Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)

Page 929 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Abwx

Lifer
Apr 2, 2011
11,885
4,873
136
I hate to break it to you but who is running CB on a handheld.
CB or anything else, if Integer perf is the point then the difference is even bigger than 60%, CB is just indicative of the huge discrepancy, i mean 60%+ is rouhgly 2 gen uplift, even the 10C 365 would make a killing, that is to say that quite costly laptops are way behing this gaming device when it comes to CPU perfs.
 
  • Like
Reactions: lightmanek

Hitman928

Diamond Member
Apr 15, 2012
6,695
12,370
136
CB or anything else, if Integer perf is the point then the difference is even bigger than 60%, CB is just indicative of the huge discrepancy, i mean 60%+ is rouhgly 2 gen uplift, even the 10C 365 would make a killing, that is to say that quite costly laptops are way behing this gaming device when it comes to CPU perfs.

nT performance doesn't really matter in that segment, that's the point. Single to lightly threaded performance and battery life are the key metrics.
 

Abwx

Lifer
Apr 2, 2011
11,885
4,873
136

Hitman928

Diamond Member
Apr 15, 2012
6,695
12,370
136
Better perfs mean longer life cycle, beside battery life is not a problem since years, it became allegedly important because that s the only selling point of a lacking competition, FI a 6850U laptop from 2022 with a 52.5Wh battery had almost 13h hours battery life in web surfing, at the time i heard no one saying that it was a major selling point.


I will reiterate, better nT performance is not meaningful in this space so your first comment doesn't apply.

Battery life for handhelds and ultra portables is absolutely a selling point, not sure why you think otherwise as it is one of the most advertised metrics for solutions in this space.

As far as your example, 13 hours battery life in 2022 was a good result but nothing unheard of, so that's probably why that particular model wasn't mentioned much, it was not a special result by the time it released.
 

Abwx

Lifer
Apr 2, 2011
11,885
4,873
136
13h with a 52.5Wh battery, a 72-75Wh one as is common with LNL would get you to 20h, exactly what is claimed a "breakthrough" for LNL.

And that was not common at the time, not even currently, dunno from where you came to this conclusion.
 

Hitman928

Diamond Member
Apr 15, 2012
6,695
12,370
136
13h with a 52.5Wh battery, a 72-75Wh one as is common with LNL would get you to 20h, exactly what is claimed a "breakthrough" for LNL.

And that was not common at the time, not even currently, dunno from where you came to this conclusion.

When did the m1 MacBook Air release? Even Intel solutions at the time could get really close to that.
 
  • Haha
Reactions: yottabit

Abwx

Lifer
Apr 2, 2011
11,885
4,873
136
When did the m1 MacBook Air release? Even Intel solutions at the time could get really close to that.
Now it s about Apple...

Intel was nowhere near those numbers, even a TGL, wich was their most efficient CPU, has only 8h battery life in the same test with a 50Wh battery, and that s with a 13.3" screen, so quite a good number but the 14" 6850U i linked has about 60% better battery life, numbers are numbers.

Notice that screens are both 1080p IPS and 60Hz, so the comparison is as fair as possible.

 

Kolifloro

Member
Mar 15, 2023
31
27
61
NBC tested a handheld with a Strix Point 370, 16500 pts in CB R23 at 28W, to compare to the much hyped LNL that barely manage 10 000pts at 32W.



That's why I previously stated my believe that Zen 6 seems to be determined to COMPETE (at performance per watt) with ARM laptops (included Apple's) ...

LNL laptops are already overthrown by Zen 5 roster ...

:)
 
  • Like
Reactions: perry mason

Hitman928

Diamond Member
Apr 15, 2012
6,695
12,370
136
Now it s about Apple...

Intel was nowhere near those numbers, even a TGL, wich was their most efficient CPU, has only 8h battery life in the same test with a 50Wh battery, and that s with a 13.3" screen, so quite a good number but the 14" 6850U i linked has about 60% better battery life, numbers are numbers.

Notice that screens are both 1080p IPS and 60Hz, so the comparison is as fair as possible.


It's in your own original link. even the previous version of the (almost same) model got fairly close in the H.264 test and did better in the WiFi and Load tests.

1738424286526.png

Then the Mac Air had already come out 2 years earlier and set the bar. So no, nothing revolutionary about that battery life in 2022. It was very good, but not a revolution or something never seen before. But go ahead and refuse to admit when you are wrong and pick another weird hill to die on, just like you always do.

Lastly, LNL models with bigger batteries all have high res (and usually) OLED screens which is why they need the bigger batteries. The fact that you get that battery life with a high res, high quality screen is kind of the point. Even LNL, though, isn't a huge revolution, but it is very good at doing what it's intended to do. It has high single thread and lightly threaded performance and great battery life for thin and light or handhelds (which was the whole point being made to begin with). LNL's problem is on the cost side.

Edit: Here's LNL with a configuration much closer to the laptop you linked to. LNL destroys it in WiFi battery life with only a 5% bigger battery.

1738425287228.png
 
Last edited:

Abwx

Lifer
Apr 2, 2011
11,885
4,873
136
It's in your own original link. even the previous version of the (almost same) model got fairly close in the H.264 test and did better in the WiFi and Load tests.

View attachment 116098

Then the Mac Air had already come out 2 years earlier and set the bar. So no, nothing revolutionary about that battery life in 2022. It was very good, but not a revolution or something never seen before. But go ahead and refuse to admit when you are wrong and pick another weird hill to die on, just like you always do.

Lastly, LNL models with bigger batteries all have high res (and usually) OLED screens which is why they need the bigger batteries. The fact that you get that battery life with a high res, high quality screen is kind of the point. Even LNL, though, isn't a huge revolution, but it is very good at doing what it's intended to do. It has high single thread and lightly threaded performance and great battery life for thin and light or handhelds (which was the whole point being made to begin with). LNL's problem is on the cost side.

H264 depend not from the CPU but from a hardware decoding block with the CPU not loaded, while the web surfing test use the CPU.

And that the previous model did as well doesnt change anything to the fact that AMD was more advanced in battery life, to the contrary it just add more weight to the claim.

Also full load battery life is nor relevant, this is dependent of the implemented CPU power wich is vastly variable from a laptop to another.

As for ST perf that s also discutable, because too high frequencies in ST are inherently limiting battery life since chips can get above 15W in ST at the expense of efficency and still poor perf improvement, FI at 4.5GHz a single core will consume 2x the power used at 3.4GHz while the exe time is only reduced by 25-30%, so 1.4-1.5x lower efficiency.
 

Hitman928

Diamond Member
Apr 15, 2012
6,695
12,370
136
H264 depend not from the CPU but from a hardware decoding block with the CPU not loaded, while the web surfing test use the CPU. And that the previous model did as well doesnt change anything to the fact that AMD was more advanced in battery life, to the contrary it just add more weight to the claim.

The decoding block sits in the SOC or SiP if using tiles/chiplets. It's a part of the "CPU" solution. But sure, let's just look at the WiFi tests at which point we are at Intel being within 10% of the AMD solution when normalizing for battery size and this comes from your own original link. Both get absolutely blown away from a Macbook available 2 years earlier. So we end up again at a very good result in general, but it wasn't talked a ton about because it wasn't that much battery life in 2022 wasn't a big deal. Conversely, LNL adds 50% battery life to that number when normalizing for battery life, so yeah, it gets talked about more. The competitors also improved in that time frame so it's not as big a deal as it would have been in 2022, but it still is a noticeable step up.

Also full load battery life is nor relevant, this is dependent of the implemented CPU power wich is vastly variable from a laptop to another.

So which is it, is nT performance important or not? Because if you say nT performance is the most important metric, then the load number becomes significant.

As for ST perf that s also discutable, because too high frequencies in ST are inherently limiting battery life since chips can get above 15W in ST at the expense of efficency and still poor perf improvement, FI at 4.5GHz a single core will consume 2x the power used at 3.4GHz while the exe time is only reduced by 25-30%, so 1.4-1.5x lower efficiency.

This doesn't add anything to the discussion. It is already accounted for in the fact that the key metrics I pointed to are single / lightly threaded performance and battery life.

You can keep arguing just to argue, but it's just going to end up back to the same spot again and again.
 

Abwx

Lifer
Apr 2, 2011
11,885
4,873
136
But sure, let's just look at the WiFi tests at which point we are at Intel being within 10% of the AMD solution when normalizing for battery size and this comes from your own original link.

I wont get in this debate further, so do us a normalisation, from the numbers you linked the 6850U last 772mn in the Wifi web surfing test with a 52.5Wh battery, the same Lenovo with TGL last 538mn with a 50Wh battery.

Notice that these are the same configuration, so it s the most relevant comparison.

The dell you re talking about not only is a different config with different components but its longer battery life is because of the CPU limited to 4.4 and single RAM channel, at 4.7 it would consume 20% more and battery life would be reduced by something like 15%, you can see the difference with the one that boost at 4.8, at this rate find an even slower CPU, no doubt that it will have even better autonomy.
 
Last edited:

Hitman928

Diamond Member
Apr 15, 2012
6,695
12,370
136
I wont get in this debate further, so do us a normalisation, from the numbers you linked the 6850U last 772mn in the Wifi web surfing test with a 52.5Wh battery, the same Lenovo with TGL last 538mn with a 50Wh battery.

You’re conveniently ignoring the other Intel solution.

Notice that these are the same configuration, so it s the most relevant comparison.

The dell you re talking about not only is a different config with different components but its longer battery life is because of the CPU limited to 4.4 and single RAM channel, at 4.7 it would consume 20% more and battery life would be reduced by something like 15%, you can see the difference with the one that boost at 4.8, at this rate find an even slower CPU, no doubt that it will have even better autonomy.

No. You’re just resorting to making up numbers to try and justify your position.
 

Abwx

Lifer
Apr 2, 2011
11,885
4,873
136
You’re conveniently ignoring the other Intel solution.



No. You’re just resorting to making up numbers to try and justify your position.

The other Intel solution use a CPU that is slower in ST than the Intel Lenovo, so it will do the task in more time even when websurfing all while using much less power.

Beside it use a single channel config, so half the memory controler is switched off,
so even not accounting for the eventual screen difference, a 13"3 for the Dell and a different brand 14" for the 6850U, these parameters have a significant footprint on the power comsumption at low loads.

U would think that comparing at same exact config with a same screen and settings is more accurate than pulling as comparison a laptop with the differences i mentioned, although like the Dell the Lenovo Intel is also single channel, so the difference with the Dell likely come also from the screen.
 

Hitman928

Diamond Member
Apr 15, 2012
6,695
12,370
136
The other Intel solution use a CPU that is slower in ST than the Intel Lenovo, so it will do the task in more time even when websurfing all while using much less power.

Beside it use a single channel config, so half the memory controler is switched off,
so even not accounting for the eventual screen difference, a 13"3 for the Dell and a different brand 14" for the 6850U, these parameters have a significant footprint on the power comsumption at low loads.

U would think that comparing at same exact config with a same screen and settings is more accurate than pulling as comparison a laptop with the differences i mentioned, although like the Dell the Lenovo Intel is also single channel, so the difference with the Dell likely come also from the screen.

I’ll give this to you, you know how to pivot harder than Shaq in the post.
 

itsmydamnation

Diamond Member
Feb 6, 2011
3,072
3,897
136
All i can say is i can get 12 odd hours in a DC* on a Zen3 lenovo slim 5 pro ( 2.5k 120hz, 75Wh, ~75% brightness) just by putting the thing in power saving mode. The STAPM management works really well ( far better then my raptor lake laptop). I have lots of apps open and often have very bursty 1T. 1T numbers are a bit of a circle jerk, larger nT numbers are as well , But 6-8 "real cores" as a power user in battery optimised config is chefs kiss. AMD pretty much had the fundamentals of that down in Zen2 (i also have an 8 core lenovo Zen2 i gave to my daughter).

if the laptop is in performance mode its ~ 2 hours , the performance difference felt in responsiveness or app performance isnt really noticeable, just having more strong threads with STAPM controlling power solves the problem.

* 12 hours straight in a Data Centre is worse then death , the joys of being a hands on network architect :p