Question Speculation: RDNA2 + CDNA Architectures thread

Page 126 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

uzzi38

Platinum Member
Oct 16, 2019
2,635
5,976
146
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
 

GodisanAtheist

Diamond Member
Nov 16, 2006
6,817
7,177
136
huh. I thought this was a reference to The Road Warrior, did some internetting...then learned a new-to-me thing about what has been happening on the internet.

I literally don't even know what is going on here.

0f1a262d2317cece28bd6b0e24ad9fd8.png

- Big Chungus. That's where we're at with this thread.

Is October 28th here yet?
 

Kuiva maa

Member
May 1, 2014
181
232
116
I'm sure AMD is quaking in their boots at the whole 3-5 games that use it. Knowing AMD they probably have been cooking up something that works with any game.

I tried Control under DLSS 2.0. The sharpenning part of DLSS is helping alleviating the muddy AA implemantation of this game but it introduces severe blocky artifacts within flat surface textures (eg when you are looking the bulletin boards and posters spread across the game). I still do not see me using it in any single player game where IQ and graphics fidelity matters. Not in that form at least.
 

Veradun

Senior member
Jul 29, 2016
564
780
136
The XBox Series X GPU is actually closer to the 2080S than the 2080ti (I know a guy who has access to one, though I haven't gotten to play with it, he works for a game studio). However, the desktop parts are clocked significantly.


The 120fps "cap"?



I'm sure AMD is quaking in their boots at the whole 3-5 games that use it. Knowing AMD they probably have been cooking up something that works with any game.
There still is the manual mode also known as "move the graphics settings sliders"
 

eek2121

Platinum Member
Aug 2, 2005
2,930
4,026
136
From VideoCardz: “AMD is set to launch three SKUs based on Navi 21. According to our sources, AMD is not launching Navi 22 or Navi 23 at the same time as Navi 21. Even the fourth variant of Big Navi should appear next year (Navi 21 XE).”

hah. Pulling an NVIDIA and sandbagging are we?
 

DisEnchantment

Golden Member
Mar 3, 2017
1,608
5,810
136
AMD is set to launch three SKUs based on Navi 21. According to our sources, AMD is not launching Navi 22 or Navi 23 at the same time as Navi 21. Even the fourth variant of Big Navi should appear next year (Navi 21 XE).

What is definitely worth pointing out is that TechPowerUP GPU-Z software is usually updated right before product launches. This might indicate that GPU-Z developer already had access to those graphics cards.
VideoCardz's sources are pretty bad these days. Thier entire Ampere stories were fed by Kopite who was right to the very end, even better than VCZ AIB sources who "confirmed 7nm"
Also TPU got some BIOS not card, he dumped the BIOS and extracted the values, which he himself said on Twitter.
 

exquisitechar

Senior member
Apr 18, 2017
657
871
136
VideoCardz's sources are pretty bad these days. Thier entire Ampere stories were fed by Kopite who was right to the very end, even better than VCZ AIB sources who "confirmed 7nm"
Also TPU got some BIOS not card, he dumped the BIOS and extracted the values, which he himself said on Twitter.
I think the Gainward (?) VCZ leak was legitimate, it's just that AIB vendors were kept in the dark about Ampere. That's why the 7nm thing didn't end up being true.
Blended as suspected.. Definition of customized and to me that means can't compare with dGPU.
I doubt that this is true.
 

Mopetar

Diamond Member
Jan 31, 2011
7,842
5,993
136
A few of the comments earlier seemed to imply that we'd be getting new GPUs shortly after the Radeon event at the end of the month.

I think it's a safe bet that Navi will be pretty good. We know that at worst it's pretty close to a 3080 and there's enough reason to suspect it could be better.

The real question is when it will drop. Is it almost right after to coincide with Zen 3 or is there going to be another gap that puts it closer to or even on Black Friday. Anyone have thoughts or any real indication of when it drops?
 
  • Like
Reactions: Tlh97 and Konan

Kuiva maa

Member
May 1, 2014
181
232
116
From VideoCardz: “AMD is set to launch three SKUs based on Navi 21. According to our sources, AMD is not launching Navi 22 or Navi 23 at the same time as Navi 21. Even the fourth variant of Big Navi should appear next year (Navi 21 XE).”

hah. Pulling an NVIDIA and sandbagging are we?

I have this nagging feeling they are not sandbagging but rather presented with a riddle on how to efficiently use their 7nm wafer allocation to optimally cover the market. I am sure they would love to get the full Radeon 6000 series within 2-3 months just like the early GCN days or earlier. Between having to also service console vendors and their own CPU range, this might have been impossible and a lost opportunity for more GPU sales.
 

Mopetar

Diamond Member
Jan 31, 2011
7,842
5,993
136
I have this nagging feeling they are not sandbagging but rather presented with a riddle on how to efficiently use their 7nm wafer allocation to optimally cover the market. I am sure they would love to get the full Radeon 6000 series within 2-3 months just like the early GCN days or earlier. Between having to also service console vendors and their own CPU range, this might have been impossible and a lost opportunity for more GPU sales.

Whatever we get at launch was a decision made months ago. It depends on existing stock as well. As much as any company would like a shiny new product in time for Black Friday it's also a good way to clean out all of the old RDNA1 and Polaris cards that could be replaced with newer Navi chips early next year.

I'm just glad that I can spend $500+ for a change and get a really great upgrade that isn't devalued due to a mining craze or awful pricing due to lack of competition. I'd just like to be able to get all of it before Christmas. I think I've been good this year. Honest.
 

amrnuke

Golden Member
Apr 24, 2019
1,181
1,772
136
After seeing how AMD seem to be downplaying the remarkable advances they made on Zen3 (it's an entire freaking new core design, every execution unit modified and optimized, etc. - producing a 19% IPC gain in 15 months), I think they're sandbagging RDNA2.

I think benchmarks on Zen 3 and RDNA2 are going to be better than what we've been expecting (I think most of us are expecting the low end of the possible range) the last couple of weeks.
 

moinmoin

Diamond Member
Jun 1, 2017
4,952
7,665
136
The real question is when it will drop. Is it almost right after to coincide with Zen 3 or is there going to be another gap that puts it closer to or even on Black Friday. Anyone have thoughts or any real indication of when it drops?
Before the consoles drop. /brokenrecord
 
  • Like
Reactions: Krteq

Panino Manino

Senior member
Jan 28, 2017
821
1,022
136
  • Like
Reactions: Tlh97 and raghu78

PhoBoChai

Member
Oct 10, 2017
119
389
106
What is he trying to pull off?
We already knew details about the SeX GPU.
"Front End is RDNA1"? Compute Units that are RDNA2 but without RT? But isn't the RT feature an integral part of the RDNA2 CU? Seriously, what is he scheming?

It's utter crap.

Front-end isn't RDNA 1 since SXS Command Processor is different and it supports Mesh/PS Shaders and VRS.

CU isn't RDNA 1 either, higher ops/clk, RT via TMUs which is part of the CU design. The texture sampler feedback in SXS also not RDNA 1.

Ppl need to be more critical of these rumors, it's a shame on the community that falls for such poorly thought rumors.

As for Videocardz, their sources have always been with AIBs. Since AIBs have been kept out of the dark this time around, they relied more on Ampere leaks from these twitter accounts, some of them are correct, others not. There is very little leaking of RDNA2 from within AMD this time, and AIBs know nothing until the launch is imminent!