AMD: Navi Speculation, Rumours and Discussion [2019-2020]

Status
Not open for further replies.
(I'm really not convinced they'd ever need 300W for the engineering sample of a card targeting 150-175W).
The first leaked Turing anything was a 450W (3*8pin) TU102 PCB.
If Navi isn't progressing well, wouldn't Lisa Su be blatantly lying to investors when stating the opposite, and isn't that a crime?
Intel&10nm saga.
So many clicks, so much bait, so much hyperbole, so little facts.. :rolleyes::runaway:
Hey, that's YT in a nutshell.
And with Intel hiring the oldskool tech press, that's all we've got.
 
Guess it depends precisely what she said and when.
Last year, may have been true until hitting issues.
She said this 4 days ago during the Q1 2019 earnings conference:

Lisa Su said:
Yeah, so, Hans, we are excited about Navi. Navi is a new architecture for us in gaming. It has a lot of new features, across the Navi architecture. Things are progressing well. We expect it to launch in the third quarter.



But they have been pretty quiet with their road map.
That's about the only perceivably good thing that has happened since Raja left: they don't constantly run their mouths off on how great their next chips will be, nor do they launch those super cringy videos of kids playing drums.
 
One could also say that YT channel is simply running with this horror story to cover that pipe dream of Navi for under $200 and under 150W competing with Vega 64 Liquid and GTX 2070.
Reality might be that Navi always targeted 225W as any mid range AMD GPU and slightly missed their target.
 
One could also say that YT channel is simply running with this horror story to cover that pipe dream of Navi for under $200 and under 150W competing with Vega 64 Liquid and GTX 2070.
Reality might be that Navi always targeted 225W as any mid range AMD GPU and slightly missed their target.
The first "leaked" information regarding performance stated Vega 56-performance level. I would be surprised if Navi wasn't able to reach Vega 56-performace at 150 W. Radeon RX 480 had 150W TDP, Vega 56 is 55 % faster. That means 55% performace boost (compared to RX 480) at the same TDP. It shouln't be problematic. E. g. RX 470 brought 80-90% performance boost compared to HD 7870 R9 370 at similar TDP level (110W/120W).
 
Last edited:
The first "leaked" information regarding performance stated Vega 56-performance level. I would be surprised if Navi wasn't able to reach Vega 56-performace at 150 W. Radeon RX 480 had 150W TDP, Vega 56 is 55 % faster. That means 55% performace boost (compared to RX 480) at the same TDP. It shouln't be problematic. E. g. RX 470 brought 80-90% performance boost compared to HD 7870 R9 370 at similar TDP level (110W/120W).

I'm not questioning that at all, just pointing towards YT speculations changing every few days to cover all the basis. One of them most likely will turn out right and youtuber will take ownership of it saying how good his sources are. We seen that happen over and over again.

Not pointing at any specific channel here, some of them get good leaks and bad leaks and some of them just post rubbish to bring in some revenue.

Navi should launch both in under 150W and over 150W products. I see Navi as Polaris replacement and expect similar power goals out of it.
Performance is to be seen, looking at my Vega 7, there is a good chance that it will perform really well if not pushed too far on the clocks.
History tells us that AMD will push for performance at the cost of power if they get a shadow of the chance to match or exceed corresponding nVidia product.
For me it doesn't matter, I can tweak their product to my liking, for Joe average I just hope they launch more optimal solution as they did with original RX470. It was power efficient and performed really good compared to Hawaii.
 
Well in factual news, if ASML EUV wafer throughput per hour is ideal and is assumed to be the production bottleneck for TSMC's 7nm+ node, then running each machine could put out over 2.5 million 100mm^2 dies in a 24 hour run period. Which is good since that's probably what Ryzen 2 and a theoretical midrange Navi die size would be (thanks 7nm!).

With TSMC apparently buying 18 EUV Twinscan machines that could mean they ramp up quickly enough to serve both AMD and Apple. Though AMD has already shipped a 7nm TSMC card so might be more comfortable with that tooling. Still, it appears to be possible that their launches this year will be 7nm+.

I saw a whole bunch of HW bugs in the LLVM commit.
https://github.com/llvm-mirror/llvm...939#diff-983f40a891aaf5604e5f0b955e4051d2R733

Someone have some idea how severe they are?

Not me, but in the link I did see the confirmation of Wavefront size of 64, and same LDS bank count (32) as GCN. So that's definitely GCN still.
 
Last edited:
Well in factual news, if ASML EUV wafer throughput per hour is ideal and is assumed to be the production bottleneck for TSMC's 7nm+ node, then running each machine could put out over 2.5 million 100mm^2 dies in a 24 hour run period. Which is good since that's probably what Ryzen 2 and a theoretical midrange Navi die size would be (thanks 7nm!).

With TSMC apparently buying 18 EUV Twinscan machines that could mean they ramp up quickly enough to serve both AMD and Apple. Though AMD has already shipped a 7nm TSMC card so might be more comfortable with that tooling. Still, it appears to be possible that their launches this year will be 7nm+.



Not me, but in the link I did see the confirmation of Wavefront size of 64, and same LDS bank count (32) as GCN. So that's definitely GCN still.
That doesn't mean much.Changing from single issue to VLIW2 does not necessarily change these parameters.
 
Well in factual news, if ASML EUV wafer throughput per hour is ideal and is assumed to be the production bottleneck for TSMC's 7nm+ node, then running each machine could put out over 2.5 million 100mm^2 dies in a 24 hour run period. Which is good since that's probably what Ryzen 2 and a theoretical midrange Navi die size would be (thanks 7nm!).

With TSMC apparently buying 18 EUV Twinscan machines that could mean they ramp up quickly enough to serve both AMD and Apple. Though AMD has already shipped a 7nm TSMC card so might be more comfortable with that tooling. Still, it appears to be possible that their launches this year will be 7nm+.
Zen 2 and Navi use 7N, not the 7nm+ EUV process. Same for Apples current chips
 
Rumor: AMD Radeon RX 3080 XT Will Challenge RTX 2070 at $339
May 7, 2019

If AMD undercuts Nvidia’s GPU pricing and does so with GPUs that lack ray tracing, it could be read as a tacit admission that Nvidia has established ray tracing as a feature that customers will pay more for. When AMD introduced Radeon VII, it deliberately didn’t price that GPU any lower than the RTX 2080, despite the fact that the Radeon VII completely lacks ray tracing. It is possible that the company will do something similar here, or choose to split the difference by pricing below the equivalent RTX GPU it intends to compete with, but not so low as to imply that Nvidia has properly priced in the value of ray tracing.
...
But as it stands, this rumor is, at best, incomplete. It implies an odd pricing structure that would require AMD to hit much higher clocks on GCN than it has ever demonstrated a capability to hit. The core counts also imply that AMD is relying heavily on efficiency gains to hit its performance targets, but efficiency gains in GPUs have been hard to come by of late. Vega was not, generally speaking, a large efficiency gain over previous versions of GCN. Could Navi change that? Yes. But historically, we’ve seen GPUs gain the most performance either by clock boosts (which GCN hasn’t been very good at) or core count increases (which this rumor implies have not occurred).
...
The proposed price structure makes limited sense without massive clock increases to drive performance in the upper tier products. And finally, it’s not clear why AMD would build two completely different chips between Navi 10 and Navi 20 if the difference between the two is just a 7 percent core count increase. This is much less of a gap than exists between the various Nvidia GPUs in their respective brackets and custom designs.
https://www.extremetech.com/gaming/290919-rumor-amd-radeon-rx-3080-xt-will-challenge-rtx-2070-at-339
 
Challenging is a strange term for me... Like, Vega 56 is already challenging 2070 IMO, in some games, it's not that far behind. And under or at 300e in europe already. The "first" Navi rumor was like vega64 perfs +10/15% but at lower price. We'll see....
 
Do we know if Navi was always planned for TSMC 7N or if it was planned for GloFo 7nm - maybe due to the WSA agreement - before GloFo ditched the 7nm node?

From what I remember GloFos 7nm node was said to be slightly ahead in comparison to TSMCs 7nm node in density and when it came to expectations regarding high clocks.

So, assuming the info in the video is correct, if Navi was planned for GloFo 7nm and developed with the design tools/specs provided by GloFo would this be something that could affect Navi in regards to clocks and thermals?
 
Status
Not open for further replies.
Back
Top