• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Discussion RDNA4 + CDNA3 Architectures Thread

Page 305 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
DisEnchantment

DisEnchantment

Golden Member
1655034287489.png
1655034259690.png

1655034485504.png

With the GFX940 patches in full swing since first week of March, it is looking like MI300 is not far in the distant future!
Usually AMD takes around 3Qs to get the support in LLVM and amdgpu. Lately, since RDNA2 the window they push to add support for new devices is much reduced to prevent leaks.
But looking at the flurry of code in LLVM, it is a lot of commits. Maybe because US Govt is starting to prepare the SW environment for El Capitan (Maybe to avoid slow bring up situation like Frontier for example)

See here for the GFX940 specific commits
github.com

History for llvm/lib/Target/AMDGPU - llvm/llvm-project

The LLVM Project is a collection of modular and reusable compiler and toolchain technologies. - History for llvm/lib/Target/AMDGPU - llvm/llvm-project
github.com github.com
Or Phoronix

More AMD "GFX940" Enablement Work Landing In LLVM - Phoronix

www.phoronix.com www.phoronix.com

There is a lot more if you know whom to follow in LLVM review chains (before getting merged to github), but I am not going to link AMD employees.

I am starting to think MI300 will launch around the same time like Hopper probably only a couple of months later!
Although I believe Hopper had problems not having a host CPU capable of doing PCIe 5 in the very near future therefore it might have gotten pushed back a bit until SPR and Genoa arrives later in 2022.
If PVC slips again I believe MI300 could launch before it :grimacing:

This is nuts, MI100/200/300 cadence is impressive.

1655034362046.png

Previous thread on CDNA2 and RDNA3 here

Question - Speculation: RDNA3 + CDNA2 Architectures Thread

Man I have been dying to make this one for a while now. First rumours for RDNA3 are here so new thread time! Just going to start off with this one for now: kopite7kimi on Twitter: "@VideoCardz Ah, I mean a simple mcm design with 10240 cores is not enough. Because the lift from RDNA2 to RDNA3...
forums.anandtech.com forums.anandtech.com
 
Last edited:
IEC said:
$729 would be DoA pricing. Even if it beat 5070 Ti in everything by a few % (which it won't, because AMD doesn't have MFG) people won't accept <20% discount over comparable nVidia product because the nV mindshare and technology advantage is so strong at this point.

Would have to approach 5080 levels of performance to justify a $729 price tag, and even then some nV diehards would just wait for a cheaper 5080.
Click to expand...
issue is 5070 Tis won't actually sell for $749 but more like $1000 same as with inflated 5080 prices at $1500 or higher.

The real issue is then 7000 series availability as even a 7800xt at $450 would be a steal, better performance/dollar.
 
beginner99 said:
issue is 5070 Tis won't actually sell for $749 but more like $1000 same as with inflated 5080 prices at $1500 or higher.

The real issue is then 7000 series availability as even a 7800xt at $450 would be a steal, better performance/dollar.
Click to expand...


According to a leak from IT Home, AMD is launching RX 9000 at a very competitive, sharp price, which will trouble Nvidia and even its own GPU, RX 7800XT. While AMD already planned to stop production in the third quarter of 2025, it wants to stop production as early as January. After this leak, we understand that AMD wants to cease production of RX 7800XT because a new competitor is on its way to outperform this GPU.

pkinsight.com

AMD RX 9000 Series Price Reportedly Kills Nvidia RTX 50 Series

AMD is looking to launch the RX 9000 at a very competitive price to outcompete Nvidia's RTX 50 models and even its RX 7800XT.
pkinsight.com pkinsight.com
 
marees said:
According to a leak from IT Home, AMD is launching RX 9000 at a very competitive, sharp price, which will trouble Nvidia and even its own GPU, RX 7800XT. While AMD already planned to stop production in the third quarter of 2025, it wants to stop production as early as January. After this leak, we understand that AMD wants to cease production of RX 7800XT because a new competitor is on its way to outperform this GPU.

pkinsight.com

AMD RX 9000 Series Price Reportedly Kills Nvidia RTX 50 Series

AMD is looking to launch the RX 9000 at a very competitive price to outcompete Nvidia's RTX 50 models and even its RX 7800XT.
pkinsight.com pkinsight.com
Click to expand...
They based it on MLID and chinese guy saying sharp.
So 499$+100$ = 599$ coz midwell. Still 599$ is good price would rather see 579$. 699$ is atrocious would mean 1k+ in europe xD
 
del42sa said:
https://wccftech.com/nvidia-mainstream-rtx-5070-rtx-5060-to-witness-launch-delay/

AMD's March RDNA4 release plans fall off, release now postponed indefinitely ...😎
Click to expand...
5070 dont matter.
If i were amd i would show 9070xt vs 5080 in 1080p and 4k (include many esports 1080p where 5080=4080) and 5070ti and 9070 vs 5070ti (show its on par even if u have to lie xD) thats it.
Then show 4 selected RT games w/o PT vs 5070ti
First party on 19th (5070ti embargo) reviews and launch week later. During the week let AIB present the cards show fsr4 deep dives etc.
599$ / 499$ and ur done shelves empty i even think 100K gpus would not be enough.

Plan B:
699$ / 599$ still good / mixed reviews low sales. FIRE SALE 2weeks later 649$ / 549$, another FIRE SALE 1-2months later 599$/ 499$ xD


1739432744686.png
 
gaav87 said:
5070 dont matter.
If i were amd i would show 9070xt vs 5080 in 1080p and 4k (include many esports 1080p where 5080=4080) and 5070ti and 9070 vs 5070ti (show its on par even if u have to lie xD) thats it.
Then show 4 selected RT games w/o PT vs 5070ti
First party on 19th (5070ti embargo) reviews and launch week later.
Click to expand...
If they want the same backlash like the 7900xtx launch, where the RDNA3 numbers were much better than in reality, that would be the way to go.

gaav87 said:
Plan B:
699$ / 599$ still good / mixed reviews low sales. FIRE SALE 2weeks later 649$ / 549$, another FIRE SALE 1-2months later 599$/ 499$ xD
Click to expand...
This is the normal AMD release strategy.
 
3100mhz boost and 2518mhz mem ? should it be 2500mhz mem ?
Leaked amd slides were with 2970mhz so +4% to that would put it 1-2% above 4080s nice.

1739447185685.png
 
gaav87 said:
3100mhz boost and 2518mhz mem ? should it be 2500mhz mem ?
Leaked amd slides were with 2970mhz so +4% to that would put it 1-2% above 4080s nice.

View attachment 116977
Click to expand...
Pixel fillrate is low?
That 7800XT
e2701b8352ccb088bf5f4955cbddf029.png



Either 285k bottlenecks hard in monster hunter or its f'in bearly faster than 6800xt and 4070s
Click to expand...
We don't know setting.
 
beginner99 said:
The real issue is then 7000 series availability as even a 7800xt at $450 would be a steal, better performance/dollar.
Click to expand...
Here in the U.S. they don't exist anymore. You want an 7800XT it is over $600.

It isn't going to matter how badly reviewers trash the new cards. Or if they are overpriced. They are going to sellout fast.
 
SolidQ said:
Pixel fillrate is low?
That 7800XT
e2701b8352ccb088bf5f4955cbddf029.png




We don't know setting.
Click to expand...
Seems very wrong, I wouldn't trust it. The 7800XT and 7700XT have 96 ROPs, so 64 on the 9070 XT would be a weird regression.

Spec wise with 64CU and a 3100MHz boost clock it should have about the same TFLOPs as a 84CU 7900XT @ 2400MHz. A 7900 XT game clock is quite a bit higher than 2400MHz though, so actual benchmarks will be interesting.
 
gaav87 said:
2518mhz mem ? should it be 2500mhz mem ?
Click to expand...
Official specs (and leaks) are always rounded, especially for memory. I think there are technical reasons like clock generators why actual memory clocks almost never exactly match the official spec.
Though this could of course also simply be an AIB OC model with a token mem OC (not that 18 mhz would do anything worthwhile for perf, so whatever).
MrTeal said:
Seems very wrong, I wouldn't trust it. The 7800XT and 7700XT have 96 ROPs, so 64 on the 9070 XT would be a weird regression.
Click to expand...
AMD likes to keep the ROP count as low as they can afford at a given performance target, so with them, I'd never rule that out.
If they could cut ROP count per SE in half and save a bunch of area and energy at a mostly insignificant performance impact, they'd probably do it.
 
Last edited:
You must log in or register to reply here.

TRENDING THREADS

Back
Top