Question Speculation: RDNA2 + CDNA Architectures thread

[uzzi38]

All die sizes are within 5mm^2. The poster here has been right on some things in the past afaik, and to his credit was the first to saying 505mm^2 for Navi21, which other people have backed up. Even still though, take the following with a pich of salt.

Navi21 - 505mm^2

Navi22 - 340mm^2

Navi23 - 240mm^2

Source is the following post: https://www.ptt.cc/bbs/PC_Shopping/M.1588075782.A.C1E.html

 

[Timorous]

Yeah. AMD is not like Nvidia or Intel, so they contrary to Nvidia or Intel do not jack up prices on their CPUs by 50$ using their dominant position, because they have no competition? 😉

Nvidia can compete though so less opportunity to raise prices. Also AMD want more mindshare in the GPU segment so pricing aggressively will help with that.

 

[DisEnchantment]

(lol, VideoCardz seriously has been dropping the ball with its rumors as of late...).

What is even more hilarious is the RT Coprocessor guy with his "die shot" (which has got nothing to do with the die but a pixelated package at best) has confirmed" Navi21 to be 15% 2080Ti perf from "AIB partners"
He "confirmed" PS5/XSX N7+
Dude has got practically zero credo.

 

[DeathReborn]

Its STILL not corrected! Hexus number for the stock 3080 is 77 frames! The 81 frames is overclocked AIB models.....

Moving from 77 fps to 81 fps is actually doubling the difference which seems unfair to say the least.

View attachment 31382

Yet the 6000 series MIGHT be on a faster CPU, comparison is worthless until they are on the same platform.

 

[swilli89]

Yet the 6000 series MIGHT be on a faster CPU, comparison is worthless until they are on the same platform.

I'm simply trying to frame the current conversation in the most accurate light possible. They are cherry picking AIB cards which seems absurd to me when Hexus also publishes the factory 3080 numbers in the same freaking charts.

And at 4k.. are we really sitting here discussing the difference between Intels fastest and AMD's fastest?

 

[Hitman928]

Yet the 6000 series MIGHT be on a faster CPU, comparison is worthless until they are on the same platform.

I highly doubt the CPU makes any difference in avg fps at 4K highest settings as long as you are using any mid to high end CPU from the last 3 years. I agree on waiting for actual reviewer benchmarks though.

 

[insertcarehere]

I'm simply trying to frame the current conversation in the most accurate light possible. They are cherry picking AIB cards which seems absurd to me when Hexus also publishes the factory 3080 numbers in the same freaking charts.

And at 4k.. are we really sitting here discussing the difference between Intels fastest and AMD's fastest?

Except Hexus's test bench is a 3950X and AMD just got done showing off how much faster the Zen 3s (paired with Big Navi) are at gaming than those CPUs, let's wait for independent benches before jumping into any comparisons....

 

[swilli89]

Except Hexus's test bench is a 3950X and AMD just got done showing off how much faster the Zen 3s (paired with Big Navi) are at gaming than those CPUs, let's wait for independent benches before jumping into any comparisons....

And I literally just posted a chart that illustrates that at 4k it really, truly doesn't matter. Of course we still need to wait. But while we are attempting to draw estimates guesses for entertainment's sake, we can ignore the CPU aspect of 4k benchmarks.

 

[Konan]

Notes for the GPU test:

 

[zinfamous]

I'd like to get stats on the resolution percentages. Does anyone know if I can just take the 4k number in pixels, get the fps per pixel and apply that to a 3440*1440 resolution?

that's what I do in my head. I'm sure it isn't exact, but "close enough" without too much thinking and I'm satisfied with convincing myself that 4k FPS are relatively similar to 3440x1440

 

[zinfamous]

Of course they won't show everything, but if they are going to bring a tease, then I am sure they brought out the best numbers they can bring out. I am very sure if they had better than 3080 numbers to show, they would have done it.

This.

3080 is available now 9well, sorta, heh). If they had something better to show that would convince people to wait for something clearly better, they would have shown it.

I believe that their two options were:

--show nothing and wait
--show the best of what you have.

 

[Timorous]

This.

3080 is available now 9well, sorta, heh). If they had something better to show that would convince people to wait for something clearly better, they would have shown it.

I believe that their two options were:

--show nothing and wait
--show the best of what you have.

Nvidia will sell everything no matter what so AMD leaving room for a surprise on oct 28th will not impact Nvidia sales.

 

[Hail The Brain Slug]

My opinion is that the entire point was simply to prove to everyone skeptical they can achieve high performance that NavII will be able to game at 4k60 ultra and nothing more. It's not supposed to showcase their max performance, flagship, or compare to the 3000 series.

 

[kurosaki]

that's what I do in my head. I'm sure it isn't exact, but "close enough" without too much thinking and I'm satisfied with convincing myself that 4k FPS are relatively similar to 3440x1440

That only applies if the die is wider than its long. If you look at red gaming techs latest post of a blurry die, it seems to me as if you would win on turning your screen 90 degrees, or buy a 1440x3440 screen instead.





.



.
.



.

😉

 

[TESKATLIPOKA]

This announcement was for Zen 3 and not for RDNA2, so It's possible they didn't use the strongest configuration based on Navi21. I think It's best to wait at least for the announcement and then third party reviews. If nothing else Zen 3 looks great.

 

[GodisanAtheist]

How do those numbers stack up against the 5700XT?

If AMD is able to double or better yet more than double the 5700XT's performance within the same power envelope on the same process, that is one hell of a feat of engineering.

 

[Vope45]

This.

3080 is available now 9well, sorta, heh). If they had something better to show that would convince people to wait for something clearly better, they would have shown it.

I believe that their two options were:

--show nothing and wait
--show the best of what you have.

But AMD sandbag !1!! Yeah no.

AMD has never sandbaged their products, ever.

 

[Veradun]

Are you guys really comparing benchmarks to slides?

 

[insertcarehere]

And I literally just posted a chart that illustrates that at 4k it really, truly doesn't matter. Of course we still need to wait. But while we are attempting to draw estimates guesses for entertainment's sake, we can ignore the CPU aspect of 4k benchmarks.

TPU's testing indicates otherwise, with a 3080-class GPU they show that even at 4k a quicker CPU (than Zen 2) is still worth a couple of frames in select titles, it's not big by any means but to say a 5900X and 3950X/10900K as identical wouldn't be accurate either.

 

[zinfamous]

Nvidia will sell everything no matter what so AMD leaving room for a surprise on oct 28th will not impact Nvidia sales.

Yes, and this same argument was made with the Vega 56 and 64 teases, RX480, etc. "Obviously, that isn't the FASTEST cards, because that isn't what this week's hype suggested!"

Look, that card that they showed today looked pretty damn good and we still don't know RAM, power, cost, and it probably still isn't final...but I imagine it is pretty close to what it will be.

I'm just going to go with what we actually know in recent years of how AMD actually teased what they had, the community insisted endlessly that there was some as-yet-unrevealed killer piece of hardware that they are just waiting to reveal in some dramatic fashion--lol, why even expect drama like this? It's just a piece of hardware, ffs--only for those to never ever materialize.

The Radeon VII wasn't even in their gameplan except for some limited number of harvested dies that they used as a "proof of 7nm" and early test for efficiency. It certainly wasn't ever planned to be a consumer card, if it was ever planned to be a product, but Nvidia, IIRC, created a little space between the 1080 and 1080ti, that, even with not-very game-friendly drivers, the Radeon VII fit somewhat decently.

 

[Leadbox]

This.

3080 is available now 9well, sorta, heh). If they had something better to show that would convince people to wait for something clearly better, they would have shown it.

I believe that their two options were:

--show nothing and wait
--show the best of what you have.

So what would be the point of the Oct 28 event then? Just to announce skus, pricing and availability?

 

[Konan]

TPU's testing indicates otherwise, with a 3080-class GPU they show that even at 4k a quicker CPU (than Zen 2) is still worth a couple of frames in select titles, it's not big by any means but to say a 5900X and 3950X/10900K as identical wouldn't be accurate either.

For benchmark comparisons, These would be at DX11 and not DX12 like what AMD showed today. (TPU test in DX11) Hence differences in straight comparison.

But good point on making note of CPU differences 🙂

 

[Stuka87]

How do those numbers stack up against the 5700XT?

If AMD is able to double or better yet more than double the 5700XT's performance within the same power envelope on the same process, that is one hell of a feat of engineering.

Roughly double what a 5700 XT will do. But think it most likely is the 72CU SKU, so they have something else to show in 3 weeks.

 

[Vope45]

Roughly double what a 5700 XT will do. But think it most likely is the 72CU SKU, so they have something else to show in 3 weeks.

That's the 80 CU die.

 

[Stuka87]

That's the 80 CU die.

Yes, the die has 80CU, but I am willing to bet its the cut down (72CU) version.

 

[GaiaHunter]

It seems to be a 60+ FPS GPU at 4K max quality.
How many of those do we currently have?

2080Ti (crazy price for new), hard to obtain 3080 and hard to obtain crazy price 3090.