Discussion AMD Cezanne/Zen 3 APU Speculation and Discussion

[dr1337]

https://twitter.com/x/status/1292844896374718470

Fresh leak out today, not much is known but at least 8cu's is confirmed. Probably an engineering sample, core count is unknown and clocks may not be final.

This is very interesting to me because cezanne is seemingly 8cu only, and it seems unlikely to me that AMD could squeeze any more performance out of vega. A cpu only upgrade of renoir may be lackluster compared to tigerlake's quite large GPU.

What do you guys think? Will zen 3 be a large enough improvement in APU form? Will it have full cache? Are there more than 8cus? Has AMD truly evolved vega yet again or is it more like rdna?

 

[uzzi38]

Those PR renders are ever only ever... haphazardly accurate... not just how they're filled in, but even to the point that the overall shape of the die is wrong. For example, the real Renoir die is taller than the render, so even though 10% sounds reasonable, it's not something I would ever advise extrapolating from these renders. That said, the Cezanne render looks significantly more grounded in reality than the Renoir render. What the heck is up with those boxes that comprise the Zen 2 cores? So that's at least a step in the right direction even though the Cezanne render adds these weird overlapping boxes inside each L3 area.

And actually, the cache area is pretty much a mess in general. Taken literally, Cezanne's L3 takes up just about as much die area relative to the Zen 3 cores as Vermeer's does, like it has 7/8ths the L3 instead of 1/2.

Seriously, whoever at AMD decided to move away from beautiful and mesmerizing die shots to half-assed generic renders... you did bad. It's a huge step backwards for marketing and PR, and it's not as if you're actually improving secrecy all that much either.

Those boxes that comprise the Zen 2 cores are actually RDNA WGPs.

...don't ask

 

[Zepp]

Seriously, whoever at AMD decided to move away from beautiful and mesmerizing die shots to half-assed generic renders... you did bad. It's a huge step backwards for marketing and PR, and it's not as if you're actually improving secrecy all that much either.

100%

I'm still hoping for someone to make a nice photogenic color shot of renoir

this is all I've seen so far https://www.flickr.com/photos/130561288@N04/albums/72157715067913276 I don't know why his renoir shots are black and white

 

[jpiniero]

Lenovo announces IdeaPad 5 series with Qualcomm, AMD or Intel CPUs - VideoCardz.com

All New Connected Consumer Experiences Start with Lenovo Introducing IdeaPad 5G, Lenovo Tab P11, plus Amazon Alexa Show Mode debuts on select Lenovo PCs Digital transformation is now more personal than ever during the new normal – renewing our collective desire for levity and to get more things...

videocardz.com

Now this is funny, Lenovo is only offering Ampere on the 16" AMD model.

 

[Panino Manino]

CES is tomorrow already!

 

[YAYgee]

Desktop APUs didn't get a mention. Over 18 months since 3400G/3200G were launched. There should be something on the horizon by now, at the very least, especially as AMD promised that DIY could get their hands on Renoir soon (but that was already 6 months ago).

According to Hallock display and video blocks have remained unchanged going from Renoir to Cezanne. Not a surprise, still not pleasant either.

 

[uzzi38]

So I figured I'd best leave this tidbit about the battery life claims here.

Also on the PCWorld stream Robert said one more thing about the configuration of the system - the display was set to 200-250 nit brightness. Somewhere in-between, he forgot the number exactly.

 

[moinmoin]

So I figured I'd best leave this tidbit about the battery life claims here.

Also on the PCWorld stream Robert said one more thing about the configuration of the system - the display was set to 200-250 nit brightness. Somewhere in-between, he forgot the number exactly.

Odd that AMD hid the comparison with 4800U in the disclaimer. So according to that with 5800U battery holds ~10% longer in video playback and ~27% longer in Mobilemark 2018.

 

[Thibsie]

Maybe because part of the 5xxx lineup is still zen2 ?

 

[movdx]

I find it amazing that Dell for one more year chose to ignore ryzen for their professional laptops

 

[uzzi38]

Odd that AMD hid the comparison with 4800U in the disclaimer. So according to that with 5800U battery holds ~10% longer in video playback and ~27% longer in Mobilemark 2018.

Probably because people would point out that 5700U = 4800U, so there's going to be a noticable difference in battery life between the LCN -U parts and CZN -U parts.

 

[coercitiv]

Lisa Su called battery life for 5000 series as "a major jump compared to our previous generation", so if they did hid the information, they did so in plain sight.

Also, I would not be quick to attribute improved battery life to the Zen 3 silicon alone. Laptop CPUs can get a nice bump in battery life from multiple sources, not all of them being directly linked to core performance. LCN-U may still surprise us, as far as a Zen 2 based SKU can anyway.

 

[LightningZ71]

Lucienne could hold a few surprises, after all, it is an actual revision at the silicon level and not just a direct rebadge.

 

[jpiniero]

I find it amazing that Dell for one more year chose to ignore ryzen for their professional laptops

Tiger Lake is probally good enough for the business market. Plus it's Dell.. they are BFF with Intel.

 

[moinmoin]

Probably because people would point out that 5700U = 4800U, so there's going to be a noticable difference in battery life between the LCN -U parts and CZN -U parts.

Video playback (which improved ~10% in CZN-U) should be down to improvements in VCN, unless it's some improvements of the system baseline (like XT-like improvements to the node, selection of components for the platform etc.), from which LCN-U may well also profit.

Also, I would not be quick to attribute improved battery life to the Zen 3 silicon alone. Laptop CPUs can get a nice bump in battery life from multiple sources, not all of them being directly linked to core performance. LCN-U may still surprise us, as far as a Zen 2 based SKU can anyway.

Lucienne could hold a few surprises, after all, it is an actual revision at the silicon level and not just a direct rebadge.

Indeed. If nothing else the LCN-U parts will show which improvements this APU gen are specific to Zen 3 and which are more generic thus usable by LCN-U based systems as well.

 

[LightningZ71]

Well, just from the numbers, the boost clocks for the CPU and GPU are up a bit between the 4800U and the 5700U, so, power and thermals allowing, it will be measurably faster, though it probably won't be noticeable.

I just wonder if the MT numbers for the mobile 5800s will improve from what we're seeing in the leaked benches. It seems that, for whatever reason, including things like Beta Bioses, the MT scores for Cezanne aren't much better than Renoir, and even show regressions in a few areas. I'm hoping that this is just teething issues, though, I wouldn't be shocked if the scales were tipped a bit in favor of ST performance as opposed to MT scores.

 

[moinmoin]

I just wonder if the MT numbers for the mobile 5800s will improve from what we're seeing in the leaked benches. It seems that, for whatever reason, including things like Beta Bioses, the MT scores for Cezanne aren't much better than Renoir, and even show regressions in a few areas.

Remember that Vermeer's clock (particularly 5950X at more than 14 cores) already regressed due to higher max power consumption within the same TDP/PPT envelope on the same node. So MT on mobile systems also being affected by this is rather natural I'd think. What will be interesting is the extend of it.

 

[randomhero]

Remember that Vermeer's clock (particularly 5950X at more than 14 cores) already regressed due to higher max power consumption within the same TDP/PPT envelope on the same node. So MT on mobile systems also being affected by this is rather natural I'd think. What will be interesting is the extend of it.

But they compensated that with substantially higher ppc.
Also, I am of opinion that high core parts are more limited by bandwidth than power.
Also tsmc processes are tilted towards mobile, read Apple, so I am not sure it's that straightforward.
Just my 2c.

 

[Shivansps]

I find it amazing that Dell for one more year chose to ignore ryzen for their professional laptops

Remember that is not only about performance, but availity as well.

 

[LightningZ71]

But they compensated that with substantially higher ppc.
Also, I am of opinion that high core parts are more limited by bandwidth than power.
Also tsmc processes are tilted towards mobile, read Apple, so I am not sure it's that straightforward.
Just my 2c.

Unfortunately, and again, this is from the leaks that MAY be based on beta bioses, etc., there appears to be regressions in the ST results on more than one of the sub-component tests from the Geekbench results. Now, this doesn't mean that it is a bad processor, just that the optimizations that were made to increase IPC in Zen3 may have a few cases where they hurt certain instruction sequences. Its still an overall improvement, but, in certain cases, it's less than ideal. As always, if you have a specific task that requires best performance, you should look for specific benchmarks of that task by sites and others.

 

[Asterox]

Zen 2/Renoir APU has only 8mb L3 Cache, compared to Zen 3 /Cezanne APU with 16mb L3 Cache.Desktop Zen 3/5000 still has 32mb L3 Cache, but Cezanne 16mb L3 Cache "it is much more comfortable."

AMD Ryzen 7 5700G 8 Core Cezanne APU Performance Leaks Out, Overclocked To 4.75 GHz & Faster Than Ryzen 7 5800X

The AMD Ryzen 7 5700G 8 Core Cezanne APU featuring the Zen 3 core architecture has been overclocked and benchmarked over at Chiphell.

wccftech.com

 

[Racan]

AMD Ryzen 9 5900HX is now PassMark’s highest ranking mobile CPU

AMD Ryzen 9 5900HX is now PassMark's highest ranking mobile CPU - VideoCardz.com

AMD Ryzen 9 5900HX Cezanne APU The Ryzen 9 5900HX is not even the fastest CPU of the Cezanne FP6 series, yet it was enough to top the PassMark mobile CPU ranking in both single and multi-threaded benchmarks. AMD introduced its first Zen3 based mobile CPUs aka Ryzen 5000H series at CES 2021. The...

videocardz.com

Both in single thread and multi

 

[Gideon]

Zen 3 APU slides leaked:

AMD Ryzen 5000 Mobile Zen3 architecture detailed - VideoCardz.com

AMD Ryzen 5000 Cezanne (Zen3) Today we are sharing AMD Ryzen 5000 mobile series architecture details. AMD has confirmed that Cezanne APU has a die size of 180mm2, a 15% increase over Renoir. The APU has 10.78 billion transistors, which is a 10% increase over its predecessor. In a new slide deck...

videocardz.com

Actually some very decent battery life improvements on top of things already known.

 

[TESKATLIPOKA]

5800U IGP up to 2.1GHz, but you gain only ~10% in TimeSpy despite boosting 20% higher. Probably the average clockspeed is either only 10% higher or bandwidth is a bottleneck.

 

[misuspita]

Just guessing. Limited by bandwidth? Maybe Cezanne really needs better ram speeds

 

[HurleyBird]

Would love to see some LPDDR4x benchmarks with the iGPU, not that I expect many notebooks to go that route.

I don't have any evidence on hand, but I'm skeptical to say the least that all CPU cores and the iGPU were locked to run at the same voltage plane in Renoir like the slide suggests. Did something get lost in translation by the time it arrived in the hands of the slide deck designer? Like the iGPU voltage being made separate to "SoC voltage", and the slide designer thinks "Well, the entire chip is an SoC, so I guess everything was running at the same voltage all the time!"