Info TOP 20 of the World's Most Powerful CPU Cores - IPC/PPC comparison

Page 3 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Antey

Junior Member
Jul 4, 2019
22
22
41
So you are taking the best results grom geekbench 5.1 (linux or windows for x86-64? i say that because the higher ones tend to be on linux) and dividing it by their frequency. but you can also do that with cinebench R15 and cinebench R20 and get different results with the same CPU. a 3800X gets 46.6 per ghz in R15 and a 112.9 in R20 and the difference between zen2 and zen will be higher in R20 than R15 (use of avx), but also different from what you can get from geekbench, and very different from the differences you can see with other workloads or benchmarks like videogames or encoding, etc. You get different results from zen (1400x) and zen (2400g) because of the cache size differences.

Are you sure that apple cpus aren't getting higher scores over android ones for similar reasons linux is getting higher results than windows? or maybe ios vs linux and windows?

just look at this

3800x on linux -> 1427/4.65 = 306


and what about this one?

3800x on windows -> 1056/4.3 = 246


is zen 2 ipc 24% higher than zen 2 ipc? if tomorrow microsoft releases a new windows version and you get a ''ppc'' of 400 with a 3950x at 4.6ghz, ipc just increased 40%?

where did you get zen 2 gets 1317 points at 4.6? did you just take the higher one you found?

and what about other cpus? try with sandy bridge and bulldozer. 171 vs 137... are you sure sandy bridge is just 25% over bulldozer? what happened there?


i don't know much about ipc and such but thats the first thing i would do if i tried to compare IPC... and thats me, and i don't know much about it, it reminds me to console fans comparing tflops between ps4, ps5, xbox, 2080ti, rx5700, radeon vii, etc...
 
Last edited:

Richie Rich

Senior member
Jul 28, 2019
421
196
76
it reminds me to console fans comparing tflops between ps4, ps5, xbox, 2080ti, rx5700, radeon vii, etc...
That's why I don't compare theoretical performance but real one measured in GeekBench 5. Especially now with Apple's transition to ARM it is useful to predict performance of new core A14 Firestorm. No doubt A14 will be the most powerful core in terms of PPC/IPC. A13 was beating Zen2 @ 4.7 GHz and got very close to 9900K @ 5 GHz. Apple's A14 could be the most powerful core in terms of absolute single thread performance. But let's see.

Another big thing is the new ISA ARMv9 with 2048-bit SVE2 vectors for A15 next year (we know ARMs Matterhorn ARMv9 arrives next year). There is a chance that A14 this year will be ARMv9 with 2048-bit SVE2 too. IIRC Apple introduced first 64-bit core (A7 cyclone 2013 with 4xALUs) two years before ARM introduced their first 64-bit Cortex A35/72.

Regarding the point of different results of 3800X: Any valid statistics has to eliminate faulty measurements. We all know about this. The pure existence of faulty results doesn't prove this comparison doesn't make sense. Dozer was useless at that time of launch day I agree. However in some modern applications like GeekBench 5.1 can provide much better results and those 25% over Sandy can be correct.
 
Last edited:

lobz

Golden Member
Feb 10, 2017
1,194
1,156
106
That's why I don't compare theoretical performance but real one measured in GeekBench 5. Especially now with Apple's transition to ARM it is useful to predict performance of new core A14 Firestorm. No doubt A14 will be the most powerful core in terms of PPC/IPC. A13 was beating Zen2 @ 4.7 GHz and got very close to 9900K @ 5 GHz. Apple's A14 could be the most powerful core in terms of absolute single thread performance. But let's see.

Another big thing is the new ISA ARMv9 with 2048-bit SVE2 vectors for A15 next year (we know ARMs Matterhorn ARMv9 arrives next year). There is a chance that A14 this year will be ARMv9 with 2048-bit SVE2 too. IIRC Apple introduced first 64-bit core (A7 cyclone 2013 with 4xALUs) two years before ARM introduced their first 64-bit Cortex A35/72.

Regarding the point of different results of 3800X: Any valid statistics has to eliminate faulty measurements. We all know about this. The pure existence of faulty results doesn't prove this comparison doesn't make sense. Dozer was useless at that time of launch day I agree. However in some modern applications like GeekBench 5.1 can provide much better results and those 25% over Sandy can be correct.
On a more serious note, don't forget SMT4 or SMT8. Possibly even SMT9, since it's ARMv9.
 

Richie Rich

Senior member
Jul 28, 2019
421
196
76
Added cores:
  • A53 - little core used in some low-end smartphones in 8-core config (Snapdragon 450)
  • A55 - used as little core in every modern Android SoC
  • A72 - "high" end Cortex core used in Snapdragon 625 or Raspberry Pi 4
  • A73 - "high" end Cortex core
  • A75 - "high" end Cortex core
  • Bulldozer - infamous AMD core

Geekbench 5.1 PPC chart 6/23/2020:
Pos
Man
CPU
Core
Year
ISA
GB5 Score
GHz
PPC (score/GHz)
Relative
Relative
1​
Apple​
A13​
Lightning​
2019​
ARMv8​
1332​
2.65​
502.64​
100%​
182%​
2​
Apple​
A12​
Vortex​
2018​
ARMv8​
1116​
2.53​
441.11​
88%​
159%​
3​
Apple​
A11​
Monsoon​
2017​
ARMv8​
933​
2.39​
390.38​
78%​
141%​
4​
ARM Cortex​
X1​
Hera​
2020​
ARMv8​
1115​
3.00​
371.69​
74%​
134%​
5​
Apple​
A10​
Hurricane​
2016​
ARMv8​
770​
2.34​
329.06​
65%​
119%​
6​
Intel​
1065G7​
Icelake​
2019​
x86-64​
1252​
3.90​
321.03​
64%​
116%​
7​
ARM Cortex​
A78​
Hercules​
2020​
ARMv8​
918​
3.00​
305.93​
61%​
111%​
8​
Apple​
A9​
Twister​
2015​
ARMv8​
564​
1.85​
304.86​
61%​
110%​
9​
AMD​
3950X​
Zen 2​
2019​
x86-64​
1317​
4.60​
286.30​
57%​
103%​
10​
ARM Cortex​
A77​
Deimos​
2019​
ARMv8​
812​
2.84​
285.92​
57%​
103%​
11​
Intel​
9900K​
Skylake​
2018​
x86-64​
1384​
5.00​
276.80​
55%​
100%​
12​
AMD​
1800X​
Zen 1​
2017​
x86-64​
1073​
3.90​
275.13​
55%​
99%​
13​
ARM Cortex​
A76​
Enyo​
2018​
ARMv8​
720​
2.84​
253.52​
50%​
92%​
14​
Intel​
4770K​
Haswell​
2013​
x86-64​
966​
3.90​
247.69​
49%​
89%​
15​
Apple​
A8​
Typhoon​
2014​
ARMv8​
323​
1.40​
230.71​
46%​
83%​
16​
Intel​
3770K​
Ivy Bridge​
2012​
x86-64​
764​
3.50​
218.29​
43%​
79%​
17​
Apple​
A7​
Cyclone​
2013​
ARMv8​
270​
1.30​
207.69​
41%​
75%​
18​
Intel​
2700K​
Sandy Bridge​
2011​
x86-64​
723​
3.50​
206.57​
41%​
75%​
19​
ARM Cortex​
A75​
Prometheus​
2017​
ARMv8​
505​
2.80​
180.36​
36%​
65%​
20​
ARM Cortex​
A73​
Artemis​
2016​
ARMv8​
380​
2.45​
155.10​
31%​
56%​
21​
Intel​
E6600​
Core2​
2006​
x86-64​
338​
2.40​
140.83​
28%​
51%​
22​
AMD​
FX-8350​
BD​
2011​
x86-64​
566​
4.20​
134.76​
27%​
49%​
23​
AMD​
Phenom 965 BE​
K10.5​
2006​
x86-64​
496​
3.70​
134.05​
27%​
48%​
24​
ARM Cortex​
A72​
Maya​
2015​
ARMv8​
260​
2.00​
130.00​
26%​
47%​
25​
AMD​
Athlon 64 X2 3800+​
K8​
2005​
x86-64​
207​
2.00​
103.50​
21%​
37%​
26​
ARM Cortex​
A55​
Ananke​
2017​
ARMv8​
178​
1.80​
98.67​
20%​
36%​
27​
ARM Cortex​
A53​
Apollo​
2012​
ARMv8​
148​
1.80​
82.22​
16%​
30%​
28​
Intel​
Pentium D​
P4​
2005​
x86-64​
228​
3.40​
67.06​
13%​
24%​



It's impressive how fast are evolving the generic Cortex cores:
  • A72 (2015) which can be found in most SBC has 1/3 of IPC of new Cortex X1 - They trippled IPC in just 5 years.
  • A73 and A75 (2017) which is inside majority of Android smart phones today has 1/2 IPC of new Cortex X1 - They doubled IPC in 3 years.

Comparison how x86 vs. Cortex cores:
  • A75 (2017) compared to Zen1 (2017) is loosing massive -34% PPC to x86. As expected.
  • A77 (2019) compared to Zen2 (2018) closed the gap and is equal in PPC. Surprising. Cortex cores caught x86 cores.
  • X1 (2020) is another +30% IPC over A77. Zen3 need to bring 30% IPC jump to stay on par with X1.

Comparison to Apple cores:
  • AMD's Zen2 core is slower than Apple's A9 from 2015.... so AMD is 4 years behind Apple
  • Intel's Sunny Cove core in Ice Lake is slower than Apple's A10 from 2016... so Intel is 3 years behind Apple
  • Cortex A77 core is slower than Apple's A9 from 2015.... but
  • New Cortex X1 core is slower than Apple's A11 from 2017 so ARM LLC is 3 years behind Apple and getting closer
 

Richie Rich

Senior member
Jul 28, 2019
421
196
76
This table is the answer to question why Apple begins transition to ARM:
  • - their Apple's PPC/IPC is best on the world and has technical advantage at least 4 years to x86
  • - other ARM cores (like A78 or X1) starts to outperform x86 CPUs very bad. So other ARM devices like MS Surface X or Samsung Galaxy Book S with new Snapdragon based on Cortex X1 would outperform Apple's x86 laptops based on Intel.

i want to see MPC from Tron in this list
Bring the GeekBench 5.1 ST result and I'll put into the chart. No problem.
 

Richie Rich

Senior member
Jul 28, 2019
421
196
76
Better is PPC/IPC performance development in time. You can see how dangerous ARM cores will be in near future:
  • Apples in-house ARM cores took PPC leadership since 2015 A9 and their IPC increase are monstrous
  • ARM Holding's generic Cortex cores reached x86 PPC levels in 2018 (A76 in Graviton2) and still accelerating, bringing even higher IPC increase than Apple
  • Intel and AMD delivers approximately same PPC/IPC increase in time. Problem is that this gain is just 1/3 of what delivers ARM cores. Clocks hit the wall so they have to bring performance only via IPC. And they are bad in this area.
  • Based on this trends whole x86 platform is already outperformed by ARM cores (valid for laptops and servers where nobody can run 4.5 GHz constant clock speeds as in desktop). If x86 vendors won't speed up their IPC development in really massive way then their future is very dark IMHO.

TOP10PPC_graph.png
 

IntelUser2000

Elite Member
Oct 14, 2003
6,825
1,387
136
XPS 13 1065G7 - 1461 @ 3.9GHz, PPC 375
Unknown Intel CPU - 1153 @ 3.2GHz, PPC 360.
Ryzen 3300X - 1401 @ 4.35GHz, PPC 322.07
EPYC 7702P
- 1075 @ 3.35GHz, PPC 320.9
i7 8700 - 290 @ 1GHz, PPC 290

-Zen 1 has 12% higher PPC than Cortex A77.
-X1 is 15% faster than Zen 1.
-Sunny Cove is just ahead of X1.
-A13 has 34% higher PPC compared to Sunny Cove.
-AMD is only 1 year behind Apple.
 
Last edited:

lobz

Golden Member
Feb 10, 2017
1,194
1,156
106
XPS 13 1065G7 - 1461 @ 3.9GHz, PPC 375
Unknown Intel CPU - 1153 @ 3.2GHz, PPC 360.
Ryzen 3300X - 1401 @ 4.35GHz, PPC 322.07
EPYC 7702P
- 1075 @ 3.35GHz, PPC 320.9
i7 8700 - 290 @ 1GHz, PPC 290

-Zen 1 has 12% higher PPC than Cortex A77.
-X1 is 15% faster than Zen 1.
-Sunny Cove is just ahead of X1.
-A13 has 34% higher PPC compared to Sunny Cove.
-AMD is only 1 year behind Apple.
This sums up the idiocy of this thread perfectly.

Again, the idea of comparing the ISAs is very exciting to me. However, the way OP's using twisted data to promote his agenda/delirium makes me sick.
 

Richie Rich

Senior member
Jul 28, 2019
421
196
76
XPS 13 1065G7 - 1461 @ 3.9GHz, PPC 375
Unknown Intel CPU - 1153 @ 3.2GHz, PPC 360.
Ryzen 3300X - 1401 @ 4.35GHz, PPC 322.07
EPYC 7702P
- 1075 @ 3.35GHz, PPC 320.9
i7 8700 - 290 @ 1GHz, PPC 290
These numbers are invalid and you know that. Ryzen 3300X@4.35 GHz cannot have higher score than Ryzen 3950X@4.6 GHz, that's pure nonsense. Ryzen 3300X is probably heavily OCed so using much higher clock than reported 4.35 GHz. That's the problem with Geekbench - you never know the clock and OC status.


-Zen 1 has 12% higher PPC than Cortex A77.
-X1 is 15% faster than Zen 1.
-Sunny Cove is just ahead of X1.
-A13 has 34% higher PPC compared to Sunny Cove.
-AMD is only 1 year behind Apple.
Even with your twisted data PPC 322 pts/GHz the AMD's Zen2 is way behind Apple A10 Hurricane from 2016 (329 pts/GHz) resulting in AMD is three years behind Apple, not one as you mentioned.

If you look at SPECint2006 PPC collected from Anandtech the GeekBench 5.1 PPC correlates to SPECint2006 PPC pretty well. Despite fact that SPECint tests only INT and not FPU like GB the PPC results are surprisingly similar:

Pos
Man
CPU
Core
Year
ISA
SPEC Score
GHz
PPC (score/GHz)
Relative
Relative
1​
Apple​
A13​
Lightning​
2019​
ARMv8​
52.82​
2.65​
19.93​
100.0%​
183.6%​
2​
Apple​
A12​
Vortex​
2018​
ARMv8​
45.32​
2.53​
17.91​
89.9%​
165.0%​
3​
Apple​
A11​
Monsoon​
2017​
ARMv8​
36.80​
2.39​
15.40​
77.2%​
141.8%​
4​
ARM Cortex​
X1​
Hera​
2020​
ARMv8​
45.76​
3.00​
15.25​
76.5%​
140.5%​
5​
ARM Cortex​
A78​
Hercules​
2020​
ARMv8​
37.66​
3.00​
12.55​
63.0%​
115.6%​
6​
Apple​
A10​
Hurricane​
2016​
ARMv8​
29.35​
2.34​
12.54​
62.9%​
115.5%​
7​
Intel​
1065G7​
Icelake​
2019​
x86-64​
47.70​
3.90​
12.23​
61.4%​
112.7%​
8​
ARM Cortex​
A77​
Deimos​
2019​
ARMv8​
33.32​
2.84​
11.73​
58.9%​
108.1%​
9​
Apple​
A9​
Twister​
2015​
ARMv8​
21.49​
1.85​
11.61​
58.3%​
107.0%​
10​
AMD​
3950X​
Zen 2​
2019​
x86-64​
50.02​
4.60​
10.87​
54.6%​
100.2%​
11​
Intel​
9900K​
Skylake​
2018​
x86-64​
54.28​
5.00​
10.86​
54.5%​
100.0%​
12​
ARM Cortex​
A76​
Enyo​
2018​
ARMv8​
26.65​
2.84​
9.38​
47.1%​
86.4%​
 

french toast

Senior member
Feb 22, 2017
968
773
106
I think the subject is very interesting, the data is very interesting, but that's it, interesting.
Nothing concrete can be taken from this for all sorts of reasons stipulated, cores are designed for different power and silicon targets, Apple can dedicate resources from phone sales to offset the the huge expense of designing such a massive core with massive cache to run at modest clocks at short bursts.
Apple cores don't have to scale to a massive gamut of form factors, TDPs and frequencies.
Apple cores are probably designed to run these benchmarks inside their cache and would likely suffer if forced to run a wide selection of apps that exceeded it's cache. it is clear Apple designs its cores and optimises the hell out of its software to run a narrow set of benchmarks extremely well.
This discussion is an interesting debate, but as others have stated nothing can be taken from this as to say 'X is better than Y'.. Not with this limited information.
I mean Linux and Windows benchmarks and different compilers throw wildly different results, the variance in software alone could throw the scores off by 50% easily, making this a fun exercise but nothing more.

Personally I think Apple could design a core for desktop that would beat out Sunny Cove and Zen 2 at 15-125w+ TDP, whether they could do this at a profit is another matter, but this data certainly doesn't show that and I doubt very much Apples current mobile cores could match top X86 current designs at their top performance running the same software across a wide selection of apps and usage scenarios, let alone having to scale the Watts and form factor like they do or make a profit on each die sold like they do.

So OP I applaud your efforts on this and I will follow this thread, please add SpecINT and SpecFP ongoing.
 

Richie Rich

Senior member
Jul 28, 2019
421
196
76
I think the subject is very interesting, the data is very interesting, but that's it, interesting.
...
So OP I applaud your efforts on this and I will follow this thread, please add SpecINT and SpecFP ongoing.
Thank a lot. It took a lot of effort to dive into and gather all data. Of course these SPEC and GeekBench data are just tip of the iceberg in the water. Everybody have to take this into account.

However I found interesting to compare how cheap smart phone SoC based on A53 stands against desktop/laptop class CPUs. Yeah, and it's really slow :D Or when somebody complains that Raspberry Pi 4 is slow garbage in compare to PC... and he is right, the A72 is really slow due to low PPC/IPC and it's 1.5 GHz nails that even more (I tried run Blender on RPI4 and compare it to my 3700X and it basicaly match the GB results, FYI render of BMW scene took 5 hours in single core :D )

Another interesting fact is that ARM has a massive range of performance you can buy today. From really slow garbage (and consuming mW) up to monstrous Apple A13 with it's massive cache system and super wide core containing 6xALUs and almost double PPC/IPC than x86 CPUs (and outperforming Ryzen 3950X@4.6 GHz).

I let everybody to make his own conclusion from the numbers. Because two years ago if somebody told me that Apple has the most advanced CPU on the world with 6xALUs then I'd laugh at him as well. But the numbers speaks for itself. BTW Apple has 6xALU monster core since 2017 (A11 Monsoon core), still unbeaten PPC-wise even today. And it looks like that Golden Cove neither Zen3/Zen4 won't able to beat that old Apple core in PPC. Pretty interesting fact estimation.

There are tons of interesting conclusions in those data.
 
Last edited:

Richie Rich

Senior member
Jul 28, 2019
421
196
76
Can't wait to see A14 in a laptop clobbering a 3950X.
No need to wait for A14. Ryzen 3950X@4.6-4.7 GHz is beaten by A13@2.65 GHz in SPECint2006 today:
  • A13 has score 52.82 pts
  • Ryzen has....... 49.02 pts
A14 will destroy any x86 CPU in ST load. Higher IPC by 10% and significantly higher clocks due to 5nm (2.65 -> 2.9 GHz is realistic expectation for iPhone, MacBooks could go up to 3.1 GHz, that's 17% up). IPC and clocks combined 1.1 * 1.17 = 29% higher ST performance. That's estimated SPECint score of 67.98 pts = massive improvement.

 

defferoo

Junior Member
Sep 28, 2015
18
10
81
as much as this is not a perfect comparison, it’s still an interesting set of observations.

I guess we’ll really find out how good Apple’s CPUs are once we get native ARM macOS binaries running on Apple Silicon. At the moment there are a lot of excuses about why we can’t compare benchmarks across desktop and mobile, but I kind of have a feeling we’ll be pretty surprised later this year.
 

Carfax83

Diamond Member
Nov 1, 2010
5,880
565
126
No need to wait for A14. Ryzen 3950X@4.6-4.7 GHz is beaten by A13@2.65 GHz in SPECint2006 today:
  • A13 has score 52.82 pts
  • Ryzen has....... 49.02 pts
A14 will destroy any x86 CPU in ST load. Higher IPC by 10% and significantly higher clocks due to 5nm (2.65 -> 2.9 GHz is realistic expectation for iPhone, MacBooks could go up to 3.1 GHz, that's 17% up). IPC and clocks combined 1.1 * 1.17 = 29% higher ST performance. That's estimated SPECint score of 67.98 pts = massive improvement.
I think you missed his point. The Apple A series is highly optimized for single threaded workloads, unlike the 3950x and other high end x86-64 CPUs which are highly optimized for multithreaded workloads as well.

So continuously quoting the SPECint single threaded benchmark as indicative of future performance and assuming that the Apple A series is going to carry its high single thread performance into other platforms are just assumptions at this point.

That said, I am glad this is happening. After all the debates we've had over ARM and x86-64, we're finally going to see who's right and who's wrong. And not only that, whatever core Apple uses for laptops and desktops in the future won't be compared against the positively decrepit Skylake cores that Intel has been using for the last thousand five years, but a much more powerful x86-64 core than anything out now.
 

DrMrLordX

Lifer
Apr 27, 2000
15,742
4,706
136
That said, I am glad this is happening. After all the debates we've had over ARM and x86-64, we're finally going to see who's right and who's wrong.
Finally. Took long enough for Apple to expose their hardware to the possibility of real testing. I sure hope Dr. Cutress and Andrei are willing to compile some of their own benches for Apple's new ARM-based Macs instead of sticking to the applications that have native, closed-source vendor support.
 

french toast

Senior member
Feb 22, 2017
968
773
106
Well zen 3 and Willow Cove will be waiting for this new Apple core.
A better comparison would probably be Apple's 2021 core vs Zen 4 on N5P and a 7nm Golden Cove?.. Let's say at 25-28w (realistic battle ground).
 
  • Like
Reactions: Tlh97

Richie Rich

Senior member
Jul 28, 2019
421
196
76
Well zen 3 and Willow Cove will be waiting for this new Apple core.
A better comparison would probably be Apple's 2021 core vs Zen 4 on N5P and a 7nm Golden Cove?.. Let's say at 25-28w (realistic battle ground).
Do not even think that any x86 CPU can compete against Apple's ARM CPUs in 25W laptop TDP.

Current A13@ 2.65 GHz in iPhone at 5W TDP is beating in ST load AMD Ryzen 3950@ 4.6 GHz in desktop 105W TDP.

Apple's MacBooks with 12-core A14 (8-big cores, 4 small cores) will devastate every desktop in ST and all 8-core Ryzens and 9900K in MT too. All this despite laptop TDP. With exception of 5GHz all core 9900K, but that's rare machine with TDP > 250W.

And you think that you will be safe with 12-core Ryzen 3900X? Those 4 little Apple efficient cores has PPC/IPC similar to Intel Sandy Bridge. There is high chance that Apple A14 with 8-big cores will threaten 12-core desktop chips.

"the new Thunder cores represent a 2.5-3x performance lead while at the same time using less than half the energy."

Let do the math and predict Geekbench score:
  • 12-core Ryzen 3900X..... PPC = 286 pts/GHz * 12 cores * 4.0 GHz all core clock = 13 728 pts
  • 8-big cores A14X .... PPC +10% over A13 = 502*1.1 = 552 pts/GHz .... * 8 cores * 3.1 = 13 694 pts
  • 4-little cores A14X .... PPC 1/3 big core = 184 pts/GHz * 4 cores * 2.0 GHz = 1 472 pts
  • Overall GB5 score for 12-core A14X will be 13694 + 1472 = 15 166 pts
When you look into Geekbench database then:

To sum up: 12-core A14X in laptop is capable to achieve Geekbench MT score 15 166 pts and beat most 16-core Ryzens 3950X. The only problem is that such a A14X would have 50W TDP and this is probably too much for laptop. So realistically MacBooks will decrease clocks and beat just 12-core desktops. 16-core Ryzens will be beaten by iMacs.


Apple A14 is the nightmare before Christmas (for x86) :D
 

ASK THE COMMUNITY