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

Page 185 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

adroc_thurston

Diamond Member
Jul 2, 2023
3,322
4,790
96
But it seems to me that Apple's (and probably Qualcomm's) approach (big L1, wide, low frequency) gives better power efficiency.
Ughhh, no.
Huge ROB and big L1d/i in Apple case is a way to sidestep some issues like horribly antiquated branch prediction (things you can't really improve much on yearly cadence).
X4 is very close to Firestorm derivatives in IPC using very x86-style caches and at half the ROB size while being a match in efficiency.
I've always wondered why neither Intel nor AMD embraced Apple's approach of letting high frequency target go and going wide with big caches
Ehhh I mean K7/8/10 had 64/64 3-cycle L1 caches.
 

TESKATLIPOKA

Platinum Member
May 1, 2020
2,428
2,914
136

If Strix Point with 50% more cores is really 35% faster at ~45-55W than Phoenix, then It would mean ~21114–21896 points in CB R23.
That's 14-19% better than 6P+8C 13700H at 115W.
Even i9-13980HX with 8P+16E cores offers only 19478 points at 55W and 23766 points at 80W, so for a comparable score, It would need ~65W.

This doesn't look very good for MTL, which is limited to 6P+8E+2LE. I have to wonder at what power draw It can offer similar performance to Strix Point.
 
Last edited:

BorisTheBlade82

Senior member
May 1, 2020
667
1,022
136
View attachment 88009
If Strix Point with 50% more cores is really 35% faster at ~45-55W than Phoenix, then It would mean ~21114–21896 points in CB R23.
That's 14-19% better than 6P+8C 13700H at 115W.
Even i9-13980HX with 8P+16E cores offers only 19478 points at 55W and 23766 points at 80W, so for a comparable score, It would need ~65W.

This doesn't look very good for MTL, which is limited to 6P+8E+2LE. I have to wonder at what power draw It can offer similar performance to Strix Point.
This will heavily depend on the competitiveness of Intel4 vs. TSMC N4 and the resulting F/V curve for MTL.
I would expect MTL to be quite competitive in MT with reasonable wattages <50w, but Strix to be the winner in ST. I am also not quite convinced that there is no power tax for their tiled approach, but those 2 LP cores might come out as the silver bullet for idle and media consumption.
 
Reactions: Saylick

TESKATLIPOKA

Platinum Member
May 1, 2020
2,428
2,914
136
This will heavily depend on the competitiveness of Intel4 vs. TSMC N4 and the resulting F/V curve for MTL.
I would expect MTL to be quite competitive in MT with reasonable wattages <50w, but Strix to be the winner in ST.
If you think Strix will win in ST, then I don't understand why you think MTL will be competitive in MT and especially at <50W. Don't forget that Strix has 50% more cores than Phoenix.

Even If Intel 4 allowed 1/2 power consumption vs Intel 7, we are still talking about 17,217-18,270 point at 45-55W for 13700H.
MTL would need to gain another 15-20% to be on par with Strix Point.
 

BorisTheBlade82

Senior member
May 1, 2020
667
1,022
136
If you think Strix will win in ST, then I don't understand why you think MTL will be competitive in MT and especially at <50W. Don't forget that Strix has 50% more cores than Phoenix.

Even If Intel 4 allowed 1/2 power consumption vs Intel 7, we are still talking about 17,217-18,270 point at 45-55W for 13700H.
MTL would need to gain another 15-20% to be on par with Strix Point.
I am quite sure that Zen5 with its IPC improvements will beat the 5.1 GHz MTL in ST. But we don't know just how efficient both SoCs will turn out and how high their individual cores will clock in CB at 45w and below. It is not impossible for Intel with their 6+8+2 cores to beat AMD with 8+4. All around Strix nevertheless looks like the better product to me.
 

FlameTail

Diamond Member
Dec 15, 2021
3,159
1,804
106
Dawg you're comparing different SoCs with different clk rates on prime cores.
Do not. do. that.
It's a product-level comparison, not really a core-level one.
Then what are we supposed to do eh?

It's not possible for us or anyone to test the core alone. The testing results will always depend on the implementation by the vendor.
 

TESKATLIPOKA

Platinum Member
May 1, 2020
2,428
2,914
136
I am quite sure that Zen5 with its IPC improvements will beat the 5.1 GHz MTL in ST. But we don't know just how efficient both SoCs will turn out and how high their individual cores will clock in CB at 45w and below. It is not impossible for Intel with their 6+8+2 cores to beat AMD with 8+4. All around Strix nevertheless looks like the better product to me.
MTL to beat Strix Point at comparable TDP? I'm pretty sure It's impossible from the rumours about MTL.

Let's look at the known facts.
14C20T 13700H loses to 8C16T 7840HS at 50W by 14%.
14C20T 13700H surpasses 8C16T 7840HS at 100W by 2%.

Now speculation time.
Let's say 13700H ported to Intel 4 node would more than halve TDP, in other words at 40W It would manage 17,727 points or 45% more than before.
Now It would be 14% faster than 7840HS at the same 40W TDP.

Strix Point will have 12C24T, which is 50% more than 7840HS.
We can't expect the same clocks, so even If Strix Point had to sacrifice 15-20% of clockspeed at 40W, we are still talking about 15,516*1.5*0.85-0.9= 18,619-19,783 points.

18,619-19,783 vs 17,727 = 5-12% faster
With 10-15% IPC gain from Zen5 that's 20,481-22,751 points.
20,481-22,751 vs 17,727 = 16-28% faster

So MTL would need to be 16-28% faster than imaginary Intel 4 Raptor Lake to match Strix Point at 40W TDP.

Where to gain that performance? Frequency was already used up by me and IPC gain at least for P cores looks like zero. That leaves us with those 2 extra E-cores, which at best would provide 10% higher score, and I am not even sure they actively can help during tests.

I summed up why I disagree with you about MTL matching Strix Point in MT performance.
 

H433x0n

Golden Member
Mar 15, 2023
1,068
1,272
96
Where to gain that performance? Frequency was already used up by me and IPC gain at least for P cores looks like zero. That leaves us with those 2 extra E-cores, which at best would provide 10% higher score, and I am not even sure they actively can help during tests.

Crestmont cores have a 5% IPC increase and depending on the SKU +100mhz clock boost.

However, I agree overall that Strix Point will outperform it though.
 

FlameTail

Diamond Member
Dec 15, 2021
3,159
1,804
106
No one outside of QC knows how their chip performs in the real world
We know from their claims we know roughly where the X Elite will be in benchmarks.

One of them being X Elite has 2x performance of Apple M2 in Cinebench 2024. There were several other comparison they showed as well.
 

soresu

Platinum Member
Dec 19, 2014
2,959
2,181
136
One of them being X Elite has 2x performance of Apple M2 in Cinebench 2024
Didn't Cinebench just become GPU bound with the replacement of the OG CPU based rendering engine with their recently acquired GPU path tracer Redshift?

If so thar 2x result would be curb stomped by the Mx Max and Pro SKUs which both have much larger GPUs than the Mx base SKU.
 

adroc_thurston

Diamond Member
Jul 2, 2023
3,322
4,790
96
Didn't Cinebench just become GPU bound with the replacement of the OG CPU based rendering engine with their recently acquired GPU path tracer Redshift?
It has a sorta usable GPU test now but no.
If so thar 2x result would be curb stomped by the Mx Max and Pro SKUs which both have much larger GPUs than the Mx base SKU.
good news everyone!
stx-halo exists.
 

poke01

Golden Member
Mar 8, 2022
1,395
1,611
106
Both Apple and AMD will be fine in 2024.

Its Intel and QC that are in trouble. Intel's MTL is meh and QC should have launched their X Elite this year not mid 2024.

If AMD can release their products in Q1 of 2024 not later in the year that would be great.
 
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |