Question Speculation: RDNA3 + CDNA2 Architectures Thread

Page 181 - 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
2,705
6,427
146

Heartbreaker

Diamond Member
Apr 3, 2006
4,340
5,464
136
There's an Asus ROG to compete with the Steam Deck now that promises 50% more performance at 15W and 100% at 35W and apparently delivers.

How much is thanks to RDNA3 and how much the Zen 4 cores help I wonder...
Other thing that makes me wonder, is how crippled it is by the SO. All these competitors that came out recently with "better" hardware disappointed being not significantly faster than the Deck.

Steam Deck was significantly behind even a Ryzen 6800U (8Core/12CU) since Steam Deck is only 4Core/8CU.
 

Panino Manino

Senior member
Jan 28, 2017
869
1,119
136
Steam Deck was significantly behind even a Ryzen 6800U (8Core/12CU) since Steam Deck is only 4Core/8CU.

In theory, not in practice.
That's why I mentioned these recent competitors that came out with more cores and CUs. No one performed as expected for the specs. The Deck is really like a "console", delivering more performance with less.
 

Heartbreaker

Diamond Member
Apr 3, 2006
4,340
5,464
136
In theory, not in practice.
That's why I mentioned these recent competitors that came out with more cores and CUs. No one performed as expected for the specs. The Deck is really like a "console", delivering more performance with less.

The ASUS numbers are all theoretical at this point.

Edit: It took a while to find a review:


At equal 15W they are very close.

But Neo can run at higher power for significantly more performance.

Perf4_hmlUkLQ.jpg
 
Last edited:

moinmoin

Diamond Member
Jun 1, 2017
5,064
8,032
136
The Deck is really like a "console", delivering more performance with less.
The Deck does have some optimizations that don't seem obvious.
 

GodisanAtheist

Diamond Member
Nov 16, 2006
7,166
7,666
136
The Deck does have some optimizations that don't seem obvious.

-Very cool article. Basically the Deck APU makes some serious CPU sacrifices to ensure the GPU is kept fed by the LPDDR5 ram, giving the chip a very favorable compute to bandwidth ratio.

There is some legit custom work that makes the Deck APU more gaming focused than your run of the mill desktop APU.
 
  • Like
Reactions: Tlh97 and moinmoin

KompuKare

Golden Member
Jul 28, 2009
1,191
1,487
136
They might also be somewhat packaging and assembly limited.

Can't help but feel that the way they approached going multi chip was excessively risky for little gain.

A far more cautious approach would have been going multi chip only for the top card and then making the top chip bigger - get a card out there which clearly wins all benchmarks, price it high and work out everything about how to go multi chip from that.

For the mid and low range, stick with monoliths.
 

moinmoin

Diamond Member
Jun 1, 2017
5,064
8,032
136
I wonder if Bergman's retirement got something to do with RDNA3 execution.
Barely 4 years at AMD does seem short for a former CEO of Synaptics (of 7+ years). Generous take would be he was brought in for a specific goal and met it.

Main optimization seems to be choking off the CPU BW so more is left for the GPU.
GPU runs at very low frequency as well. It's all an interesting balance.

They might also be somewhat packaging and assembly limited.
At this point they shouldn't be. I'm not seeing the other market that bottlenecks them. PS5? Naw.
 
  • Like
Reactions: Tlh97 and Leeea

Ajay

Lifer
Jan 8, 2001
16,094
8,109
136
The price probably doesn't excite AMD at all. Fair or not I'm guessing they would feel like they would have to sell it for less than the $599 of the 4070.
So, basically, their margins are so poor it's not worth it. AMD can use it's wafer allocation for other, higher margin, chips. Also, that means an entire generation of GPUs (RDNA3 dGPUs) are a financial failure. I think Bergman blew it, and that's why he's out (Lisa Su -> hey Bergman, just a thought, you could retire now and enjoy your family and such, or ....).
 
  • Like
Reactions: Tlh97 and Leeea

Mopetar

Diamond Member
Jan 31, 2011
8,113
6,768
136
I expect we'll start to see information start to trickle out by the end of the month. The rumors at the time of launch said that they had to do some significant redesigns and that we wouldn't see N32 until the middle of 2023.

I don't know if there's anything major prior to Computex, but AMD probably wants to get as much free press out of any announcement as they can.
 
  • Like
Reactions: Tlh97 and coercitiv

gdansk

Platinum Member
Feb 8, 2011
2,962
4,493
136
I am very skeptical of a significant re-spin.
Energy/money better spent on RDNA4. Market conditions aren't good and it was known even at the end of the last year when such a decision would have been made. Why spend more to fix a product that won't sell much even if it wasn't a dud.
 

Mopetar

Diamond Member
Jan 31, 2011
8,113
6,768
136
Unless RDNA4 was planned to be developed and ship not long after RDNA3, AMD is leaving a huge hole in their product line by scrapping N32.

Depending on how much the architectures have in common the fixes they would need for RDNA3 would also be needed for RDNA4 as well.
 

jpiniero

Lifer
Oct 1, 2010
15,223
5,768
136
Unless RDNA4 was planned to be developed and ship not long after RDNA3, AMD is leaving a huge hole in their product line by scrapping N32.

Depending on how much the architectures have in common the fixes they would need for RDNA3 would also be needed for RDNA4 as well.

They have a product. It's unloading the remaining stock of RDNA 2.
 

Ajay

Lifer
Jan 8, 2001
16,094
8,109
136
I am very skeptical of a significant re-spin.
Energy/money better spent on RDNA4. Market conditions aren't good and it was known even at the end of the last year when such a decision would have been made. Why spend more to fix a product that won't sell much even if it wasn't a dud.
Unless RDNA4 was planned to be developed and ship not long after RDNA3, AMD is leaving a huge hole in their product line by scrapping N32.

Depending on how much the architectures have in common the fixes they would need for RDNA3 would also be needed for RDNA4 as well.
Well, with the market conditions being poor, especially for AMD GPUs, they can ship or not ship RDNA3 and it may not matter much. 'Wasting' any addition resources on RDNA3 in a down market wouldn't make sense. That said, it's unlike AMD would skip a re-spin if they've already done the engineering for it. They can just keep clearing the channel, sell whatever they can make the most money on, and then ship RDNA4 as planned (FAD presentation Graph hints at end 2023/beginning 2024, but is vague as usual). Last note, the FAD presentation was around 1/3rd RDNA and 2/3rds CDNA - so it's clear that AMD's high performance commercial offerings are really where it is at for them. Not that that is a surprise.