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

[Richie Rich]

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 to 9900K​	Relative to Zen3​
1​	Nuvia​	(Est.)​	Phoenix (Est.)​	2021​	ARMv9.0​	2001​	3.00​	667.00​	241.0%​	194.1%​
2​	Apple​	A15 (est.)​	(Est.)​	2021​	ARMv9.0​	1925​	3.00​	641.70​	231.8%​	186.8%​
3​	Apple​	A14 (est.)​	Firestorm​	2020​	ARMv8.6​	1562​	2.80​	558.00​	201.6%​	162.4%​
4​	Apple​	A13​	Lightning​	2019​	ARMv8.4​	1332​	2.65​	502.64​	181.6%​	146.3%​
5​	Apple​	A12​	Vortex​	2018​	ARMv8.3​	1116​	2.53​	441.11​	159.4%​	128.4%​
6​	ARM Cortex​	V1 (est.)​	Zeus​	2020​	ARMv8.6​	1287​	3.00​	428.87​	154.9%​	124.8%​
7​	ARM Cortex​	N2 (est.)​	Perseus​	2021​	ARMv9.0​	1201​	3.00​	400.28​	144.6%​	116.5%​
8​	Apple​	A11​	Monsoon​	2017​	ARMv8.2​	933​	2.39​	390.38​	141.0%​	113.6%​
9​	Intel​	(Est.)​	Golden Cove (Est.)​	2021​	x86-64​	1780​	4.60​	386.98​	139.8%​	112.6%​
10​	ARM Cortex​	X1​	Hera​	2020​	ARMv8.2​	1115​	3.00​	371.69​	134.3%​	108.2%​
11​	AMD​	5900X (Est.)​	Zen 3 (Est.)​	2020​	x86-64​	1683​	4.90​	343.57​	124.1%​	100.0%​
12​	Apple​	A10​	Hurricane​	2016​	ARMv8.1​	770​	2.34​	329.06​	118.9%​	95.8%​
13​	Intel​	1065G7​	Icelake​	2019​	x86-64​	1252​	3.90​	321.03​	116.0%​	93.4%​
14​	ARM Cortex​	A78​	Hercules​	2020​	ARMv8.2​	918​	3.00​	305.93​	110.5%​	89.0%​
15​	Apple​	A9​	Twister​	2015​	ARMv8.0​	564​	1.85​	304.86​	110.1%​	88.7%​
16​	AMD​	3950X​	Zen 2​	2019​	x86-64​	1317​	4.60​	286.30​	103.4%​	83.3%​
17​	ARM Cortex​	A77​	Deimos​	2019​	ARMv8.2​	812​	2.84​	285.92​	103.3%​	83.2%​
18​	Intel​	9900K​	Coffee LakeR​	2018​	x86-64​	1384​	5.00​	276.80​	100.0%​	80.6%​
19​	Intel​	10900K​	Comet Lake​	2020​	x86-64​	1465​	5.30​	276.42​	99.9%​	80.5%​
20​	Intel​	6700K​	Skylake​	2015​	x86-64​	1032​	4.00​	258.00​	93.2%​	75.1%​
21​	ARM Cortex​	A76​	Enyo​	2018​	ARMv8.2​	720​	2.84​	253.52​	91.6%​	73.8%​
22​	Intel​	4770K​	Haswell​	2013​	x86-64​	966​	3.90​	247.69​	89.5%​	72.1%​
23​	AMD​	1800X​	Zen 1​	2017​	x86-64​	935​	3.90​	239.74​	86.6%​	69.8%​
24​	Apple​	A13​	Thunder​	2019​	ARMv8.4​	400​	1.73​	231.25​	83.5%​	67.3%​
25​	Apple​	A8​	Typhoon​	2014​	ARMv8.0​	323​	1.40​	230.71​	83.4%​	67.2%​
26​	Intel​	3770K​	Ivy Bridge​	2012​	x86-64​	764​	3.50​	218.29​	78.9%​	63.5%​
27​	Apple​	A7​	Cyclone​	2013​	ARMv8.0​	270​	1.30​	207.69​	75.0%​	60.5%​
28​	Intel​	2700K​	Sandy Bridge​	2011​	x86-64​	723​	3.50​	206.57​	74.6%​	60.1%​
29​	ARM Cortex​	A75​	Prometheus​	2017​	ARMv8.2​	505​	2.80​	180.36​	65.2%​	52.5%​
30​	ARM Cortex​	A73​	Artemis​	2016​	ARMv8.0​	380​	2.45​	155.10​	56.0%​	45.1%​
31​	ARM Cortex​	A72​	Maya​	2015​	ARMv8.0​	259​	1.80​	143.89​	52.0%​	41.9%​
32​	Intel​	E6600​	Core2​	2006​	x86-64​	338​	2.40​	140.83​	50.9%​	41.0%​
33​	AMD​	FX-8350​	BD​	2011​	x86-64​	566​	4.20​	134.76​	48.7%​	39.2%​
34​	AMD​	Phenom 965 BE​	K10.5​	2006​	x86-64​	496​	3.70​	134.05​	48.4%​	39.0%​
35​	ARM Cortex​	A57 (est.)​	Atlas​	0​	ARMv8.0​	222​	1.80​	123.33​	44.6%​	35.9%​
36​	ARM Cortex​	A15 (est.)​	Eagle​	0​	ARMv7 32-bit​	188​	1.80​	104.65​	37.8%​	30.5%​
37​	AMD​	Athlon 64 X2 3800+​	K8​	2005​	x86-64​	207​	2.00​	103.50​	37.4%​	30.1%​
38​	ARM Cortex​	A17 (est.)​	​	0​	ARMv7 32-bit​	182​	1.80​	100.91​	36.5%​	29.4%​
39​	ARM Cortex​	A55​	Ananke​	2017​	ARMv8.2​	155​	1.60​	96.88​	35.0%​	28.2%​
40​	ARM Cortex​	A53​	Apollo​	2012​	ARMv8.0​	148​	1.80​	82.22​	29.7%​	23.9%​
41​	Intel​	Pentium D​	P4​	2005​	x86-64​	228​	3.40​	67.06​	24.2%​	19.5%​
42​	ARM Cortex​	A7 (est.)​	Kingfisher​	0​	ARMv7 32-bit​	101​	1.80​	56.06​	20.3%​	16.3%​

TOP 10 - Performance Per Area comparison at ISO-clock (PPA/GHz)

Copied from locked thread. They try to avoid people to see this comparison how x86 is so bad.[/B]

Pos​	Man​	CPU​	Core​	Core Area mm2​	Year​	ISA​	SPEC PPA/Ghz​	Relative​
1​	ARM Cortex​	A78​	Hercules​	1.33​	2020​	ARMv8​	9.41​	100.0%​
2​	ARM Cortex​	A77​	Deimos​	1.40​	2019​	ARMv8​	8.36​	88.8%​
3​	ARM Cortex​	A76​	Enyo​	1.20​	2018​	ARMv8​	7.82​	83.1%​
4​	ARM Cortex​	X1​	Hera​	2.11​	2020​	ARMv8​	7.24​	76.9%​
5​	Apple​	A12​	Vortex​	4.03​	2018​	ARMv8​	4.44​	47.2%​
6​	Apple​	A13​	Lightning​	4.53​	2019​	ARMv8​	4.40​	46.7%​
7​	AMD​	3950X​	Zen 2​	3.60​	2019​	x86-64​	3.02​	32.1%​

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

GeekBench5.1 comparison from 6/22/2020:

• added Cortex X1 and A78 performance projections from Andrei here
• 2020 awaiting new Apple A14 Firestorm core and Zen3 core

Updated:



EDIT:
Please note to stop endless discussion about PPC frequency scaling: To have fair and clean comparison I will use only the top (high clocked) version from each core as representation for top performance.

 

[itsmydamnation]

Don't treat us like idiots. You know full well that Apple chips have been benchmarked via SPEC, via browsers, via my own Mathematica tests, and all results are consistent with GB5.
To keep whining that "GB5 is a single mobile focused benchmark" says a lot more about you than about GB5 or Apple chips.

If that's what you took from his post you just made yourself actually look like an idiot because you obviously lack comprehension. Is your name Donald?

User insults are not allowed. Additionally, this is not
the P&N forum. If you want to talk about Donald, post
there.

AT Mod Usandthem

 

[lobz]

@lobz Is that 2nd reply meant for me or Rich?

Oh my God, how has that happened 🤣 - corrected it!

 

[DrMrLordX]

So no, Blender is a far inferior benchmark to SPEC unless rendering is the main thing you care about.

Except that Blender can't be gamed by compilers (oops) and that unlike SPEC, Blender is actually multithreaded while SPEC just loads multiple instances of single-threaded benchmarks. Would you like to be the next one who gets to lecture the forum on why reviewers like Anandtech should scrap their entire benchmark suite and just run SPEC and Geekbench 5?

 

[Doug S]

Except that Blender can't be gamed by compilers (oops) and that unlike SPEC, Blender is actually multithreaded while SPEC just loads multiple instances of single-threaded benchmarks. Would you like to be the next one who gets to lecture the forum on why reviewers like Anandtech should scrap their entire benchmark suite and just run SPEC and Geekbench 5?

If Anandtech was running Blender as their only benchmark I would be saying they need to scrap that and go with a wider variety of tests. Either SPEC or GB would be superior to that single benchmark. I'm not saying SPEC or GB are better than an entire benchmark suite, of course they aren't, I'm saying either is better than Blender alone. It should be obvious to anyone with half a brain that running a benchmark suite that contains a couple dozen tests that all do different things is better than running a single test that does only one thing. Getting a couple dozen numbers to look at is better than getting one number to look at.

SPEC2006 has had compiler writers break a couple of the benchmarks, but those benchmarks were fixed/removed in the next version and AFAIK none of SPEC2017's tests have been broken yet. If you want to look at a single benchmark number that's game-proof, instead of Blender look at the gcc/llvm numbers in SPEC/GB5 and ignore the rest. Compilers have lots of difficult to predict branches and a large icache footprint, and provide probably the closest analogue to modern browser/GUI code.

 

[DrMrLordX]

If Anandtech was running Blender as their only benchmark I would be saying they need to scrap that and go with a wider variety of tests.

They aren't. Are we just now learning how sites like Anandtech conduct their benchmarks, and why?

 

[itsmydamnation]

So AMD claimed they have wider execution for Zen3 ( we will have to wait and see if more issue ports or just more capability per port) and it only accounted for ~3% of the ~20% IPC increase. With the biggest gains from decode and load+store. This must be really disappointing news for the arbitrary unit counters

 

[Carfax83]

So AMD claimed they have wider execution for Zen3 ( we will have to wait and see if more issue ports or just more capability per port) and it only accounted for ~3% of the ~20% IPC increase.

What does more "capability per port" mean in this context?

 

[Doug S]

They aren't. Are we just now learning how sites like Anandtech conduct their benchmarks, and why?

You're the one who threw out the ridiculous strawmen, then try to criticize me when I start a sentence with IF.

Is your day job political spin?

 

[itsmydamnation]

What does more "capability per port" mean in this context?

so AMD has 4 ALU ports but not all operations exist on all ports. AMD could increase number of issue ports or make each ALU more symmetric/capable . Both could be classed as widening the execution engine. My bet is more issue ports, it is probably cheaper to have extra simple ALU's that can absorb simple arithmetic/logic ops leaving more fully functional ALU's to handle complex ops then it is to make all ALU's handle all complex ops for example.

 

[tamz_msc]

Except that Blender can't be gamed by compilers (oops) and that unlike SPEC, Blender is actually multithreaded while SPEC just loads multiple instances of single-threaded benchmarks.

This isn't even remotely true.

 

[DrMrLordX]

So AMD claimed they have wider execution for Zen3 ( we will have to wait and see if more issue ports or just more capability per port) and it only accounted for ~3% of the ~20% IPC increase. With the biggest gains from decode and load+store. This must be really disappointing news for the arbitrary unit counters

Oh don't worry. Facts never stopped them.

Is your day job political spin?

No. Is yours?

How many people posting here honestly looked at anything I posted and thought, "gee this man is endorsing benchmarking a CPU with only Blender" when I have gone out of my way multiple times to indicate that any CPU-to-CPU comparison should involve multiple different applications, preferably those relevant to the target readership.

How could anyone possibly post here on Anandtech and not be familiar with how CPU benchmarking is normally carried out on a site like AT? Anyone who has followed AT over the decades should immediately be able to spot the problem with an IPC list that uses only one data point as its basis.

This isn't even remotely true.

Elaborate. I was upbraided in the Graviton2 thread for thinking otherwise.

 

[tamz_msc]

Elaborate. I was upbraided in the Graviton2 thread for thinking otherwise.

Anandtech(and others) run in in that way(multiple instances of SPEC_rate) because it saves them time and has lower memory requirements. SPEC_speed is purely multithreaded and is an inverse ratio of system wall time over the reference time.

 

[Spartak]

I'm missing Tiger Lake in this list? How is willow cove on average vs sunny cove?

 

[DrMrLordX]

Anandtech(and others) run in in that way(multiple instances of SPEC_rate) because it saves them time and has lower memory requirements. SPEC_speed is purely multithreaded and is an inverse ratio of system wall time over the reference time.

Ah okay. See that improves the quality of your reply considerably. It's hard getting anyone to actually run SPEC_speed for their comparisons. To the best of my knowledge, SPEC_speed data hasn't been used in any of the IPC threads to date.

I'm missing Tiger Lake in this list? How is willow cove on average vs sunny cove?

I noticed it isn't there. Willow Cove actually has variable IPC vs Sunny Cove thanks to the way Intel changed the cache layout. It's faster in some things at the same clocks and slower in others.

 

[Spartak]

Ah okay. See that improves the quality of your reply considerably. It's hard getting anyone to actually run SPEC_speed for their comparisons. To the best of my knowledge, SPEC_speed data hasn't been used in any of the IPC threads to date.



I noticed it isn't there. Willow Cove actually has variable IPC vs Sunny Cove thanks to the way Intel changed the cache layout. It's faster in some things at the same clocks and slower in others.

yes i know but on average I believe it was about 2-3% faster in Geekbench..?

what i also dont understand in this list is the big performance delta between skylake and coffee lake (~7%). What I remember between kaby lake and skylake IPC was identical and coffee lake vs kaby/skylake was about 2-3%. Something is off.

 

[DrMrLordX]

what i also dont understand in this list is the big performance delta between skylake and coffee lake (~7%). What I remember between kaby lake and skylake IPC was identical and coffee lake vs kaby/skylake was about 2-3%. Something is off.

Well if the IPC is derived from ST scores only, then GB5 may be feeding off the larger L3 cache of the 9900k.

 

[Spartak]

Well if the IPC is derived from ST scores only, then GB5 may be feeding off the larger L3 cache of the 9900k.

My point is this 7% difference doesn't show in the geekbench 5 processor charts so it seems the OP is cherry picking GB5 results and not the median results provided by Geekbench themselves.

The 9900K scores an average of 1334, which scales pretty good from the baseline 3.6Ghz 1000 score for skylake processors.
The 6700K scores 1150.

1334/5 gives 266,8 and 1150/4.2 gives 273,8. So using the median results gives the 6700K a 3% higher IPC result than the 9900K.

Bottom line, the OP shouldnt be using cherry picked results since his results are 10% off from those provided by GB themselves.

 

[DrMrLordX]

My point is this 7% difference doesn't show in the geekbench 5 processor charts so it seems the OP is cherry picking GB5 results and not the median results provided by Geekbench themselves.

The 9900K scores an average of 1334, which scales pretty good from the baseline 3.6Ghz 1000 score for skylake processors.
The 6700K scores 1150.

1334/5 gives 266,8 and 1150/4.2 gives 273,8. So using the median results gives the 6700K a 3% higher IPC result than the 9900K.

Bottom line, the OP shouldnt be using cherry picked results since his results are 10% off from those provided by GB themselves.

Your post highlights much of what's wrong with the idea of using results from disparate clockspeeds in a single application that (for the purpose of this discussion) only runs on one core. Even within the same uarch, you have the absurd notion that a 6700k is somehow a faster/better CPU than a 9900k (hint: it isn't). Of course, on top of that, he's cherry-picking results while also rejecting allegedly cherry-picked results from other posters.

 

[Thala]

Intel's cache ratio is ~9/10, not 1/4. That was really just a side note anyway, the main thing is that the whole core was custom designed for high frequency compared to a standard A72 core.

There were NO design changes to the architecture, only backend changes. It was a standard A72 architecture wise.

 

[Thala]

If we assume that Apple's A14 is running at 2.99/3.00 GHz as the leaked GB5 benchmarks indicate, they should be able to reach at least 3.6 GHz in a higher power design. HPC cells will buy you about 10%

In TSMC N7 using low Vth cells gives your more than just 10%. And don't let me get started on the memory cells. There is easily factor 2 between the slowest low leakage and fastest high leakage cells.

You also completely ignoring the fact that a standard Cortex A72 could be clocked up to 4.2GHz as has been demonstrated by TSMC - of course with doing some backend optimizations.

 

[Doug S]

You also completely ignoring the fact that a standard Cortex A72 could be clocked up to 4.2GHz as has been demonstrated by TSMC - of course with doing some backend optimizations.

That is a lot less impressive than you make it out to be, because it was designed for what, 28nm? Of course it will clock higher in an improved process - especially when you can cherry pick the die and don't have to disclose how much power it was fed or what sort of cooling it was provided.