Question Performanc per Area and Performance per Watt of Apple Firestorm vs Zen 3 vs Intel Golden Cove ?

[FlameTail]

Can someone calculate the [1] Performance per Area and [2] Performance per Watt of the following cores:

1) Apple Firestorm
2) AMD Zen 3
3) Intel Golden Cove

Thanks in advance : )

Edit : I Wouldn't mind values of Sunny Cove if Golden Cove isn't available.

 

[Saylick]

Rename it single thread test then if you want, but running two threads and calling it a "single core" test is down there with correcting spelling like replying "*you're" to someone when you can't come up with any actual defense against their position.

I mean, you wouldn't be wrong... and with our former AT author Andrei nonetheless:

https://twitter.com/x/status/1334233879134298115

 

[Doug S]

I mean, you wouldn't be wrong... and with our former AT author Andrei nonetheless:

https://twitter.com/x/status/1334233879134298115

OMG that's classic! Have a few beers on me for posting that and making me literally laugh out loud!

 

[Doug S]

I agree...but in this case its valid since the OP requested info on performance/die area per core and thats exactly this because SMT is built on the core

Well what is "performance per area"? You are either measuring the performance of a single thread or of a multiple threads. Just because a core has SMT doesn't mean running two threads on it is a valid comparison to a core without it. If you want to go that way, what about cores that share higher levels of cache with other cores? If they can use the ENTIRE shared cache (whether that's L2, L3 or whatever) in a "single core" test then you should include the entirety of that cache when deciding how big that core is and all those calculated core sizes are way low.

This all goes to show how dumb the entire concept of "performance per area" is.

 

[nicalandia]

This all goes to show how dumb the entire concept of "performance per area" is.

It's the best I could come up as the OP requested, since they are pretty difference CPU implementations wee need a starting point

 

[Cardyak]

Neither of these concepts is incorrect, they are just measuring two different things:

If you want to determine how fast a single thread can be processed, use single thread performance: 1 thread into 1 core

If you want to determine how fast a single core is, then utilise SMT (if available): >=1 thread (s) into 1 core

A lot of the anger and confusion comes from people assuming one measurement over the other. It’s incredibly important to clarify *exactly* what data you’re trying to capture.

 

[nicalandia]

Lets use ST to test Performance Per Area at stock speed, also would like to compare it with Apple M1 Firestorm Performance Core and AMD Zen 3 for reference





Intel Golden Cove core with L2$ as measured by Locuza is 7.04 mm2 and it gets 1,937 points in GB5
Apple Firestorm core with L2$ measured by Semianalysis is 3.83 mm2 and gets 1,745 points in GB5
AMD Zen 3 core with L2$ as measured by Locuza is 4.27 mm2 and it gets 1,506 points in GB5
Intel Gracemont core with L2$ as measured by Locuza is 2.19 mm2 and gets 1,168 in GB5


Performance/mm2

1st place is Intel Gracemont core with 532 Geekbench5 points per mm2
2nd place is Apple Firestorm core with 455.6 Geekbench5 points per mm2
3rd place is AMD Zen3 core with 352.7 Geekbench5 points per mm2
4th place is Intel Golden Cove with 275.1 Geekbench5 points per mm2


***************************************************************************************



Why stop at Golden Cove vs Firestorm? Let's include Intel Gracemont efficiency core and Apple efficiency Icestorm core

Anandtech SPEC2006 ST Suit for Floating Point and Integer Benchmarks

SPEC2006 - 453.povray ST

Still the Gracemont core reign supreme in Performance/Area

First Place:
Intel Gracemont core with L2$ area is 2.2 mm2
tested under SPEC2006 - 453.povray gets 59.50 points so the performance point per area is 27.04

Second Place:
Apple Firestorm core with L2$ area is 3.83 mm2 tested under SPEC2006 - 453.povray gets 88.80 points so the performance points per area is 23.18

Third Place:
Intel Golden Cove core with L2$ area is 7.04 mm2. Tested under SPEC2006 - 453.povray gets 117.7 points so the performance points per area is 16.71

Fourth Place:
Apple Icestorm core with L2$ area is 1.445 mm2. Tested under SPEC2006 - 453.povray gets 23.72 points so the performance points per area is 16.41

 

[Abwx]

Okay, I found a better source for Apple Firestorm Core + L2, the one at Techpowerup I found earlier had a rough diagram that made the L2 larger than expected and the 9.1 was for the two cores and the big L2$, here is a better diagram that I was able to confirm(Pixel/mm2 wise), also there is detailed chart per block.

Here is the diagram.

Apple Firestorm Core on the A14 SOC size: 3.76 mm2 per Core with L2$

View attachment 54705

Source: https://semiwiki.com/semiconductor-...sis-terrifying-implications-for-the-industry/




AMD Zen3 Core on Cezanne SOC size: 6.4 mm2 with L3$ and 4.2 mm2 with L2$ only
View attachment 54706


Source: https://videocardz.com/newz/amd-ryzen-5-5600g-cezanne-apu-die-has-been-pictured-up-close



Intel Golden Cove core on Alder Lake SOC size: 7.04 mm2 with L2$ and 9.4 with L3$, the Ring bus is Huge compared to Zen 3
View attachment 54713



Some benchmarks(CBR23 and Geekbench 5) and Final numbers.

View attachment 54719

View attachment 54720




Apple Firestorm core area: 3.76 mm2 per core with L2$ only(no system cache)
CBR23: 408.5 points per area,
GB5: 463.3 points per area
11 Watts peak power CBR23 ST: 139.6 points per watt


Intel Golden Cove core area: 7.04 mm2 per core with L2$ only
CBR23: 254.5 points per area
GB5: 229.7 points per area
45 Watts peak power CBR23 ST: 39 points per watt

AMD Zen3 Core area: 4.2 mm2 per core with L2$ only
CBR23: 362 points per area
GB5: 384.2 points per area
45 Watts peak power CBR23 ST: 33.8 points per watt

45W for a single core..?.
Did you actually check the ST average power during a CB riun.??

 

[nicalandia]

45W for a single core..?.
Did you actually check the ST average power during a CB riun.??

It was not me, it was Anandtech review. Let me check

 

[Abwx]

It was not me, it was Anandtech review. Let me check

View attachment 57205

It s about 30W for the core of a 11980HK, and you used the same 45W figure for "estimating" Zen 3 efficency.

So your number for Intel efficency is 33% lower than what is stated by AT and surely something like 50% for Zen 3 since it consume less than 30W in ST.

 

[nicalandia]

It s about 30W for the core of a 11980HK, and you used the same 45W figure for "estimating" Zen 3 efficency.

So your number for Intel efficency is 33% lower than what is stated by AT and surely something like 50% for Zen 3 since it consume less than 30W in ST.

My numbers might be off, really. Do you know how long(in seconds) it takes a CBR23 ST run? We can plot that with this info

 

[Abwx]

My numbers might be off, really. Do you know how long(in seconds) it takes a CBR23 ST run? We can plot that with this info

View attachment 57212

Comparing monolithic chips indeed allow better estimation of a single core power.
In CB ST a 12900K core use 36W@5.2, a 12700K is at 32W@4.9 and a i5 12500 is at 27W@4.6
Comparatively a Ryzen 5700G core use 18W@4.6.

Intel Core i9-12900K, i7-12700K & i5-12600K im Test: Leistungsaufnahme, Temperaturen und OC

Intel Alder Lake im Test: Leistungsaufnahme, Temperaturen und OC / Leistungsaufnahme in Spielen / Leistungsaufnahme in Anwendungen

www.computerbase.de

Intel Core i5-12400 und Core i5-12500 im Test: Leistungsaufnahme, Temperatur und Overclocking

Core i5-12400 und i5-12500 im Test: Leistungsaufnahme, Temperatur und Overclocking / Overclocking: Über den BCLK auf DDR5-Boards möglich

www.computerbase.de

 

[eek2121]

Okay, I found a better source for Apple Firestorm Core + L2, the one at Techpowerup I found earlier had a rough diagram that made the L2 larger than expected and the 9.1 was for the two cores and the big L2$, here is a better diagram that I was able to confirm(Pixel/mm2 wise), also there is detailed chart per block.

Here is the diagram.

Apple Firestorm Core on the A14 SOC size: 3.76 mm2 per Core with L2$

View attachment 54705

Source: https://semiwiki.com/semiconductor-...sis-terrifying-implications-for-the-industry/




AMD Zen3 Core on Cezanne SOC size: 6.4 mm2 with L3$ and 4.2 mm2 with L2$ only
View attachment 54706


Source: https://videocardz.com/newz/amd-ryzen-5-5600g-cezanne-apu-die-has-been-pictured-up-close



Intel Golden Cove core on Alder Lake SOC size: 7.04 mm2 with L2$ and 9.4 with L3$, the Ring bus is Huge compared to Zen 3
View attachment 54713



Some benchmarks(CBR23 and Geekbench 5) and Final numbers.

View attachment 54719

View attachment 54720




Apple Firestorm core area: 3.76 mm2 per core with L2$ only(no system cache)
CBR23: 408.5 points per area,
GB5: 463.3 points per area
11 Watts peak power CBR23 ST: 139.6 points per watt


Intel Golden Cove core area: 7.04 mm2 per core with L2$ only
CBR23: 254.5 points per area
GB5: 229.7 points per area
45 Watts peak power CBR23 ST: 39 points per watt

AMD Zen3 Core area: 4.2 mm2 per core with L2$ only
CBR23: 362 points per area
GB5: 384.2 points per area
45 Watts peak power CBR23 ST: 33.8 points per watt

Find me a Zen 3 chip that uses 45W of power for a single core. That is your first issue.

EDIT: My 5950X uses 6-12W for CBR23 on a single core, FYI.

 

[nicalandia]

Comparing monolithic chips indeed allow better estimation of a single core power.
In CB ST a 12900K core use 36W@5.2, a 12700K is at 32W@4.9 and a i5 12500 is at 27W@4.6
Comparatively a Ryzen 5700G core use 18W@4.6.

Intel Core i9-12900K, i7-12700K & i5-12600K im Test: Leistungsaufnahme, Temperaturen und OC

Intel Alder Lake im Test: Leistungsaufnahme, Temperaturen und OC / Leistungsaufnahme in Spielen / Leistungsaufnahme in Anwendungen

www.computerbase.de

Intel Core i5-12400 und Core i5-12500 im Test: Leistungsaufnahme, Temperatur und Overclocking

Core i5-12400 und i5-12500 im Test: Leistungsaufnahme, Temperatur und Overclocking / Overclocking: Über den BCLK auf DDR5-Boards möglich

www.computerbase.de

Could you do me a big Favor. I hate clicking on those links, it takes me to a website that is not in English and I just don't know how to get past the first pop up. So kindly post the graphs here.

 

[Abwx]

Could you do me a big Favor. I hate clicking on those links, it takes me to a website that is not in English and I just don't know how to get past the first pop up. So kindly post the graphs here.

This site, and german generaly, is well translated by google.

There a graph for idle, CB ST, CB MT and Prime 95 powers.

(CB ST power minus idle power) x 0.775 = single core power.

That being said i posted the numbers, so why bother doing the same work..?

 

[eek2121]

Comparing monolithic chips indeed allow better estimation of a single core power.
In CB ST a 12900K core use 36W@5.2, a 12700K is at 32W@4.9 and a i5 12500 is at 27W@4.6
Comparatively a Ryzen 5700G core use 18W@4.6.

Intel Core i9-12900K, i7-12700K & i5-12600K im Test: Leistungsaufnahme, Temperaturen und OC

Intel Alder Lake im Test: Leistungsaufnahme, Temperaturen und OC / Leistungsaufnahme in Spielen / Leistungsaufnahme in Anwendungen

www.computerbase.de

Intel Core i5-12400 und Core i5-12500 im Test: Leistungsaufnahme, Temperatur und Overclocking

Core i5-12400 und i5-12500 im Test: Leistungsaufnahme, Temperatur und Overclocking / Overclocking: Über den BCLK auf DDR5-Boards möglich

www.computerbase.de

Thinking back here, you are REALLY cherry picking. You are comparing a mobile SoC (M1) with a high-end desktop chip (12900k) and a budget desktop chip (5700g). The actual proper comparison would have been the M1 vs. the 5980hs and one of the most efficient mobile Alder Lake chips. Desktop and budget chips are in an entirely different world than the M1 Pro/Max.

 

[Abwx]

Thinking back here, you are REALLY cherry picking. You are comparing a mobile SoC (M1) with a high-end desktop chip (12900k) and a budget desktop chip (5700g). The actual proper comparison would have been the M1 vs. the 5980hs and one of the most efficient mobile Alder Lake chips. Desktop and budget chips are in an entirely different world than the M1 Pro/Max.

Dunno for the M1 but there s a comparison within the numbers i posted wich is very relevant :

i5 12500 is at 27W@4.6
Comparatively a Ryzen 5700G core use 18W@4.6.

Laptop chips will work around this frequency in ST, actually even at 5GHz for Intel chips, and being a mobile SKU will change nothing, in ST a mobile chip core will consume as much as a DT SKU core at the same frequency.

 

[leoneazzurro]

That would be true if voltage and binning would have been the same between mobile and desktop, which rarely is. Generally mobile cores have a firmware with a different voltage pattern. If voltage is the same, then both desktop and mobile would consume the same.

 

[Asterox]

Blender CPU rendering, if the data is legit everything is very clean there.

 

[nicalandia]

Blender CPU rendering, if the data is legit everything is very clean there.

View attachment 57546

Impressive

 

[moinmoin]

Normalized to imaginary watt/seconds estimated Joules, lower is better:

​	6900HS​	12900HK​
25W​	6300​	8750​
30W​	7290​	8700​
40W​	8120​	9040​
50W​	9300​	9600​
60W​	10620​	10380​
70W​	11970​	11340​

So 12900HK has worse efficiency at 25W than at 30W, and 6900HS is by far the most efficient at 25W.

 

[deasd]

I'm just shocking that big+little design in X86 area is not as efficient as we thought, where's problem? 6P+8E can't win in sub 60 watts already a big failure to me....

 

[Hitman928]

Normalized to imaginary watt/seconds, lower is better:

​	6900HS​	12900HK​
25W​	6300​	8750​
30W​	7290​	8700​
40W​	8120​	9040​
50W​	9300​	9600​
60W​	10620​	10380​
70W​	11970​	11340​

So 12900HK has worse efficiency at 25W than at 30W, and 6900HS is by far the most efficient at 25W.

I think you mean watts*seconds which is not imaginary, it is Joules. What you are showing is total energy used to complete the render.

 

[moinmoin]

I think you mean watts*seconds which is not imaginary, it is Joules. What you are showing is total energy used to complete the render.

Depends on how the watts were measured/limited. I wouldn't trust the benchmarks to have ran through exactly at the given wattage for the whole of the duration. Estimated joules then.

 

[Hitman928]

Depends on how the watts were measured/limited. I wouldn't trust the benchmarks to have ran through exactly at the given wattage for the whole of the duration. Estimated joules then.

True, it is an estimate, but it should be a pretty accurate one given the nature of the load which can fully load the cores and can scale well beyond the core counts present in these chips. Other than that, there will be a very short amount of time at the beginning and end where the power consumed drops as the render is started and completed, but it will not be a long enough to have an appreciable effect on the results.

 

[Hitman928]

Here's a quick plot of the energy used table from @moinmoin