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

[DisEnchantment]

Speculate at will

 

[Det0x]

Pbo+co?

Take a look at the first 40w screenshot --> PBO+CO was enabled

 

[CouncilorIrissa]

But it doesn't make sense that stock ES is better than PBO scores that AMD shows officially. Also

AMDs slide says Ryzen 9000 is 7° cooler at same power. And now this ES runs at 230W with 60°?!?. 7950X needed a chiller to even get to 230W at 95°. Most reviews with AIO measured a Max powerdraw of 190-200W for 7950X because it simply ran into the 95° Limit. So this ES looks to be more like 40-45° cooler at same power. Very suspicious.

I wonder if there's some weird offset that causes the temps to be reported N degrees lower than they actually are.

 

[techjunkie123]

I wonder if there's some weird offset that causes the temps to be reported N degrees lower than they actually are.

Yeah I'm more likely to believe temperature reporting being off on the ES...maybe the sensors aren't configured properly idk .

 

[Josh128]

Dont know if this is my requested result or not, but just saw this:

 

[CouncilorIrissa]

Dont know if this is my requested result or not, but just saw this:

View attachment 103250

I wouldn't trust anything from that account.

 

[CouncilorIrissa]

Yeah I'm more likely to believe temperature reporting being off on the ES...maybe the sensors aren't configured properly idk .

yeah, low wattage runs report temperatures in the low 30s, that's impossible.

 

[therealmongo]

Has it been confirmed the process node used for the CCD and IOD in 9000 series ??
nvm

 

[techjunkie123]

Has it been confirmed the process node used for the CCD and IOD in 9000 series ??

CCD: N4P
IOD: N6

 

[DavidC1]

Zen 5 mobile seem to have worse performance per clock vs desktop zen 5 in geekbench 6. Hx 370: 2860@5.1ghz

Mobile is slower than desktop per clock. That's normal.

 

[cebri1]

From @Asus marketing material

 

[igor_kavinski]

Take a look at the first 40w screenshot --> PBO+CO was enabled
View attachment 103247

My bad. I forgot about the first screenshot.

 

[igor_kavinski]

Dont know if this is my requested result or not, but just saw this:

View attachment 103250

No, yours is coming. Soon. Hopefully.

 

[Philste]

My bad. I forgot about the first screenshot.

Me too. So the numbers from AMD event (42k stock, 44k PBO, 45.5K Curve Shaper) could be true. But this still is like the best ES ever.

 

[igor_kavinski]

But this still is like the best ES ever.

Or it's just grateful for the royal treatment (custom watercooling)

 

[gaav87]

14900kf
1.45V
417.6W
P-cores 6ghz / E-cores: 4.8
R23: 45K

 

[Hail The Brain Slug]

14900kf
1.45V
417.6W
P-cores 6ghz / E-cores: 4.8
R23: 45K

I wonder how many seconds it can operate like this before it dies?

 

[StefanR5R]

There is no reason to marvel over these temperature readouts (from an ES, to boot).

And now this ES runs at 230W with 60°?!?

Have you ever seen temperature readouts of EPYC chips @400 W under decent air cooling? Let alone water cooling.

CPU temperature readouts are sensitive to

• where they are measured (edit, since temperature is not uniform over the die),
• at which clock speed the spot which is measured is driven (edit, since logic uses nonlinearly more power at increased clock speed), and
• how the firmware translates measurements to user-exposed values.¹

yeah, low wattage runs report temperatures in the low 30s, that's impossible.

Do you happen to know the ambient temperature which the leaker maintained?

Edit:
¹) Many of you will remember that AMD once did something else: Let the CPU firmware report stupidly increased temperature values, so that cooler fans tend to spin faster.

Edit 2:
Who of you has ever soldered his own temperature sensor onto the naked die in order to learn the "real" temperature of his CPU?

 

[Det0x]

Me too. So the numbers from AMD event (42k stock, 44k PBO, 45.5K Curve Shaper) could be true. But this still is like the best ES ever.

Think it only comes down to that those two amd guys arent really good tuners. They simply have bad efficiency/optimizations on all their runs compared to "real overclockers" (that video is only for showcase and PR afterall)
We went throught the same with Zen4(TR) before release, where they showed LN2 runs and "xtreme OC", only to get smashed by prettymuch everyone on hwbot (even those with much worse cooling)

But that stock 42k i can believe, its only their PBO/PBO CS numbers i dont have too much faith in..
And who's to say AMD didnt change their "TDP formula" for this release ?
"170w TDP" could very well mean a 200w PPT limit this time around

Remember TDP for cpus is a made up number

AMD Ryzen TDP Explained: Deep-Dive on TDP Definitions & What Cooler Manufacturers Think | GamersNexus

stub Thermal Design Power, or TDP, is a term used by AMD and Intel to refer in an extremely broad sense to the rate at which a CPU cooler must dissipate heat from the chip to allow it to perform as advertised. Sort of. Depending on the specific formula and product, this number often ends up a...

gamersnexus.net

 

[Hail The Brain Slug]

yeah, low wattage runs report temperatures in the low 30s, that's impossible.

I guess you didn't look at my 7950X on an air cooler screenshots.

40W:


60W:

 

[Saylick]

I wonder how many seconds it can operate like this before it dies?

Inb4 new AMD marketing metric vs. Intel: how many seconds of benchmarking the CPU can run before it goes kaput.

 

[Abwx]

Those 22% should also apply to Cinebench to some extent :

Ryzen 9000/AI 300: Details zu Zen 5, RDNA 3.5, Zen 6(c) & Zen 7

Ryzen 9000 kommt am 31. Juli. Schon heute gibt es weitere technische Details zu den Desktop-CPUs und Ryzen AI 300 Strix Point für Notebooks.

www.computerbase.de

 

[gaav87]

I wonder how many seconds it can operate like this before it dies?

seconds

Think it only comes down to that those two amd guys arent really good tuners. They simply have bad efficiency/optimations on all their runs compared to "real overclockers" (that video is only for showcase and PR afterall)
We went throught the same with Zen4(TR) before release, where they showed LN2 runs and "xtreme OC", only to get smashed by prettymuch everyone on hwbot (even those with much worse cooling)

But that stock 42k i can believe, its only their PBO/PBO CS numbers i dont have too much faith in..
And whois to say AMD didnt change their "TDP formula" for this release ?
"170w TDP" could very well mean a 200w PPT limit this time around

Remember TDP for cpus is a made up number

AMD Ryzen TDP Explained: Deep-Dive on TDP Definitions & What Cooler Manufacturers Think | GamersNexus

stub Thermal Design Power, or TDP, is a term used by AMD and Intel to refer in an extremely broad sense to the rate at which a CPU cooler must dissipate heat from the chip to allow it to perform as advertised. Sort of. Depending on the specific formula and product, this number often ends up a...

gamersnexus.net

This screenshot that is beeing spammed on wccftech is incorrect:
At 13:24 in the GamersNexus video the guy said hes running 5ghz all core at 1.35V (but he could run it at 1.1V at the current temp of -69C(nice) with a score of 43375 and he said i quote:
"This is what you could call mission mode STOCK mode"

 

[Abwx]

This screenshot that is beeing spammed on wccftech is incorrect:
At 13:39 in the GamersNexus video the guy said hes running 5ghz all core at 1.35V (but he could run it at 1.1V at the current temp of -69C(nice) with a score of 43375 and he said i quote:
"This is what you could call mission mode STOCK mode"

At very low temp mosfets transistors conduction is lower than at ambiant, so with such low temps you cant reach clocks as high as when the silicon is at say 30°C and at a same voltage.

As temp increase the mosfets conduct better up to a given temp at wich their conduction decrease again, that could explain the relatively low score, for the time i keep on my estimation of 46k points at 230W PPT that i made before Igor made a run at this power.

 

[Det0x]

seconds

This screenshot that is beeing spammed on wccftech is incorrect:
At 13:24 in the GamersNexus video the guy said hes running 5ghz all core at 1.35V (but he could run it at 1.1V at the current temp of -69C(nice) with a score of 43375 and he said i quote:
"This is what you could call mission mode STOCK mode"

Not that it matters much, but in the ES 200w PPT screenshot you can see the cpu running ~5.1ghz @ ~1.080v
*edit*
Didnt hassan post on X about a stock 9950X using 19x watt running cooler awhile back ? (before all these ES leaks from this forum)
Think i saw it earlier today but cant find now

 

[adroc_thurston]

Didnt hassan post on X about a 9950X using 19x watt running cooler awhile back ? (before all these ES leaks from this forum)

That's indeed what it draws.
Seems thermally-limited which is kinda funny.