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]

These tuning variances between OEMs are exactly why I compared the laptop to its direct predecessor.

On mixed workloads the new G14 certainly doesn't seem to be an improvement in battery life vs the old G14.

1600p display vs 1080p display.

 

[uzzi38]

Well yeah, assuming the YT playback test is playing back a 4K video, it is most likely using AV1 decode.

It can't be hardware decode AV1 vs software decode. No chance the Cezanne laptops would even manage 6 hours in that case.

 

[moinmoin]

Hardware Unboxed host.

Ok, so it's from

He says "depending on how the OEM implements the ports AMD's USB4 ports will also support Thunderbold devices" and proceeds that a crucial feature is still awaiting certification. But don't the OEMs need the certification, not AMD?

Any idea what that crucial feature could be? Looking at the Intel table the possible missing features that may need certification by AMD itself would be Thunderbolt networking, wake from sleep, and "Intel VT-d based DMA protection".

 

[Asterox]

It can't be hardware decode AV1 vs software decode. No chance the Cezanne laptops would even manage 6 hours in that case.

If Youtube(defoult setings)robots detect AV1 decoding capable hardware, it will send AV1 video stream if available.But most Youtube videos still offer only VP9 video stream.

Page 31 - Zen 2 APUs/"Renoir" discussion thread

Page 31 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

VP9 or AV1 same story, Rembrandt APU has redesigned SoC +latest VCN hardware decoder.When everything is taken into account, with Rembrandt APU the battery should have the same duration regardless of Youtube VP9 or AV1 video playback or stream.

One important detail, AV1 video for same video quality needs significantly less bandwidth or bitrate.
My internet connection is not super fast, and Youtube AV1 4K/60 video stream(if availabile) suits me much better.It is Desktop PC not laptop, my Youtube video settings is Always prefer AV1.

- VP9 4K/60, hiccups and video playback buffering, Firefox or Chrome same story

- AV1 4K/60, video playback has no problems

 

[tamz_msc]

It can't be hardware decode AV1 vs software decode. No chance the Cezanne laptops would even manage 6 hours in that case.

If AV1 decode isn't present, YT falls back to VP9, and Vega on iGPU has full VP9 decode, though not as efficient as Intel I think.

 

[uzzi38]

If AV1 decode isn't present, YT falls back to VP9, and Vega on iGPU has full VP9 decode, though not as efficient as Intel I think.

That's probably what's going on then, and that makes a lot more sense.

Maybe this is a more controversial take then, but I think that's a very valid result then. This feels like an actual potential use-case. Of course it's not representative the battery life improvement in every regard, but it's still an area with a massive improvement.

 

[Shivansps]

If AV1 decode isn't present, YT falls back to VP9, and Vega on iGPU has full VP9 decode, though not as efficient as Intel I think.

That does not always happens. I had seem AV1 streams on non AV1 capable hardware.

 

[igor_kavinski]

Woah. Intel's in trouble.

EDIT: Ryzen 6000 Blew Me Away - YouTube

 

[igor_kavinski]

"Better" leakage? I think they meant to say "lower" leakage.

 

[BorisTheBlade82]

View attachment 57606

"Better" leakage? I think they meant to say "lower" leakage.

"Better" in this context obviously means "lower". "higher" as the opposite would be described as "worse".

 

[igor_kavinski]

AMD Ryzen 6000 "Rembrandt" Laptop Review Roundup | VideoCardz.com



10 hours battery life from a gaming laptop. Now that's progress.

 

[coercitiv]

"Better" leakage? I think they meant to say "lower" leakage.

What does "better performance" mean for you?

 

[igor_kavinski]

What does "better performance" mean for you?

Performancewise, better means higher. But leakagewise, it just seems weird to me that better would be taken to mean lower because I've never seen better and leakage used together before. Better leakage characteristics? That seems to make more sense.

 

[coercitiv]

Better leakage characteristics? That seems to make more sense.

What does "better leakage characteristics" mean in the following two contexts:

• a presentation about a low power mobile CPU
• a presentation about a CPU binned for overclocking

That's the beauty of "better", it allows you to guide your audience even if they don't necessarily understand the term. At the very least they know something is better.

 

[Thibsie]

What does "better performance" mean for you?

Still a weird choice of words.

 

[Mopetar]

It does seem weird because to the average person "leakage" is not a term with a positive connotation. Saying "lower leakage" (even though that may not always mean better in all circumstances) does make more sense from their perspective.

Take other words or phrases with a similar negative connotation and add "better" to see how weird it sounds. Better murder rates? That almost seems to be hiding a sinister element when compared with "lower murder rates". If you said "better molestations" (as opposed to fewer) to someone they're probably thinking about reporting you to the police.

Even though "better leakage" technically works, I think this is clearly as case where better is worse.

 

[moinmoin]

View attachment 57605

Woah. Intel's in trouble.

EDIT: Ryzen 6000 Blew Me Away - YouTube

See also

Page 2 - Question - Performanc per Area and Performance per Watt of Apple Firestorm vs Zen 3 vs Intel Golden Cove ?

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

and

Page 2 - Question - Performanc per Area and Performance per Watt of Apple Firestorm vs Zen 3 vs Intel Golden Cove ?

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

 

[Zucker2k]

See also

Page 2 - Question - Performanc per Area and Performance per Watt of Apple Firestorm vs Zen 3 vs Intel Golden Cove ?

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

and

Page 2 - Question - Performanc per Area and Performance per Watt of Apple Firestorm vs Zen 3 vs Intel Golden Cove ?

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

This performance per watt scaling is quite different from others we've seen in other reviews. Also, the plotting seems to be off as the 12900HK is the better performer at 70w.

 

[desrever]

This performance per watt scaling is quite different from others we've seen in other reviews. Also, the plotting seems to be off as the 12900HK is the better performer at 70w.

That graph shows the the 6900HS wins in efficiency under ~55W. It's total energy consumption for the test vs power limit so lower is better at any given limit. Intel wins at higher power which is expected.

 

[moinmoin]

Also, the plotting seems to be off as the 12900HK is the better performer at 70w.

Lower is better. 12900HK is faster at both 60W and 70W so uses less Joules for the Blender render accordingly.

 

[Zucker2k]

Yes, I see it's joules.

 

[DisEnchantment]

The SoC optimizations AMD did with RMB are very impressive. But not close to what is happening in mobile SoCs. But it is a great start.

I am excited to see what AMD will do with Zen 4 on N5.
If they continue this trend, the future looks good for x86 in laptops and handhelds.
2x efficiency on Zen4 plus new efficiency optimizations sounds so sweet for a steam deck successor or a 2023 laptop.
That efficiency knee right up to 4+ GHz on N5 means the competition will sweat quite a bit.
If they secured enough supply, they stand to take a good chunk of market share next year.

 

[uzzi38]

This performance per watt scaling is quite different from others we've seen in other reviews. Also, the plotting seems to be off as the 12900HK is the better performer at 70w.

What reviews are you taking as a basis? The reviews where the 12700H scores ~13k points at 45W like HUB's, or the reviews where it scores ~10.4k at 45W?

https://media.discordapp.net/attachments/682675911488700419/938888327401185350/v2-8806beed78e0f0c3a5104b68847b5297_r.png

How many times must I repeat that results are totally all over the place lmfao.

And before anyone thinks this laptop must be throttling or something, the default ootb configuration is 70W PL1 and 55W PL2.

There are just huge variations on a chip by chip basis, and that's normal for different laptops. Even within the exact same chassis you can have wildly different results: see Jarrod Tech's 12900HK vs HUB's 12700H.

 

[tamz_msc]

Chinese review of 6800H and 6600H is up:

AMD锐龙6000系列移动处理器评测：惊喜不止核显

由于时间关系只完成了用户最关心的iGPU部分，搭配独显的表现后面看有没有时间再测了。 目录介绍之前已经介绍过了，但应该有一些人对这一代全新的锐龙6000系不那么熟悉，所以就再简单过一遍。 本文内容仅包含标压的…

zhuanlan.zhihu.com

Biggest takeaway for me is that the performance drop on battery power is not as bad as previous gen. Still not as good as Intel's, but a step in the right direction.

Oh and the 12 CU iGPU is at MX450 level, which confirms my predictions.

 

[Abwx]

.Still not as good as Intel's, but a step in the right direction.

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..?.