Question AMD Rembrandt/Zen 3+ APU Speculation and Discussion

[izaic3]

Alright, so we've had some leaks so far. I don't know if any of it's been confirmed yet, as it's pretty early, but here is what I've surmised so far (massive grain of salt of course):

• Zen 3 "enhanced"
• DDR5 and LPDDR5 support
• PCIE4
• 6nm
• CVML
• Navi 2/RDNA/RDNA 2
• 12 CU
• Socket FP7/AM5
• Possible USB4
• 2021(unlikely) -2022 (more likely) release date

If if turns out to have RDNA 2 and 12 CU, I could see iGPU performance potentially almost doubling over Cezanne.

If I've made any mistakes or gotten anything wrong, please let me know. I'd also love to hear more knowledgeable people weigh in on their expectations.

 

[uzzi38]

We have a score boys:

https://twitter.com/x/status/1450868227773501452

Single channel DDR5 and the performance is somewhere in the RX560/GTX1050 region (the 560 is the lower score here)

What'd I say huh? The performance uplift is greater than the improvement to memory bandwidth

This is actually less memory bandwidth than dual channel DDR4-3200 and it still beats the average 4800U score by 50%

 

[Mopetar]

What I'd be most interested in is whether the move to RDNA2 graphics technology also coincides with the inclusion of something like a shared L3/infinity cache between the CPU and GPU cores. APUs being particularly bandwidth constrained would see a big improvement on the graphical workload side and we've already seen many applications where Zen 3 has done well because of the larger cache available.

 

[uzzi38]

Rembrandt tape-out was less than 6 months after Cezanne, I would be very surprised if it's a 2022 release.

It's also a brand new platform with plenty of validation etc to be done. Tonne of new platform features - USB4, PCIe4 etc etc. Plus, a yearly cadence is a big selling point amongst OEMs, they like that kind of consistency.

I dunno, my bets still on CES next year

 

[NTMBK]

I don't entirely agree with this.

APU with multiple channels of memory in MCM is just going to outcompete on cost and power any dGPU + discrete memory with overlapping performance level.

At some point, you just can't compete offering same performance at higher cost and higher power consumption.

The only reason this continues to exist is Intel + AMD + OEMs operating like a cartel, preventing this solution from getting to the market.

But Apple is not controlled by this same cartel.

AMD aren't the cartel, they're the ones who are hamstrung by dumb OEM decisions. Who keeps putting single channel memory alongside APUs? Who keeps putting in dGPUs that are worse than the integrated graphics? It's not AMD.

Apple control the whole stack, so they can push this through. They also only target the high margin end of the market, so they can afford to spend billions of transistors on an enormous system level cache, and a memory bus wider than a 3080. They have a huge die on a cutting edge process, using advanced packaging techniques. This thing is going to be PRICEY. You can't churn out $500 HP craptops with those sort of design choices.

 

[soresu]

I'm agree, it is strange to say that Zen 2 is only an "evolution" of Zen. But it is not my opinion, it is the AMD's point of view, about their Zen µArchitectured famlily.

Here is an an old interview (end of 2019) of Forrest Norrod (AMD's senior vice president and general manager):



They share the same family number (17h). This seems clear.

True, though family number is not necessarily an accurate indication of difference - even CDNA2 still retains a number similar to Vega (GFX9) despite the large amount of changes made to differentiate it from the OG Vega design.

Also Norrod is not the best authority on the subject of µArch details, or how to classify them - he is a sales/marketing specialist, rather than R&D like Mike Papermaster, Mike Clark, or Suzanne Plummer.

Given his speciality and the timeframe being after Zen 2 was released I think it isn't surprising that he would downplay Zen 2 in favor of playing up the mystery of the coming Zen 3.

I think the product family number has more to do with compiler targets, although what exactly needs to change for a compiler to need a different target I don't know - off the top of my head I don't know what instruction set changes happened between Zen 1/+ and 2.

I think cache infrastructure plays some part in compiler targets too - anyone knowledgable please feel free to disabuse me of that 😅.

 

[uzzi38]

The iGPU is a little underwhelming, I expected a bit more than +71% over the Vega based 5900HS with DDR4-3200. It's a big improvement sure, it's just that there was a big hype about RDNA2. Give Vega 8 50% more units, shrink it to 6nm, add DDR5/LPDDR5 support and it can't be much worse than this.

1. No it wouldn't. Too few ROPs.

2. Add 50% more units and the GPU power alone would be approaching 60W under load. Without including CPU nor SoC power consumption.

3. You're heavily overestimating the performance boost from memory. You have been from the very beginning. You just refuse to listen.

 

[Markfw]

Meteor Lake is sometime in 2H 2023. Hopefully its iGPU is competent according to the rumors.

Who cares about meteor lake, this is a Zen 3 thread.

Anyway, I got my 6850U laptop today, but too busy to mess with it, with buying Zen 4 stuff and posting about it, and deal with trolls like tamz_msc.

Will post about it later. Anything specifically you want to know ?

 

[Spicy]

The chips that are in production are mobile variants. There is a TINY chance we could see a desktop release, but I am expecting, barring leaked information to the contrary, to not see a desktop variant until at least 12 months from the release of Zen 3 APUs.

I think Rembrandt will be lauchned on the AM5 platform sooner.
3 points:
1. Zen 4 (Raphael) is expected to be launched Q4 2022. Usually, AMD releases its APU version (Phoenix G) 2-3 quarters later. So, Rembrandt G should be launched prior to Raphael, to have a decent life time. I bet Q2 2022 (thanks to point 2).
2. Some rumors predict AM5 motherboards will be released Q2 2022. Which CPU will we be able to buy with this MB? Raphael? Not available. Rembrandt G? I think that.

Intel Z690 'Alder Lake' Motherboards Expected in Q4 2021, Z790 'Raptor Lake' In Q3 2022, AMD AM5 'Ryzen' Motherboards In Q2 2022

Expected launch dates of next-generation AMD Ryzen AM5 and Intel Z690 Alder Lake / Z790 Raptor Lake platforms have leaked out.

wccftech.com

3. A leaked document reveals that the AM5 platform will begin its career with 3 processor types: models 40h, models 60h and models 70h

AMD Socket AM5 "Zen 4" Processors to have RDNA2 Integrated Graphics Across the Lineup

The first desktop processors to implement AMD's "Zen 4" microarchitecture will feature integrated graphics as standard across the lineup, according to a Chips and Cheese report citing leaked AMD design documents. Currently, most of the Socket AM4 desktop processor lineup lacks integrated...

www.techpowerup.com

- Models 40h have 20 pcie lines and a USB 4 controller.
Rembrandt is known for having 20 PCIe 4.0 lines and an integreted USB 4 controller.

AMD Rembrandt to feature (6nm?) Zen3 cores, support DDR5-5200 memory - VideoCardz.com

First leaks on AMD’s next-generation APUs. AMD Rembrandt; 6nm Zen3 cores? Further speculation on AMD’s next-generation APUs codenamed Rembrand has emerged yesterday, after @patrickschur_ provided a new update to his previous leaks. AMD Rembrand is a codename for next-generation APU, planned for...

videocardz.com

So Type 1 / 40h is Rembrandt G (~Q2 2022).

- Models 60h have 28 pcies lines but no USB 4 port.
Raphael is known for having 28 PCIe 4.0 lines and an other leaked document reveals that 4 of them can be connected to an external USB 4 controler, deported on the motherboard (so, not integrated to the processor).

Mockup of AMD's Zen 4 Raphael CPU Appears With Potential Specs

New era for AMD CPUs

www.tomshardware.com

No PCIe Gen5 for "Raphael," Says Gigabyte's Leaked Socket AM5 Documentation

AMD might fall behind Intel on PCI-Express Gen 5 support, say sources familiar with the recent GIGABYTE ransomware attack and ensuing leak of confidential documents. If you recall, AMD had extensively marketed the fact that it was first-to-market with PCI-Express Gen 4, over a year ahead of...

www.techpowerup.com

So, Type 2 / 60h is Raphael (~Q4 2022)

- Models 70h again have 20 pcie lines and USB 4 ports. This is typically an other G APU.
So, Type 3 / 70h should be Phoenix (~H1 2023)

I bet Q1 2022: 6000U/H mobile versions, Q2 2022: 6000G Pro/OEM versions, Q3 2022: 6000G retail versions (like Cezanne 2021 roadmap).

The next refresh Zen 3D will be named 6000X, Raphael: 7000X and Phoenix 7000U/H/G.

Raphael will have an integreted Graphic Processor, but it will not confront Rembrandt G/Phoenix G, because its iGP will be very tiny.
Here we know that Raphael's iGP will be integreted to the IODie and looks small (while the size on the picture is not representative of the reality). So it will not be a dedicated RDNA chiplet.

Old AMD Ryzen 'Raphael' Zen 4 Desktop CPU & AM5 Platform Slides Leaked, Show Just How Much Has Changed Since 2020

In a new video published by Gamers Nexus, slides regarding AMD's Zen 4 powered Ryzen 'Raphael' and the AM5 platform have leaked out.

wccftech.com

And here, we know that AMD is working on a very small RDNA 1 iGP (without RT).

Radeon "Cyan Skillfish" Navi 1x APU Submitted For Linux 5.15 Plus Many Fixes - Phoronix

www.phoronix.com

I bet this is what we will find on 7000X Zen 4 Raphael. Enough for home office, web and movies. For gaming / 3D? Buy a G APU or a discret GPU.

(Of course, this is only my personal feeling/analysis. May be wrong!)

 

[TESKATLIPOKA]

You're looking at about 80GB/sec in bandwidth with worse latency numbers than the Vega 8 APUs and no infinity cache, and I'm seeing people expecting better performance than dGPUs with over 100GB/sec of bandwidth.

That is overly optimistic.

Really? Let's see.

Link

Ryzen 7 5700G specs: 8C16T 3.8/4.6GHz 65W TDP ; IGP: 2000Mhz 8CU(512SP:32TMU:8ROPS?) (2048 GFLOPs)
Memory: 2x 8GB DDR4-3200 (dual channel) memory @ 16-16-16-36 -> 51.2 GB/s for both CPU+IGP.

RX 560 specs: 16CU(1024SP:64TMU:16ROPs); Frequency: 1168(1244) MHz -> 2611GFLOPs; BW: 112GB/s; TDP: 80W

RX 560 has 119% higher BW just for Itself, TDP is also higher but is for the whole card, GFLOPs is 27.5% higher, Pixel Fillrate(ROPs) is most likely ~2.5x higher, and you gain only 33-35% better performance. It doesn't look like the limited bandwidth is negatively affecting the performance to me, maybe by just a few %.

dGPUs vs 5700g IGP:
RX 560 is 35% faster and 33% faster at 99th percentile.
GTX 1050 is 61% faster and 48% faster at 99th percentile.
GTX 1050 Ti is 79% faster and 68% faster at 99th percentile.
GTX 1650 is 128% faster and 115% faster at 99th percentile.

Now what about Rembrandt?
Example: 8C16T Zen3+ 65W; 12CU(768SP, 48TMU, 16ROPs), I will keep It at 2GHz so that would mean 3072 GFLOPs.
Memory: DDR5 4800Mhz -> 76.8 GB/s.
Just this alone would mean 50% higher GPU performance.

Now let's calculate IPC between Polaris and RDNA2.
RX 6600XT review
RX 580(36CU;2304SP:144TMU:32ROPs; 256bit 8GHz GDDR5; 185W TBP) vs RX 6600XT(32CU; 2048Sp:128TMU:64ROPs; 128bit 16GHz GDDR6 + 32MB IF; 160W)
Performance: 48 vs 99 at Full HD (+106%)
GFLOPs: 6175(1340MHz) vs 10,605(2589MHz) ( +72%)
Bandwidth: 256 GB/s vs 256 GB/s + 32MB IF
I got 20% better IPC.

Let's calculate performance over 5700g IGP: 100*1.5*1.2 = 180%.
Now It would look like this:
Rembrandt IGP is 33-35% faster than RX 560.
Rembrandt IGP is 12-22% faster than GTX1050.
Rembrandt IGP is 1-7% faster than GTX1050 Ti.
GTX 1650 is 19-27% faster than Rembrandt IGP.
I got this with only 4800MHz DDR5 and 2GHz IGP. What would happen If we use 6400Mhz(+33% BW) and 2600MHz clocks(+30%)?

 

[moinmoin]

I thought AMD learnt the lesson with cezanne and U models but I was wrong.

At least Barcelo is not part of the 6000 series so one can't accidentally get a last gen chip looking at the model number unlike with Lucienne.

 

[nicalandia]

I doubt there is any significant IPC increase. 6800U has 4.7Ghz vs 4.4Ghz (5800U) for ST boost , that's ~7% higher clock for Cinebench 1T

There is a clear and substantial IPC increase even with half of the MB less than the Desktop 5800

 

[Hans de Vries]

The interesting take-away is that: Even when operating with a higher 28W max TDP, the 6800U has a longer battery live as the 15W TDP 5800U in a general use case. With all those new power management features, adaptive power management framework and new sleep states it requires less watts over time in the general use case.

 

[leoneazzurro]

https://twitter.com/x/status/1483837240212168704

Apparently, Rembrandt takes quite a performance hit at 15W. However, I think people are still misunderstanding why Rembrandt was tested at 28W to begin with. Apple started this trend. Their current laptop chips are 30W. AMD and Intel are just changing things up to match. Cores are getting too big to run at peak performance when capped to 15W. Capping to 15W means sacrificing single core performance. Battery and cooling tech has advanced far enough that running this chips in thin/light laptops should rarely be an issue.

He has no Rembrandt to test, he tested Cezanne and he found that "there is a substantial hit" at 15W vs 25W but he did not specify how much, the only testing he spoke about was Rainbow Six which was 66FPS instead of 74FPS. Which means a 12% more on Cezanne going from 15W to 25W. Which also means that instead of 2x we would get a 1,78x comparing a 28W Rembrandt to a 25W Cezanne which is not a +100'% increase but still damn good at practically same power level and probably better bettery life. Techepifany lately is all AMD bashing, often without much sense.

 

[tamz_msc]

Total non sense, if Intel chips perf tank less in battery mode it means that it consume a lot more power than RMB, what matters is the total throughput over battery life time, or do you mean that running half the time at 10% higher perf is the better behaviour..?.

It does not use a "lot more" power to give the same performance plugged vs unplugged. I want the same CPU performance for light, bursty loads whether the laptop is plugged or unplugged. This was not possible on earlier CPUs from AMD. For example Renoir dropped 40% performance in Geekbench when unplugged. The only thing that changes on Intel when you plug it in is PL1, PL2 remains the same. So boosting behaviour remains the same.

 

[tamz_msc]

This worries me a little actually. The way this worked on Renoir and Cezanne is that short bursts to inefficient frequencies are avoided on battery unless the workload takes a longer time. (At least on Renoir this behavior is changeable by the user.) If Rembrandt changes that I hope it doesn't do so while using more energy for short bursts.

The 10 s delay to boost that was present on Renoir (dunno if it's also there on Cezanne) was a big reason for sluggishness on battery power.

 

[uzzi38]

Yes, I know, this is Van Gogh, not Rembrandt. But look at the comparisons vs the 4900HS here and tell me again @mikk and tell me with a straight face that scaling up Vega is a good choice.

 

[izaic3]

AMD Ryzen 7 6800U Efficiency Review - Zen3+ beats Intel Alder Lake

We had to wait quite a while but we finally got our hands on AMD’s latest mobile processors from the Ryzen 6000U generation. We compare the Zen3+ cores of the new Ryzen 7 6800U with Intel’s current Alder Lake-P platform as well as Apple’s M1 Pro SoCs. Who builds the best chip for compact laptops?

www.notebookcheck.net

This talks about the efficiency of the 6800u.

"The efficiency advantage of the new Ryzen 7 6800U over the current Alder Lake models is much bigger in multi-core tests and contrary to the single-core tests, there is no real performance advantage anymore, either. Even the Core i7-1260P with a much higher power consumption is just slightly ahead in the benchmarks and our results once again show how inefficient Alder Lake is at higher consumption values. The Core i5-1240P in the Yoga Slim 7 Pro consumes much more power than the Core i7-1260P in the Yoga 9i 14, but this is hardly noticeable in the performance.

It is also interesting to see how Alder Lake-P chips perform at lower power levels when the cooling performance is limited. This is the case for the Samsung Galaxy Book 2 Pro 13, where the CPU is quickly limited to just 20W. The efficiency increase is very small compared to the gains we saw for the Ryzen 7 6800U at 12W, which means Intel’s hybrid architecture with the 10 nm cores is not really suited for very compact systems. These are not good news for the upcoming Alder Lake-U CPUs, which are positioned in the 15W class. However, Intel still lists a maximum consumption of 55W, even though these chips are only equipped with 2 P-cores. Based on our results so far, we do not believe the Alder Lake-U systems will be very competitive, neither in terms of performance, nor efficiency.

The new AMD Ryzen 7 6800U leaves a very good impression so far. The single-core performance is not as good as on the new Alder Lake chips, but the multi-core performance is comparable and the new Ryzen CPU is way more efficient. AMD also managed to improve the integrated GPU, but we will take a closer look at the new RDNA2 GPU in a separate article.

We expect more laptops with Ryzen 6000U CPUs in the next couple of weeks and the real rivals for the 6000U models, the Alder Lake-U chips like the Core i7-1265U, should be available in a few weeks as well. They use a similar hybrid architecture as other Alder Lake chips, but with fewer P-cores. Intel still lists a maximum consumption of 55W, which worries us a bit. The high consumption values do not only result in shorter battery runtimes (even with simple workloads), but we also observed more frequent fan activity in our reviews so far.

Based on our results so far, the AMD Ryzen 6000U chips are the best mobile processors for Windows laptops right now. The big question is if manufacturers will use the AMD chips for their premium laptops and if they will be available in larger quantities, where Intel has an advantage right now."

 

[Abwx]

English version of NBC 6800U review with some interesting numbers, 1600 pts in CB R15 at about 27W and 1040pts at 12W.

According to the reviewer the competing 1260P require 38W, 40% more power, to perform the same score than RMB.

https://www.notebookcheck.net/Asus-Zenbook-S-13-OLED-laptop-review-Subnotebook-impresses-with-fast-AMD-Ryzen-7-6800U.633487.0.html

 

[tamz_msc]

That is entirely OEM-configurable and even done with some Alder Lake laptops recently as well.

PS: Rembrandt doesn't need it to win on battery life, just FYI.

Intel laptops do not ignore Windows performance settings. AMD laptops do, and that behavior was arguably worse in the past generations.

That's almost a userbenchmark kind of comment. Did they comment anything on the Intel way of doing boost way past the CPU power rating ?

There is no "boosting way past the CPU power rating" with Intel. There's PL1 and PL2. Once you fix the power profile, and provided the thermal design is adequate, only PL1 changes depending on whether you are on battery or AC. PL2 doesn't. What the Windows performance plan changes is the responsiveness of the boost behavior. On AMD while on battery, the latter is slow even if you set it to best performance mode, or at least it used to be.

This is not something you will see in Cinebench runs, because those kind of tests always fall back to PL1 on laptops. You will notice it on browser benchmarks like Speedometer 2.0, Office stuff, light compilation/light compute like MATLAB or Mathematica, and Geekbench.

AMD laptops having better battery life - that's the reason why - the experience on battery is disparate from the experience on AC.

 

[Markfw]

It gets better... AFTER I cancelled it, it still says order delayed, but now it will be here Next Tuesday ,, as I got a UPS ship notice ! Before the one from Micro center ! So now I will have 2, to play with and decide.

Edit: I just looked. at 5:02 pm I got an email saying it was delayed. at 8:01 pm I got this "Hi Mark, Your package is on the way! From LENOVO Scheduled Delivery Monday 09/26/2022 10:30 AM - 2:30 PM Track Your Package "

 

[ahimsa42]

laptops are the way to go imo. small & portable they also come with keyboard, monitor, camera & microphone. i went from desktop to laptop only starting with the 200U and have not looked back. now if only i were able to find the 6800U model i want i'll be set for 3+ years until strix point.

 

[DrMrLordX]

Actually, I've been looking at the older models those, since I have those accessories. I might consider that when it comes out. Thanks for letting me know about this one that uses Rembrandt!

If it's in your budget, go for it. The 6900HX one looks very nice.

Also while I can't vouch for these guys, I saw them advertised by Google shopping:

Morefine M600 Mini PC AMD R9-6900HX

- MOREFINE

morefine.com

These units don't seem to be shipping yet.

 

[Kaluan]

A first tangible design for refreshed or rebranded mobile Rembrandt SKUs:

ASUS ROG Zephyrus G14 with AMD Ryzen 7000HS CPU has been pictured as well - VideoCardz.com

The first gaming laptop with Ryzen 7000HS? ASUS ROG Zephyrus G14 a Ryzen 7000 treatment. ASUS new 14-inch gaming laptop is most likely set to be revealed soon. The first photograph of this new product has just been discovered at South Korean Energy Agency, as the laptop goes through the...

videocardz.com

Likely called Ryzen 7735H/HS, so similar or identical to 6800H/HS. Also seems to come with a Radeon RX 7000S dGPU. Dunno if Navi32 or 33 based.

 

[izaic3]

Just picked up a Thinkpad P16s with the 6850u. Let me know if you have any questions.

 

[tamz_msc]

Just picked up a Thinkpad P16s with the 6850u. Let me know if you have any questions.

Well you can do the honors by testing the performance delta on battery and plugged in. Set Best Performance in Power Mode in Settings -> System -> Power & battery. Keep everything else on default. Run Geekbench 5 and Speedometer 2.0. Report the Geekbench ST score and Speedometer result when on battery and when plugged in.