• 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 RDNA 5 / UDNA (CDNA Next) speculation

Page 110 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
That's another bad assumption. It will sell for more than an RTX 5090 if they caught the customer they hunted. But not to us. There's no way you build a chip that big on the hope to sell it to DLSS enthusiasts.
People who buy high end GPUs for gaming (like myself) run 4k+ monitors and want to enjoy 4k native 100 FPS+ perf. Only handful of games (UE5 based, CP2027 PT) really require DLSS and usually Quality works just fine: 1440P -> 4K upscaling works fine in FSR too, where DLSS works better is much lower reses, but that's not the situation for high end GPUs.

All new games will have support for a form of FSR, so new high end GPU will run very fast all older games native (as the nature intended) and if necessary FSR can be used for new ones.

That's why I am open to buying AT0 - as long as it outperforms 5090 by 30%+, DLSS is not a factor, we are not talking 60/70 class here where it indeed can be a very large factor.
 
They did it half a year ago.
I see Kepler_L2 disagreeing with you on this and I'm trying to reconcile who has the more current inside knowledge. Because Kepler_L2 says that AMD will still launch the AT0 RDNA 5 flagship GPU for gaming to compete with NVIDIA in high-end gaming GPUs for the next generation. I'm curious to know how you came to your conclusion.
 
Never had any after the RV770 suicide.

This take makes absolutely no sense at all.

That was 18 years ago and AMD kept their dGPU marketshare above 30% for another 7 years and 4 new architectures after RV770's release.


1772805589573.png


Nvidia would only start to distance themselves with Maxwell.



I see Kepler_L2 disagreeing with you on this and I'm trying to reconcile who has the more current inside knowledge.
@Kepler_L2 has a solid track record. It's not even close.
 
I see Kepler_L2 disagreeing with you on this and I'm trying to reconcile who has the more current inside knowledge. Because Kepler_L2 says that AMD will still launch the AT0 RDNA 5 flagship GPU for gaming to compete with NVIDIA in high-end gaming GPUs for the next generation. I'm curious to know how you came to your conclusion.
it will launch but probably a prosumer version for dual use as AI & Gaming, just like Radeon VII

So it won't compete directly with nvidia — in gaming
 
it will launch but probably a prosumer version for dual use as AI & Gaming, just like Radeon VII

So it won't compete directly with nvidia — in gaming

If desktop AT0 competes in performance against the 6090 then it competes directly with nvidia in gaming, as the 6090 like the other x090/Titan cards should be a prosumer option.

The Radeon VII couldn't be a prosumer card in the same range as the Titan RTX because it was neither large/wide enough (less transistors than the RTX 2080 TU104) nor was it performant enough (it lagged in performance against the RTX 2080 despite using brand new 7nm from TSMC).
 
Last edited:
I see Kepler_L2 disagreeing with you on this and I'm trying to reconcile who has the more current inside knowledge. Because Kepler_L2 says that AMD will still launch the AT0 RDNA 5 flagship GPU for gaming to compete with NVIDIA in high-end gaming GPUs for the next generation. I'm curious to know how you came to your conclusion.
The die is alive and well. Client product is dead.
That was 18 years ago and AMD kept their dGPU marketshare above 30% for another 7 years and 4 new architectures after RV770's release.
I too love loss leaders.
RV770 made Radeon a poverty brand, a place from which there was no return. RIP.
 
If desktop AT0 competes in performance against the 6090 then it competes directly with nvidia in gaming, as the 6090 like the other x090/Titan cards should be a prosumer option.

The Radeon VII couldn't be a prosumer card in the same range as the Titan RTX because it was neither large/wide enough (less transistors than the RTX 2080 TU104) nor was it performant enough (it lagged in performance against the RTX 2080 despite using brand new 7nm from TSMC).
6090 is a client part, and no, AMD isn't trying to compete with NV anymore.
 
If desktop AT0 competes in performance against the 6090 then it competes directly with nvidia in gaming, as the 6090 like the other x090/Titan cards should be a prosumer option.
As an RTX 5090 owner, it is 100% a consumer gaming GPU. Especially when you play games at 4K with Path Tracing enabled and you are still needing DLSS Quality Mode to get over 60FPS with max settings on games like Cyberpunk 2077. The only GPU that you could call "prosumer" would be the RTX PRO 6000. So, if AMD is going to use the RDNA 5 AT0 flagship GPU to compete against the equivalent of the RTX PRO 6000, then I would say the AT0 is a "client only" or enterprise only GPU; otherwise, it will 100% still be in the running as a high-end gaming GPU just like the NVIDIA xx90 models are.
 
As an RTX 5090 owner, it is 100% a consumer gaming GPU. Especially when you play games at 4K with Path Tracing enabled and you are still needing DLSS Quality Mode to get over 60FPS with max settings on games like Cyberpunk 2077. The only GPU that you could call "prosumer" would be the RTX PRO 6000. So, if AMD is going to use the RDNA 5 AT0 flagship GPU to compete against the equivalent of the RTX PRO 6000, then I would say the AT0 is a "client only" or enterprise only GPU; otherwise, it will 100% still be in the running as a high-end gaming GPU just like the NVIDIA xx90 models are.
It's a cloud VDI part.
 
As an RTX 5090 owner, it is 100% a consumer gaming GPU. Especially when you play games at 4K with Path Tracing enabled and you are still needing DLSS Quality Mode to get over 60FPS with max settings on games like Cyberpunk 2077. The only GPU that you could call "prosumer" would be the RTX PRO 6000. So, if AMD is going to use the RDNA 5 AT0 flagship GPU to compete against the equivalent of the RTX PRO 6000, then I would say the AT0 is a "client only" or enterprise only GPU; otherwise, it will 100% still be in the running as a high-end gaming GPU just like the NVIDIA xx90 models are.

As a RTX 5090 owner, it is 100% as much of a prosumer GPU as all the Titan dGPUs that were sold throughout the last 13 years.
The RTX Pro 6000 is a workstation GPU released for $8.5K, 4x the MSRP of the 5090. It's has all of the "pro" from professional and none of the "sumer" from consumer.

You running into games that saturate the card doesn't make the card less of a prosumer, that's not what it means. We've always had games that couldn't run maxed out on new hardware.


1772813414184.png




It's a cloud VDI part.
Nah. Consumer part isn't cancelled. Repeating it more doesn't make it real.

Unless AT0 is cancelled entirely it will 100% release to consumers, even if it's a token release with no real volume like the Radeon VII.
 
Oh really? That's interesting. So, AMD is going to use the AT0 for Virtual Desktop Infrastructure for enterprise rather than a discrete GPU for your general consumer? I guess that makes sense.
AT0 seems more like a GPU designed for the iterative upgrade of XCloud.
 
Oh really? That's interesting. So, AMD is going to use the AT0 for Virtual Desktop Infrastructure for enterprise rather than a discrete GPU for your general consumer? I guess that makes sense.
not AMD. Microsoft. the contracts are signed & its a done deal

it is azure xcloud infrastructure. a single AT0 will virtualize into 8 xcloud sessions (~8x series S2)
 
You know you can run IP roadmap without torching pallets of cash on dumb client graphics misadventures?
No, you can't run a credible IP roadmap without people actually shipping silicon and learning from their mistakes, and you certainly can't do it for over a decade with zero revenues. AMD lost to Nvidia because their main business was self-bulldozered, only maybe in last 5-6 years they finally had the money to improve GPU side of things, and sadly they underinvested.
Nah you just want AT0 to exist to exist in the consumer space, just so that you could buy 6090 for cheaper
Why do you think people who can afford to buy 5090 won't be able to do so with 6090? I am on 4090 but only because I refuse to upgrade to only 30%+ above my current card, when 4090 came out I bought it straight as upgrade to 3090, and before that I was on Titan: double perf gets my money, so if AMD releases AT0 and it is 30%+ above 5090 then this should be enough for me to upgrade, not least because RDNA5 seems to have exciting list of tech that will last for a bit, and being used in PS6 it will make ports better (unless Sony stops them, which sure looks that way now).

Bottom line is this: -90 series buyers are price pretty insensitive, so drop your juvenile argument that at best works with 60/70 series.

And also, what's your problem with more competition to keep prices lower for consumers? Obviously that should be encouraged, otherwise we end up with monopoly - like Nvidia.

It's a cloud VDI part.

And Zen5 is a server chip, so what - stop selling consumer chips?
 
Last edited:
They're about to.
They might for Zen 6, for reasons that you know very well - crazy server chip demand _and_ it's on N2 which is also used for high margin MI45x, but will they stop selling consumer Zen5 and reset market share to 0?

No, they won't - sales will reduce naturally though due to high memory prices, so they'll just move them to EPYCs.

Yes you can and no you don't need loss leader parts for that.
No, that's can't be done in a PLC in current times (or 10 years ago).
 
Back
Top