Discussion RDNA4 + CDNA3 Architectures Thread

[DisEnchantment]

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

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

Or Phoronix

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

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

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



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...

socialtechwork.com

 

[Kepler_L2]

Execution hasn't been a thing at Radeon recently.

How does one cancel everything that was complex about the lineup and still end up delaying everything.

RDOA4 indeed.

They did move a lot of engineers over to MI series chasing that AI money. But still this doesn't look good.

 

[Mahboi]

Should we expect the PS5 Pro to be delayed with it then?

 

[CouncilorIrissa]

Should we expect the PS5 Pro to be delayed with it then?

Not really, different IP I guess.

 

[Mahboi]

Also, the gcc patches and mesa patches have already started being fired. Doesn't seem like software will be delayed, so can we assume they found another problem in the silicon again?

 

[CouncilorIrissa]

Also, the gcc patches and mesa patches have already started being fired. Doesn't seem like software will be delayed, so can we assume they found another problem in the silicon again?

Would be a certified Radeon moment indeed.

It's almost as if RDNA2 was worked on by a different company. Every other GPU gen is a dumpster fire after dumpster fire.

 

[eek2121]

Also, the gcc patches and mesa patches have already started being fired. Doesn't seem like software will be delayed, so can we assume they found another problem in the silicon again?

No? If this is true (dug briefly but unable to find more info or a source) it might be due to all the roadmap changes.

It would not surprise me if true. Maybe others will know more. I haven’t been able to keep up with anything due to the flood of Computex news.

 

[Kepler_L2]

Should we expect the PS5 Pro to be delayed with it then?

No, that has been done for a while.

 

[Ajay]

They did move a lot of engineers over to MI series chasing that AI money. But still this doesn't look good.

Thanks. I wondered about this. Can't say I blame AMD for the transfers, but yeah, bad look for Radeon anyway.

 

[ToTTenTranz]

Oof

https://twitter.com/x/status/1798062738322919911

Wow this is terrible.

They're launching only 2 non-enthusiast chips after cancelling most of the lineup, and they're still delaying the new architecture?!

What does this even mean for RDNA5? Is it now coming in 2026 or later?

 

[branch_suggestion]

See you at CES.

 

[adroc_thurston]

Oof

https://twitter.com/x/status/1798062738322919911

Unlikely, they already dumped ASIC definitions into the kernel.

 

[jpiniero]

Wow this is terrible.

They're launching only 2 non-enthusiast chips after cancelling most of the lineup, and they're still delaying the new architecture?!

They want to get RDNA2/3 supply gone before releasing RDNA4?

 

[soresu]

They're launching only 2 non-enthusiast chips after cancelling most of the lineup, and they're still delaying the new architecture?!

Those non enthusiast chips make up market segments that comprise most discrete GPU sales.

 

[adroc_thurston]

Those non enthusiast chips make up market segments that comprise most discrete GPU sales.

That's not the point.
Less dies == less bringup complexity.

 

[soresu]

That's not the point.
Less dies == less bringup complexity.

An important point to be sure, but not what I was referring to.

 

[Saylick]

Wow, that sucks. For all of us, as consumers. Welp, here’s to hoping Intel Battlemage picks up the slack.

 

[TESKATLIPOKA]

See you at CES.

Which year?

 

[GodisanAtheist]

Wow this is terrible.

They're launching only 2 non-enthusiast chips after cancelling most of the lineup, and they're still delaying the new architecture?!

What does this even mean for RDNA5? Is it now coming in 2026 or later?

Gonna channel my best Adroc for this one... Ok here goes nothing...

Plans change

 

[deasd]

Just wow. I believed there would be RDNA4 launch in 2h2024 and possibly a preview at Computex. Nothing happened and being told it would be until next year? Does AMD still has flagship model of RDNA4 or cancelled?

I find it hard to believe most of the leaks are wrong in 2024.

 

[SolidQ]

They seems want announce it with RTX 5070/5060 and sell RTX cards, instead yourself.

 

[poke01]

I think we'll see RDNA4 this year. Later in year, around oct/nov.

 

[leoneazzurro]

To me, it has not so much sense. If reason is price, you can always adjust it later. If the product is ready, then delaying it means only losing potential sales. A real reason would be excess inventory of older cards that must be cleaned up before launching the new ones (which could be).

 

[coercitiv]

A real reason would be excess inventory of older cards that must be cleaned up before launching the new ones (which could be).

Combined with this would be Nvidia waiting until next year to introduce 5070 and under. How would AMD know how to price their products without guidance from professionals?!

 

[Mahboi]

I sure hope that it is just



being Wccftech.

Because another 4 months of completely pointless delay for a low midrange product which already has its mesa updates + Valve compiler updates out is just ridiculous.

 

[linkgoron]

Unless RDNA4 is also broken like RDNA3 in some unexpected way, I'd be very surprised if they delayed it. Code has landed in a lot of places, and AMD has no reason to delay. Waiting for the 5070 would be ridiculous.