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

Page 651 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
I

igor_kavinski

Lifer
Jul 27, 2020
27,953
19,100
146
Timmah! said:
Regarding scaling better with additional power, granted , it scales where zen4 stopped to scale. The scaling is still rather insignificant, its not like gain additional 50 percent of perf going from 160W to 230. Its maybe 10~15 percent and while thats more than almost nothing before, its still not something extremely impressive, that would take zen5 out of sight of Zen4/RPL perf-wise. And thats really the most important thing imo.
Click to expand...
It's just AMD being cautious and not trying to bite off more than they can chew. The X3D is supposed to be the special SKU of this generation. The competition by the end of the year will be super interesting to watch and unless Intel manages some miracle, this will be AMD's year.
 
  • Like
Reactions: Joe NYC, Tlh97 and Saylick
C

CouncilorIrissa

Senior member
Jul 28, 2023
722
2,675
106
Finally the GB6 run with Strix boosting properly and it breaks the 2.9k score for the first time.

ASUSTeK COMPUTER INC. ProArt P16 H7606WI_H7606WI - Geekbench

Benchmark results for an ASUSTeK COMPUTER INC. ProArt P16 H7606WI_H7606WI with an AMD Ryzen AI 9 HX 370 processor.
browser.geekbench.com

ASUSTeK COMPUTER INC. ProArt P16 H7606WI_H7606WI - Geekbench

Benchmark results for an ASUSTeK COMPUTER INC. ProArt P16 H7606WI_H7606WI with an AMD Ryzen AI 9 HX 370 processor.
browser.geekbench.com
"frequencies": [
5137,
5137,
5135,
5136,
4914,
4939,
5137,
5135,
5137,
5135,
5137,
5138,
5135,
5138,
5137,
5137,
5136,
5135,
5137,
5134,
5137,
5138,
5135,
5132,
5135,
5137,
5138,
5136,
5134,
5138,
5137,
5138,
5137,
5136,
5099,
5138,
5137,
5136,
5136,
5137,
5138,
5137,
5138,
4966,
5138,
5138,
5136,
5137,
5137,
5136,
5136,
5137
]
 
  • Like
Reactions: Elfear, Tarkin77, Josh128 and 8 others
S

StefanR5R

Elite Member
Dec 10, 2016
6,667
10,545
136
Timmah! said:
IMO the one reason why the CCDs are the same size as Zen4 and only transistor count increase comes from the newer denser process, is cause AMD dont thinks they can sell more expensive than Z4 and wont be giving any additional performance increase for “free” - as in potentially bigger CCDs could have been even faster, but AMD would expect to be paid for every additional mm2, and people would not be willing to pay….

Or i am completely wrong and its the same size cause any bigger would not fit under IHS :)
Click to expand...
Yes, you are completely wrong ;-) and it is precisely because bigger CCDs would not fit under the IHS. And by that I refer to 1 sIOD + up to 16 CCDs having to fit under Turin's IHS. Granite Ridge's IHS would certainly have room for a little bigger CCDs still. (Although maybe Granite Ridge and Turin might receive CCDs at differing steppings, I presume that the design is the same, just like it was shared between desktop and server in Zen1...Zen4.)

BTW, it has been speculated here that the density increase of Granite Ridge's CCD vs. Raphael's is not only thanks to the process bump, but also due to denser L3$ by means of tweaks which are not yet clear.
 
  • Like
Reactions: Joe NYC, Tlh97, Timmah! and 1 other person
Tup3x

Tup3x

Golden Member
Dec 31, 2016
1,272
1,405
136
CouncilorIrissa said:
Finally the GB6 run with Strix boosting properly and it breaks the 2.9k score for the first time.

ASUSTeK COMPUTER INC. ProArt P16 H7606WI_H7606WI - Geekbench

Benchmark results for an ASUSTeK COMPUTER INC. ProArt P16 H7606WI_H7606WI with an AMD Ryzen AI 9 HX 370 processor.
browser.geekbench.com

ASUSTeK COMPUTER INC. ProArt P16 H7606WI_H7606WI - Geekbench

Benchmark results for an ASUSTeK COMPUTER INC. ProArt P16 H7606WI_H7606WI with an AMD Ryzen AI 9 HX 370 processor.
browser.geekbench.com
"frequencies": [
5137,
5137,
5135,
5136,
4914,
4939,
5137,
5135,
5137,
5135,
5137,
5138,
5135,
5138,
5137,
5137,
5136,
5135,
5137,
5134,
5137,
5138,
5135,
5132,
5135,
5137,
5138,
5136,
5134,
5138,
5137,
5138,
5137,
5136,
5099,
5138,
5137,
5136,
5136,
5137,
5138,
5137,
5138,
4966,
5138,
5138,
5136,
5137,
5137,
5136,
5136,
5137
]
Click to expand...
Probably high performance mode. That laptop also has high TDP so your typical business laptop is not going to be anywhere close.
EDIT: Well, power plan was indeed set to performance.
 
Philste

Philste

Senior member
Oct 13, 2023
299
474
96
CouncilorIrissa said:
Finally the GB6 run with Strix boosting properly and it breaks the 2.9k score for the first time.
Click to expand...
This should be the final score. David Huang got exactly 3k for the 5.0GHz SKU at Linux. Linux typically has 5% advantage, so it adds up exactly. Also the newest Lunar Lake result has basically exactly the same per clock performance.
 
Last edited:
  • Like
Reactions: Elfear, Mopetar, Josh128 and 3 others
gaav87

gaav87

Senior member
Apr 27, 2024
659
1,279
96
mpumalanga said:
I don't see any gaming numbers in your message.

CapFrameX is on the same level as userbenchmark with regards to AMD. Best to ignore both unless you want to read something really wrong and misleading.
Click to expand...
Thunder 57 said:
Quick search seems to suggest that yes, they are Userbenchmark-lite, if you will.
Click to expand...
Speaking of UB. What will the poor chap do when zen5 hits and is faster in every single of his tests ? xD Will he sort the CPU's by his ranking this time ? xD
 
  • Haha
  • Like
Reactions: Mopetar, Thunder 57, lightmanek and 3 others
I

igor_kavinski

Lifer
Jul 27, 2020
27,953
19,100
146

Focusing on the MT tests and keeping in mind that the Ryzen has 33% fewer threads:

File compression: Good
Navigation: Decent
HTML Browser: Almost on par
PDF renderer: Very good
Photo Library: Good
Clang: OK (was expecting better here)
Text processing: Outstanding
Asset compression: Good
Object detection: Good
Background blur: Outstanding
Horizon detection: Good
Object remover: Good
HDR: Very good
Photo filter: Good
Raytracer: Good
Structure from Motion: Good

In ST, the Ryzen knocks it out of the park in HDR, Object detection and text processing while being decent in Clang, PDF renderer and HTML5 browser.

Excellent MT performance against Intel despite only 24MB L3 cache.
 
  • Like
Reactions: lightmanek, Joe NYC and Tlh97
C

CouncilorIrissa

Senior member
Jul 28, 2023
722
2,675
106
Philste said:
This should be the final score. David Huang got exactly 3k for the 5.0GHz SKU at Linux. Linux typically has 5% advantage, so it adds up exactly. Also the newest Lunar Lake result has basically exactly the same per clock performance.
Click to expand...
That's probably it for the LPDDR5 devices, yes.
Some DDR5 devices will probably score around 3050-3100 (there are absolutely cracked Hawk Point runs doing 2700 with DDR5, whereas LPDDR5 runs hover around 2550)
 
  • Like
Reactions: igor_kavinski and Tlh97
T

Timmah!

Golden Member
Jul 24, 2010
1,571
935
136
StefanR5R said:
Yes, you are completely wrong ;-) and it is precisely because bigger CCDs would not fit under the IHS. And by that I refer to 1 sIOD + up to 16 CCDs having to fit under Turin's IHS. Granite Ridge's IHS would certainly have room for a little bigger CCDs still. (Although maybe Granite Ridge and Turin might receive CCDs at differing steppings, I presume that the design is the same, just like it was shared between desktop and server in Zen1...Zen4.)

BTW, it has been speculated here that the density increase of Granite Ridge's CCD vs. Raphael's is not only thanks to the process bump, but also due to denser L3$ by means of tweaks which are not yet clear.
Click to expand...
Well then its settled, did not consider Turin for sure, thx for correcting me.
 
J

Joe NYC

Diamond Member
Jun 26, 2021
3,630
5,173
136
Josh128 said:
Regarding CapframeX's not so cryptic tweet. I can believe it. The main reason I believe why is the same IOD & memory limitations of Zen 4. When looking at Zen 4 memory scaling testing for games, 4800 to 6000 shows a real, tangible difference due to the increase being in the same 1:1 ratio. Increases from 6000 1:2 to 8000 1:2 are either very small, or nothing at all. When you look at Intel scaling, they seem to achieve that same boost that AMD gets from 4800 to 6000 all the way to 8000+. No matter how good the core, its going to be limited in this regard. Until AMD improve mem controller and subsystem, the will always be behind the 8 ball in this regard.
Click to expand...

Interesting how people are still comparing the unstable Raptor Lake performance to stable Zen performance.

Specifically, one of the ways to "stabilize" Raptor Lake is to turn down the memory speed, all the way down to 4800.

Also, on the Zen 5 side, apparently the internal IF clock can go to 2400, from 2000, so memory speed performance increases can go up to 7200. Although I am not sure if that works under current chipsets or only the upcoming new ones.
 
  • Like
Reactions: Tlh97, carancho, igor_kavinski and 1 other person
You must log in or register to reply here.

TRENDING THREADS

COMPANY
RESOURCES
FOLLOW
Top Bottom