Question Speculation: RDNA3 + CDNA2 Architectures Thread

[uzzi38]

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 is much bigger than from RDNA1 to RDNA2. We should expect many big improvements in GFX11. 🤔" / Twitter

 

[Shmee]

Hmm yeah, it could just be that current drivers are more locked down. Hopefully that will change and we will have more control. I guess we can always hope for continued driver improvements, which can include better OC support.

 

[Karnak]

before anybody mentions about pcie slot being 75W, the spec only allows 5.5A max on 12V, which is 66W

That's why I'm a bit surprised that they're pushing it all the way up to 355W. There's... not much headroom left. Hopefully it won't be a Polaris 2.0 situation.

If there's a single game which will exceed 366W I bet there'll be another [redacted].

Profanity is not allowed in the tech forums.

AT Mod Usandthem

 

[Hail The Brain Slug]

That's why I'm a bit surprised that they're pushing it all the way up to 355W. There's... not much headroom left. Hopefully it won't be a Polaris 2.0 situation.

If there's a single game which will exceed 366W I bet there'll be another shitstorm.

They already let the 6950XT go over 400W and I haven't seen anyone complaining. They likely keep the slot power draw locked down to prevent another RX480 situation, though. I have drawn >800W sustained on my 6900XT with a 3x8 pin and no slot issues whatsoever.

 

[PJVol]

but from what perspective is it a 4080 competitor?

From it's announced price I suppose, what else can indicate a competitor?

 

[Karnak]

They already let the 6950XT go over 400W and I haven't seen anyone complaining.

Out of the box or raising the PL manually? According to TPU max. power consumption is somewhere around 340W.

I have drawn >800W sustained on my 6900XT with a 3x8 pin and no slot issues whatsoever.

Yeah it's no problem at all but it's still out of spec.

 

[A///]

8 pins can deliver more power than 150 watts each. 150 watts is the steady safe amount. its specced for this taking into account the number of #2 quality power supplies.

 

[Hail The Brain Slug]

Out of the box or raising the PL manually? According to TPU max. power consumption is somewhere around 340W.

It's a 335W power limit, allowing +20% to be user configured.

Yeah it's no problem at all but it's still out of spec.

That wasn't my point, my point was the last time I saw anyone riled up about exceeding spec power delivery was the RX480 burning out pcie slot commotion that happened back in 2016. I assume AMD has locked the slot draw down to prevent anything like that from ever happening again, even if it was only one or a few isolated incidents.

 

[Joe NYC]

8 pins can deliver more power than 150 watts each. 150 watts is the steady safe amount. its specced for this taking into account the number of #2 quality power supplies.

You can get additional 75 Watts from the PCI bus.

 

[mrcmtl]

From it's announced price I suppose, what else can indicate a competitor?

Of course they will say it's a 4080 competitor. If they said it was a 4090 competitor, then we'll "probably" say they've failed.

 

[gdansk]

Of course they will say it's a 4080 competitor. If they said it was a 4090 competitor, then we'll "probably" say they've failed.

AMD priced it to compete with the 4080, AMD say it competes with the 4080. The specs appear to be between the 4080 and the 4090.
Only the "RDNA3 is the messiah" hype train said it was a 4090 competitor.

 

[A///]

You can get additional 75 Watts from the PCI bus.

this is widely known. An 8 pin can draw much more than most people can imagine. 150 watts is standardized on atx to make up for bad psus. A good psu in theory can supply more than 250 watts through a single 8 pin. even more than 300 in spike instances.

 

[DiogoDX]

You can get additional 75 Watts from the PCI bus.

After the RX480 fiasco I don't think that AMD will ever draw much power from the pci-e slot again. They will just draw a little more than 150W from the 8 pin cables and will be fine.

Just remember that AMD lauched a 295X2 that was a 500W dual GPU with 2X8pin and the card worked fine on good PSUs.

 

[Carfax83]

RT has taken longer to go from unaplayable to its current state because it is a lot more complex but it is still not a default on option because in many games turning it on makes the game unplayable, even on a 4090 at native.

I would say that the RTX 4090 can easily play most games with ray tracing enabled at 4K native without DLSS. Like AA and AF, ray tracing is scalable. The ray tracing quality we have now in games pales in comparison to that which is used in big budget movies that require rendering farms, but those are rendered offline from what I understand so performance is secondary to quality.

Over at TPU, the RTX 4090 fails to hit 60 FPS in only one title at 4K native with RT, and that's Cyberpunk 2077.

TechpowerUp RTX 4090 review

Over at Techspot it was the same, but with Dying Light 2 added as well. The RTX 4090 is just a hair under 60 FPS with RT enabled.

Techspot RTX 4090 review

So right now, today I do not think RT is common enough or performant enough to be a primary purchasing consideration. By next gen I expect the shift will have happened and then RT performance does become a primary purchasing consideration.

Well to me the RTX 4090 is a full blown 4K RT capable card in every sense of the word, especially with DLSS 3.0. However, it remains to be seen what kind of RT performance the mainstream versions will have. The mainstream GPUs are of much greater importance when it comes to industry adoption.

 

[Carfax83]

My system is even more outdated than yours, but Zen 4 V-Cache is the one I want, and it is worth the wait.

Zen 4 3D is definitely going to be a great gaming CPU, but so is Raptor Lake and it's already available. Also, one thing I like about Raptor Lake compared to Zen 4 is that it can utilize ultra high speed DDR5. No reviewer has yet to benchmark Raptor Lake with DDR5 7200 and above memory speeds using an RTX 4090. Should be interesting!

 

[Unreal123]

Something has gone very wrong.

https://twitter.com/x/status/1588652253048868864

 

[biostud]

Something has gone very wrong.

https://twitter.com/x/status/1588652253048868864

That could very well be the case. If the goal were a 3GHz GPU.

 

[GodisanAtheist]

I think its pretty obvious something went wrong and we're seeing the clean-up crew.

Maybe:

- There was supposed to be another SKU above the 7900XTX with dual GCD and it just wasn't performing.
- Drivers, drivers, drivers
- Clockspeeds can't go up due to desynch or something with the MCDs or interconnect

The idea AMD would come out swinging with the 6900XT and then settle back into the role of second best in the following generation is obviously the punchline of a bad joke.

 

[A///]

I'd say there's another card in the pipeline. People old enough remember amd and before the buy ati used to have multiple high end cards. most consumers aren't getting 4090s and they weren't buying 3090s. The 1600 price tag for founders minus tax is a large cost to foot that most people won't buy unless they need the very best with the cash flow to do it. nvidia's 4090 is plagued with problems at the moment. it is the butt of jokes online. Amd targetting the 4080 16 gb and below is another blow to nvidia.


people like this kepler moron are upset over being deceived when you should never put faith in leaks because they're leaks and they can be wrong even if they've been right in the past. I don't recall amd stating they were targetting 3 ghz clock speeds. the current clocks setup they have is something they would have planned for long ago, before the 3 ghz bs flew out of someone's bum.

 

[Timorous]

Something has gone very wrong.

https://twitter.com/x/status/1588652253048868864

Assuming the leakers had solid design target information it just looks like they missed the clockspeed target at the reference TBP so had to back it off and accept lower performance.

If those 3.5Ghz clocks were a genuine design goal then it missed by one heck of a lot and that sort of delta would move it from the current performance of between the 4090 and 4080 16G in raster to clearly ahead of the 4090 in raster and probably on par with the 4080 16G in RT. That would have meant a $1,200 price tag most likely and for that you get the same RT performance but you get $1,600 + worth of raster performance.

 

[Timorous]

people like this kepler moron are upset over being deceived when you should never put faith in leaks because they're leaks and they can be wrong even if they've been right in the past. I don't recall amd stating they were targetting 3 ghz clock speeds. the current clocks setup they have is something they would have planned for long ago, before the 3 ghz bs flew out of someone's bum.

The setup is something they would have planned for as is the target TBP but if they had a 1.6x perf/watt target vs the 6900XT stock vs stock and had planned on a 355W TBP it means they were targeting 4090 raster performance or there abouts and it seems like they missed the internal goal and even the advertised hit of 1.54x perf/watt was not stock vs stock but was 300W 7900XTX vs stock 6900XT so it looks like they missed by quite a bit which suggests the v/f curve they have is not what they were expecting hence lower clocks.

Given how succesful they have been with clocks in RDNA2 and in Zen 4 it is a bit of a surprise but I expect it will get resolved for a refresh product or for RDNA4.

 

[Unreal123]

The setup is something they would have planned for as is the target TBP but if they had a 1.6x perf/watt target vs the 6900XT stock vs stock and had planned on a 355W TBP it means they were targeting 4090 raster performance or there abouts and it seems like they missed the internal goal and even the advertised hit of 1.54x perf/watt was not stock vs stock but was 300W 7900XTX vs stock 6900XT so it looks like they missed by quite a bit which suggests the v/f curve they have is not what they were expecting hence lower clocks.

Given how succesful they have been with clocks in RDNA2 and in Zen 4 it is a bit of a surprise but I expect it will get resolved for a refresh product or for RDNA4.

Kepler is saying on Reddit that AMD underestimated Rtx 4090. It is giving same performance as AMDRX 7900 XTX hoped. He said normally generation performance increase 50% max but nvidia did 80% this time and caught AMD of guard. AMD never thought Rtx 4090 will be this fast.

 

[Timorous]

Kepler is saying on Reddit that AMD underestimated Rtx 4090. It is giving same performance as AMDRX 7900 XTX hoped. He said normally generation performance increase 50% max but nvidia did 80% this time and caught AMD of guard. AMD never thought Rtx 4090 will be this fast.

That is also understandable. If 4090 hit top end of NV projections and 7900XTX is about middle of what AMD expected then that puts you behind a bit relatively.

The 4080 is really cut down though so lets wait and see those reviews because I am not convinced it will scale as well in RT as the 4090 does and that will make things very interesting when comparing the 4080 and the 7900XTX

 

[Unreal123]

That is also understandable. If 4090 hit top end of NV projections and 7900XTX is about middle of what AMD expected then that puts you behind a bit relatively.

The 4080 is really cut down though so lets wait and see those reviews because I am not convinced it will scale as well in RT as the 4090 does and that will make things very interesting when comparing the 4080 and the 7900XTX

AMD expected to be max 50% faster than rtx 3090 but according to Kepler When Reviews came of rtx 4090 than majority of reviews even said it is 100% faster in some game on raster compared to rtx 3090. He also said that Rtx 4090 lack of hype also played big role of convicing Amd that rtx 4090 will be a turd.

Kepler also said some media played big part in downplaying rtx 4090 and convicing Amd to be beat Nvidia this gen.

 

[AtenRa]

AMD expected to be max 50% faster than rtx 3090 but according to Kepler When Reviews came of rtx 4090 than majority of reviews even said it is 100% faster in some game on raster compared to rtx 3090. He also said that Rtx 4090 lack of hype also played big role of convicing Amd that rtx 4090 will be a turd.

Kepler also said some media played big part in downplaying rtx 4090 and convicing Amd to be beat Nvidia this gen.

Yes AMD relies on media to know the tech specs of the competition lol

 

[TESKATLIPOKA]

Kepler is saying on Reddit that AMD underestimated Rtx 4090. It is giving same performance as AMDRX 7900 XTX hoped. He said normally generation performance increase 50% max but nvidia did 80% this time and caught AMD of guard. AMD never thought Rtx 4090 will be this fast.

Why should we care about what he says? He is no leaker, no insider. He is just a fan, who acts like he knows some insiders, and was one of the guys hyping up RDNA3.
RDNA3 launch just ended up embarrassing him.
His clown tweet is more than enough proof of that.