AMD is preparing Ryzen 5000 APUs for laptops, but not all will be Zen 3-based

mongeese

Posts: 643   +123
Staff
Something to look forward to: AMD kicked off an excellent year of tech with the release of Renoir, the Ryzen 4000 series mobile APUs. Now their replacements are due: but lightning rarely strikes twice, and it appears that AMD isn't ready for another revolution.

The new Ryzen mobile series is suspected to have a mix of Zen 2 and Zen 3 parts, according to different leaks as shown in the table below (courtesy of NotebookCheck).

Code-named Lucienne (pronounced lucy-en), a feminine French name meaning 'light,' when the processors release, they'll join AMD's desktop parts in going by the Ryzen 5000 series nomenclature. But Lucienne parts won't use the new Zen 3 architecture, instead, they'll use Zen 2 and stick to the same Vega graphics.

It is suspected Lucienne will be offered in budget and mainstream laptops, while the 'Cezanne' Ryzen 5000 APUs will get the upgrade to Zen 3 and offer bigger boosts in performance, especially in single-thread operation, with an expected price premium to go along with it.

CPU Architecture Cores / Threads Base / Boost
Clock (GHz)
L3 Cache
(MB)
GPU Cores GPU Clock
(GHz)
Ryzen 3 5300U Zen 2 (Lucienne) 4 / 8 2.6 / 3.85 4 6 1.5
Ryzen 3 5400U Zen 3 (Cezanne) 4 / 8 2.6 / 4.0 8 6 1.6
Ryzen 5 5500U Zen 2 (Lucienne) 6 / 12 2.1 / 4.0 8 7 1.8
Ryzen 5 5600U Zen 3 (Cezanne) 6 / 12 2.3 / 4.2 12 7 1.8
Ryzen 7 5700U Zen 2 (Lucienne) 8 / 16 1.8 / 4.3 8 8 1.9
Ryzen 7 5800U Zen 3 (Cezanne) 8 / 16 2.0 / 4.4 16 8 2.0

The Ryzen 7 5700U appeared in the Geekbench 5 database this week several times with the CPU and GPU components getting tested. The CPU was revealed to have 8 cores, 16 threads, and operate with a base clock of 1.8 GHz and a maximum boost clock of 4.33 GHz.

For context, the 4800U also has 8 cores, 16 threads and runs a 1.8 GHz base clock and a 4.2 GHz boost clock. And if you'll forgive me for ignoring the boring numbers, we'll skip to the context on the performance: the 5700U got more or less the same benchmark scores that the 4800U gets (bear in mind that mobile scores are very variable).

The GPU inside the 5700U has eight compute units, just like the 4800U, but at 1.9 GHz, they were running 150 MHz faster than the 4800U's. There's no meaningful improvement in the scores, though.

Just recently, the Ryzen 5 5500U joined its faster sibling in the database, albeit with just one entry. This processor has 6 cores, 12 threads and runs a 2.1 GHz base clock, and can turbo up to at least 4.04 GHz. Familiar? The 4500U has a 2.3 GHz base clock and a 4.0 GHz boost clock, with the same core configuration. Both the 4500U and the 5500U get similar scores, too.

In short, AMD is refreshing Renoir and giving it a 5000 name. A refresh means that Lucienne will be a little bit better than Renoir, but probably not upgrade worthy. Cezanne, on the other hand, should be more interesting and offer more performance. The processors may see a release soon, too, possibly January.

Permalink to story.

 
Really AMD...skipped the 4000 series for Zen 3 to align mobile and desktop naming and now they release a Zen 2 core based APU as....Ryzen 5000.

Even if this was potentially due to OEM pressure, I don‘t like this and find it misleading if true as the difference between a 5500U and 5600U should be much greater than the naming implies.

 
For every new Zen architecture, Threadripper came 5 months later and mobile at least 10 months later. Having so early rumors about mobile makes me wonder if this time it will be different. It would benefit AMD a lot to switch it up and release mobile first, where they actually have some competition from Intel.
 
Do you mean the 4600u where you are comparing the 4500u to the 5500u because the 4500u is a 6 core 6 thread cpu I thought.
 
Why does AMD use Vega and not use RDNA in their APUs? Is RDNA not suitable for an APU or is there a different reason?
 
Really AMD...skipped the 4000 series for Zen 3 to align mobile and desktop naming and now they release a Zen 2 core based APU as....Ryzen 5000.

Even if this was potentially due to OEM pressure, I don‘t like this and find it misleading if true as the difference between a 5500U and 5600U should be much greater than the naming implies.
If performance is there why should anyone care if it's zen2 or zen3 based?
 
No 1660AF
No 3300X
No Renoir
No Radeon 7
No Big Navi
And Zen 3 desktop scraps.

AMD needs consistency.
This isn't even close.
 
If performance is there why should anyone care if it's zen2 or zen3 based?
Because there is a potential 20% difference as well as lower power usage with zen 3.

Amazingly, I know, people don't want to pay current gen rpcies for last gen performance.

Amd should have just skipped the odd numbers altogether. What is the point of having zen 2? They come from the same foundry ffs.
 
Because there is a potential 20% difference as well as lower power usage with zen 3.

Amazingly, I know, people don't want to pay current gen rpcies for last gen performance.

Amd should have just skipped the odd numbers altogether. What is the point of having zen 2? They come from the same foundry ffs.

But what 'gen' a processor belongs to is completely arbitrary. What matters is the performance/$, and given Zen3 has a ~16% larger die than Zen2 (https://wccftech.com/amd-ryzen-5000...gh-res-die-shots-close-ups-pictured-detailed/) it is highly likely Zen3-based Cezanne will be larger than Renoir and therefore more costly to manufacture. So it makes sense that AMD separate Cezanne to the higher tier models and leave Renoir at the lower tier models.
 
I don't like it when they do these rebadging sessions. AMD is guilty of this on the GPU side as well, along with nVidia. So you can no longer trust the model of the device and you have to do a deep dive into the specifications to find out if you're actually getting what you're hoping for.
 
If AMD is indeed doing this, that's incredibly confusing for customers, considering that, for example, a 5600U is likely to be better at most tasks than a 5700U.
 
...code-named Lucienne, a feminine French name meaning 'light,'...
Lucienne was the illegitimate daughter of Renoir. I wonder if AMD is trying to intimate something scandalous about the chip's provenance?

It would benefit AMD a lot to switch it up and release mobile first, where they actually have some competition from Intel.
That's what they did with the 4000 series. The mobile chips have been out a while, but no desktop parts yet to be seen.

people don't want to pay current gen rpcies for last gen performance....there is a potential 20% difference as well as lower power usage with zen 3.
That figure comes from benchmarks on specific pre-existing cpus. A new 5000-series cpu will be a different chip, and whether its Zen 2 or Zen 3, the performance and efficiency could be higher or lower than that 20% figure. Moral of the story: judge each chip by its own benchmarks, not what family AMD assigns it to.
 
Why does AMD use Vega and not use RDNA in their APUs? Is RDNA not suitable for an APU or is there a different reason?
When comparing like-for-like components, layouts using the Vega architecture are a little smaller than RDNA (although not as dense). For example, the Vega 20 chip in the Radeon VII has a die area of 331 mm2, whereas as the Navi 10 chip in the Radeon RX 5800 XT has an area of 251 mm2 - but the former has 60 Compute Units to the latter's 40. So while 331 mm2 is 32% larger than 251 mm2, 60 Navi CUs would take up 50% more space than 40 CUs -- this doesn't mean that a 60 CU Navi chip would 376 mm2, but it does explain why the Vega 20 chip is so much bigger (and the 4096-bit wide HBM2 memory system also requires more die space too, compared to a 256-bit wide GDDR6 system).

And for APUs, an RDNA GPU isn't going to be magically a lot better than a Vega one, as we're only talking about a processor that's sporting just 8 CUs and using system memory. Yes, RDNA is a lot more efficient than Vega (as clearly seen by comparing the performance of the RX 5800 XT to the Radeon VII), it's not going to make much difference when you're entirely ROP and bandwidth limited.

That all said, I suspect the real reason why AMD have stuck with Vega in their APUs is that they only have so much development resources, in their Compute and Graphics divisions. Perhaps the APU designers have been under tighter deadlines, smaller budgets, etc than the teams working on the chiplet setups.

The Xbox Series X/S and PlayStation 5 are using a Zen 2 + RDNA 2 monolithic chip, of course, and now that design is complete, we may perhaps see the next round of APUs, sporting Zen 3, using RDNA 2 as well. I think AMD may skip RNDA entirely for these products, given the density and power gains they've made with the new architecture.
 
That's what they did with the 4000 series. The mobile chips have been out a while, but no desktop parts yet to be seen.
I mean mobile before Threadripper for a specific architecture, not before Desktop APUs. That is something they have never done. Threadripper is so ahead of Intel's HEDT that it wouldn't matter much if it gets delayed this time. Meanwhile, if Zen 3 mobile could be ready for Q2 laptops, that would be an enormous hit on Intel.
 
When they skipped a number for Ryzen 5000 on the desktop, I hoped this meant an end to confusion. I see I was over-optimistic, if they're not going to reserve 5000 numbers strictly for Zen 3 chips whether on desktop or mobile.
 
Just do your research when you buy anything, if you don’t and you get burned that’s on you not the company who sold it.
 
If performance is there why should anyone care if it's zen2 or zen3 based?
For technical follow up. Even a non technical person who wants to buy one of these products needs to be able to follow the differences without getting lost in the history of the sku naming. I agree with Irata, this will be annoying. I think they should have just released Lucienne on its own as a continuation of the zen 2 line without mixing it up with cezanne, as it is also going to dampen some of that shine that AMD has worked so hard to build.
 
Why does AMD use Vega and not use RDNA in their APUs? Is RDNA not suitable for an APU or is there a different reason?
From what I gather Vega has better compute performance over Navi, and the reason AMD created their APU product lines was to improve video encoding and other compute tasks. It's quite possible we will see Navi CU's in the next line of CPU's but I'm not sure.
 
If AMD is indeed doing this, that's incredibly confusing for customers, considering that, for example, a 5600U is likely to be better at most tasks than a 5700U.
Let's just wait on the benchmarks. The 4000 series mobile chips were very different from the desktop chips so I'm assuming that perf/$ scaling across the 5000 series will be ok.
 
"Let the buyer beware"

Lovely sentiment and undoubtedly a truism; however if I am ever nailed by deceptive advertising or marketing, I generally remove that vendor or manufacturer from consideration for future product purchases.

Doesn't matter much now but the days when I was building 8 computers a day for speech recognition clients, refusing to use certain brand of mainboard because of deceptive specs or advertising cost somebody a fair amount of money.

Do unto others..... .
 
Lovely sentiment and undoubtedly a truism; however if I am ever nailed by deceptive advertising or marketing, I generally remove that vendor or manufacturer from consideration...
The only thing a higher part number indicates is that the part is newer. AMD has never stated that any specific part # or series corresponds to a specific internal microarchitecture. If they take a four-year old CPU and rebrand it as "5000-series", then I'll call it deceptive marketing. Until then, they're not responsible for people's preconceptions.
 
Back