the wafer randomly and you can't and will never know where it will happen. Which part will be affected. So what is going to happen with the dies that are below 2060 super die size or better 2070 super? (keep in mind it is a different die than 2060 super since it is just like 2080 and 2080 super). So what if it is not 2080 nor 2070 super? It can't be 2070 cause these are no longer in production and use different chip. The range of defects in such a big die like TU104 can be of dozens or even more. On top of that there are tensor cores and RT cores which can be with defects as well. how can you match it up to get the current product line?
I don't think yields is much of a reason for this.

 

[bug]

I don't think yields is much of a reason for this.

That is why we disagree.
That and me being unable to follow your wall of text.

 

[ratirt]

That is why we disagree.
That and me being unable to follow your wall of text.

Well, as you can see, and hopefully can admit, it is complicated. I've got still a lot more arguments because it is not as simple as you describe it. Maybe that is why we disagree in the first place.

 

[bug]

Well, as you can see, and hopefully can admit, it is complicated. I've got still a lot more arguments because it is not as simple as you describe it. Maybe that is why we disagree in the first place.

Oh, but it is simple. You just need to complicate it because you're angry the Super cards rained on Navi's parade

 

[ratirt]

Oh, but it is simple. You just need to complicate it because you're angry the Super cards rained on Navi's parade

You are adorable But no and I'm not angry because of Navi And yet you have just said another argument that popped into my head. Maybe the release is not related to the yields by any means which in my opinion is totally wrong but because of Navi. well, at least in part.