• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."

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

Page 13 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Thala

Golden Member
Nov 12, 2014
1,095
430
136
I understand that, but my point was that Intel and AMD prefer to have a higher operating frequencies compared to Apple. Intel and AMD both seem to be targeting around 5ghz with their CPUs. If they didn't think frequency was important, why didn't they design a CPU with lower clock speeds and higher IPC over the years? Same for IBM.
Thats because it is relatively easy to increase the frequency if you have the power headroom to do so. You can see that the cores operate in a similar frequency range at the same voltage and process.
 

Carfax83

Diamond Member
Nov 1, 2010
5,926
695
126
That is because anyone can submit scores using settings way outside of specification. In fact the highest scores are always coming from modders trying to push the limits to the extreme.
This is the highest Tiger Lake score I could find for single core. Seems like it was run on a Taiwanese OEM laptop:

Tiger Lake GB5 1620
 

Hitman928

Platinum Member
Apr 15, 2012
2,914
2,436
136
Why is this bogus? Looks pretty accurate to me.
It is not easy to see, because Geekbench database entries do not give enough information and in many cases this information is not even reliable.
Your other argument, that if a number is not coming from a review, because the reviewers do not use Geekbench i can absolutely follow. Such entries should at least marked in the table and the methodology how the number is obtained should be described. I do consider using the median as relatively reliable - but calculating the median is quite a bit more work.
So which is it, is the table accurate or can we not trust the numbers? What is the methodology for creating the table, what are the quality controls to make sure the numbers are accurate? What is the variability in scores? The numbers in the table are there completely on the whim of the person who created it with zero quality control. If you had actually read through the thread, you would know this.

Also,


All 3 on laptops using Renoir cores, all within boost spec, all showing higher ppc than what's in the table for Zen2 and that's with reduced L3 from the desktop products.


All 3 on server CPUs and server motherboards, all within boost spec, all showing much higher ppc than what's in the table for Zen2.

You can look for yourself, these are not abnormal scores. So are you going to argue that people are modding laptops and servers left and right just so they can score average scores on a mobile benchmarking app?

Edit: This doesn't even touch upon the fact that he's using a benchmark app designed primarily for phones and tablets to be the ultimate decider of IPC across all CPU segments.
 
Last edited:

Hitman928

Platinum Member
Apr 15, 2012
2,914
2,436
136
Thats because it is relatively easy to increase the frequency if you have the power headroom to do so. You can see that the cores operate in a similar frequency range at the same voltage and process.
The V/f curve is not the same for every architecture on the same process. The Apple CPUs start hitting severe diminishing returns on voltage increases before AMD/Intel CPUs do. Trying to push past that wall to match AMD and Intel CPUs in frequency will not go well, not without design changes to the Apple CPUs.
 

Thala

Golden Member
Nov 12, 2014
1,095
430
136
So which is it, is the table accurate or can we not trust the numbers? What is the methodology for creating the table, what are the quality controls to make sure the numbers are accurate? What is the variability in scores? The numbers in the table are there completely on the whim of the person who created it with zero quality control. If you had actually read through the thread, you would know this.
It is up to the OP the explain the scores not me - i did remember he mentioned reviews, which should anyway be the minimum standard.

Yet again your are linking random Geekbench entries.
 
  • Like
Reactions: Lodix

Thala

Golden Member
Nov 12, 2014
1,095
430
136
The Apple CPUs start hitting severe diminishing returns on voltage increases before AMD/Intel CPUs do. Trying to push past that wall to match AMD and Intel CPUs in frequency will not go well, not without design changes to the Apple CPUs.
And your evidence for this is what?
 
  • Like
Reactions: Lodix

Hitman928

Platinum Member
Apr 15, 2012
2,914
2,436
136
It is up to the OP the explain the scores not me - i did remember he mentioned reviews, which should anyway be the minimum standard.

Yet again your are linking random Geekbench entries.
I have provided more examples and data behind where the scores came from than the O.P. ever has. So do you stand by your statement that the table looks accurate or not?
 

Zucker2k

Golden Member
Feb 15, 2006
1,233
593
136
People wanted to cheat. I remember one guy downclocked his x86 CPU to 2.6 GHz similar to Apple, probably boosted uncore and DDR mem to maximum and claimed score higher about +20% than my table. This is cheating. Apple core can also provide higher IPC at 1 GHz. GB database is full of tweaked and OC'ed systems with wrongly reported frequency. I try to use GB results from reviews as much as possible to get reasonable numbers. I stated this table is chart of IPC under normal maximum performance to avoid down clocking (cheating).
So, basically x86 can gain 20%+ ipc in this benchmark by just reducing clocks? Meaning this app scales so poorly that the lower the clocks, the higher the ipc? That's a very unusual choice of benchmark app because the lowest clocked chips have an inherent advantage.
 

Thala

Golden Member
Nov 12, 2014
1,095
430
136
I pretty much doubt that this is a legit curve for N7P and A12, because this would mean Apple runs with VCC of around 1.1V, which is way outside the overdrive zone one would use for ultra mobile. Second for what cell distribution is this curve? For what corner is this curve - slow/slow perhaps?
Due to your semiconductor knowledge, i am confident you are aware of, that one can shift such a curve to the right by not using low-leakage/high-vt cells?

Anyway, this is pretty much off-topic.

So do you stand by your statement that the table looks accurate or not?
Sure, at first glance and without cross-checking any numbers in particular the order at least looks reasonable. There are minor things which are little surprising, as for instance Zen2 being in front of Cortex A77 - but its reasonably close.

ps. In addition i do not like that unreleased products are part of the table.
 
Last edited:

Carfax83

Diamond Member
Nov 1, 2010
5,926
695
126
Thats because it is relatively easy to increase the frequency if you have the power headroom to do so. You can see that the cores operate in a similar frequency range at the same voltage and process.
But the CPU microarchitecture has to be designed for high frequencies in mind no? It's not just a question of them pumping more voltage into the CPU for it to hit higher frequencies.
 

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
20,810
8,970
136
I pretty much doubt that this is a legit curve for N7P and A12, because this would mean Apple runs with VCC of around 1.1V, which is way outside the overdrive zone one would use for ultra mobile. Second for what cell distribution is this curve? For what corner is this curve - slow/slow perhaps?
Due to your semiconductor knowledge, i am confident you are aware of, that one can shift such a curve to the right by not using low-leakage/high-vt cells?

Anyway, this is pretty much off-topic.



Sure, at first glance and without cross-checking any numbers in particular the order at least looks reasonable. There are minor things which are little surprising, as for instance Zen2 being in front of Cortex A77 - but its reasonably close.

ps. In addition i do not like that unreleased products are part of the table.
Did you read my reply on the last page ? all of this becomes a mute question if you are comparing apples to apples.
 
  • Like
Reactions: Tlh97 and Drazick

Carfax83

Diamond Member
Nov 1, 2010
5,926
695
126
I pretty much doubt that this is a legit curve for N7P and A12, because this would mean Apple runs with VCC of around 1.1V, which is way outside the overdrive zone one would use for ultra mobile.
How often would the A12/A13/A14 be hitting max frequency for long enough periods of time to cause thermal problems on a strictly mobile platform?

Very little if any I'd wager. Unless the owner is gaming for long periods of time or running consecutive benchmarks, the CPU will probably never be sucking down such a high voltage on a consistent basis.
 
  • Like
Reactions: Tlh97

coercitiv

Diamond Member
Jan 24, 2014
3,936
4,229
136
Yet again your are linking random Geekbench entries.
First defending GB as a good indicator of overall CPU performance (which imho is fair game), then pushing for lack of reliability of the benchmark when it comes to even multiple "random" server class entries.

So, just to make it clear - GB5 is a good tool for estimating server CPU performance but only as long as the result comes from a reputable source, unless the CPU is unreleased in which case an anonymous forum used called Richie Rich is to be considered reputable source. Scientific criteria on one side, personal preference on the other.

FYI - the OP refused x86 server class GB5 results not based on reliability, but rather based on lower clocks used by most x86 server CPUs combined with the arbitrary rule of having only 1 entry per architecture.

So, basically x86 can gain 20%+ ipc in this benchmark by just reducing clocks? Meaning this app scales so poorly that the lower the clocks, the higher the ipc? That's a very unusual choice of benchmark app because the lowest clocked chips have an inherent advantage.
It's not the app's fault, think about what reducing clocks does to a computer with respect to CPU / memory subsystem relation. We're not comparing absolute performance or at least some proper relative indicator such as perf/watt (which would also put Apple cores first, mind you), it's an artificial comparison of PPC that heavily favors doing the work slowly so that the memory subsystem can keep up.

The reason for the PPC ranking is Apple and ARM products have traditionally been mobile oriented, and the main argument brought against them as server/desktop replacements in the forums was scaling (frequency, core count, interconnect). So therefore, perf/watt while being stunning for Apple at least, was not enough to lead imagination to new heights. Instead of waiting for more actual server/desktop silicon from Nuvia/Apple and other entitites, the OP created his own narrative in which revolutionary high performance ARM cores are always imminent, always at the next corner, and to support this scenario he chose to rely on PPC. This is also the reason we started seeing Nuvia and Apple future product estimates, an uncontrollable desire to make predictions happen.

You can witness his cognitive dissonance at work in the Apple A14 thread where even the remote possibility of A14 providing a generational leap in performance based mostly on higher clocks got him instantly tilted: first he rejected the GB results because of the clocks (he had no issue with the score), then quickly moved back to fantasy land with the wider A15 16c beast and the A19 12XALU juggernaut. If the present doesn't fit expectations, move to the future.

Semiconductor knowledge as well as this:
FYI, Thala was already presented with this information early this year.
 

naukkis

Senior member
Jun 5, 2002
365
208
116
, it's an artificial comparison of PPC that heavily favors doing the work slowly so that the memory subsystem can keep up.
But with Apple core's that's not the case. They are equal in performance wise to higher clocked x86-cpu's. Instead donwnclocked x86-cpus are those artificial comparison points, making cpu's slower to improve IPC.
 
  • Like
Reactions: Richie Rich

naukkis

Senior member
Jun 5, 2002
365
208
116
And Intel is extremely desperate to chase higher IPC - but with x86 that results to very big cpu cores which aren't power efficient - which makes them unpractical to about every use case. Willow Cove cores are too big to be competitive in laptop space - and with Golden cove cores are too big to be competitive even in desktop-class - resulting they have to rely big.Little scheme even on desktop. Intel should have also been thinking about buying ARM holdings - they will have very rough years to come against nVidia server products.
 
  • Like
Reactions: Richie Rich

Zucker2k

Golden Member
Feb 15, 2006
1,233
593
136
But with Apple core's that's not the case. They are equal in performance wise to higher clocked x86-cpu's. Instead donwnclocked x86-cpus are those artificial comparison points, making cpu's slower to improve IPC.
You're not going to gain 20%+ ipc on x86 platform by downclocking by 1GHz in Cinebench. What gives?
 

Richie Rich

Senior member
Jul 28, 2019
470
227
76
So which is it, is the table accurate or can we not trust the numbers? What is the methodology for creating the table, what are the quality controls to make sure the numbers are accurate? What is the variability in scores? The numbers in the table are there completely on the whim of the person who created it with zero quality control. If you had actually read through the thread, you would know this.

Also,


All 3 on laptops using Renoir cores, all within boost spec, all showing higher ppc than what's in the table for Zen2 and that's with reduced L3 from the desktop products.


All 3 on server CPUs and server motherboards, all within boost spec, all showing much higher ppc than what's in the table for Zen2.

You can look for yourself, these are not abnormal scores. So are you going to argue that people are modding laptops and servers left and right just so they can score average scores on a mobile benchmarking app?

Edit: This doesn't even touch upon the fact that he's using a benchmark app designed primarily for phones and tablets to be the ultimate decider of IPC across all CPU segments.
  • You claim that Renoir has 1254 pts @ 1.8 GHz ..... resulting in 697 pts/GHz. 16-core Ryzen 3950X has only 286 pts/GHz. So you claim that Renoir has 2.4x higher IPC? Really? Why AMD bother with Zen3 +15% IPC development when they can install Renoir everywhere and go 10-year vacation?
  • You claim Epyc has 1094 pts @ 2.25 GHz .... resulting in 486 pts/GHz, which is 1.7x higher IPC than 3950X. No comment.
  • you claim this is how we should do the IPC comparison?


This is exactly the reason why I refused to update the IPC table with all these crazy non-sense low-clocked numbers. I told people: create your own and better IPC table. Show me how better you are. None. Silence. Obviously they like to hate however smart enough knowing that their own IPC table full of these crazy results would discredit them, not me. So they stick with hate.

Running CPU at max speed:
  • has eliminates frequency uncertainty.
  • deliver max performance and that's we are searching for
  • means we compare IPC at max performance for given uarch - that's the goal

It's clear that most people is angry due to Apple's almost double IPC and much bigger 6xALU architecture in compare to their poor 4xALU x86 CPUs. All those results denials, doubting, claiming AISC circuits and hidden aggressivity is just like when you take a toy from little kid. x86 CPU sucks last 5years. Apple is the new uarch leader now. Get over it. And get used to it because Nuvia Phoenix with 8xALU+4xBr = 12-wide uarch is coming soon.
;)
 

coercitiv

Diamond Member
Jan 24, 2014
3,936
4,229
136
But with Apple core's that's not the case. They are equal in performance wise to higher clocked x86-cpu's. Instead donwnclocked x86-cpus are those artificial comparison points, making cpu's slower to improve IPC.
If and when Apple launches higher clocked desktop replacement SKUs based on the same arch used in the iPad, will you honestly ask the OP to use the lower PPC scores of the higher clocked SKUs instead? Think about that, the desktop part would likely crush the phone part in every single performance benchmark, yet the phone CPU will still win in PPC. Do you remember the time when the OP repeatedly tried to convince us that a 2.6Ghz Apple laptop CPU with 8 cores would be faster than competing x86 CPUs because consumer loads would scale better with core count than ST performance? You probably don't, because it was just another of his failed attempts to change the narrative towards a hypothetical winning scenario of his.

That's the problem with focusing on the winner, the ranking system criteria become opportunistic. You have someone like Thala who warns against using random benchmark numbers but openly embraces performance estimates from a stranger. This thread is a church of science, no longer separates belief from rigorous testing & reasoning.

Personally I don't care who wins the performance race, actually I kinda' rooted for ARM, until the latest developments it was the more open and arguably competition friendly ISA. Now it's more of a wash, but still ok either way.
 
Apr 30, 2020
26
64
51
Thats because it is relatively easy to increase the frequency if you have the power headroom to do so. You can see that the cores operate in a similar frequency range at the same voltage and process.
That is really not true at all. Core design has a HUGE impact on clocks. Look back at AMD's Winchester vs. Orleans cores. Same process - but one maxed out at 2200 MHz, the other at 2600 MHz at the same power. Or how about we compare two different architectures on the same node - AMD Zen1 vs. AMD Polaris. Polaris struggles to achieve much more than ~1.3 GHz no matter how much power you give it.

Just because an Apple A-whatever can run at 2.6 GHz, doesn't means it'll hit 4 GHz even with all the power headroom in the world. If that were true, we'd already have 4-5 GHz "fast" ARM desktop CPUs. But we don't.
You claim that Renoir has 1254 pts @ 1.8 GHz ..... resulting in 697 pts/GHz. 16-core Ryzen 3950X has only 286 pts/GHz. So you claim that Renoir has 2.4x higher IPC? Really? Why AMD bother with Zen3 +15% IPC development when they can install Renoir everywhere and go 10-year vacation?
  • You claim Epyc has 1094 pts @ 2.25 GHz .... resulting in 486 pts/GHz, which is 1.7x higher IPC than 3950X. No comment.
  • you claim this is how we should do the IPC comparison?
This is exactly the reason why I refused to update the IPC table with all these crazy non-sense low-clocked numbers. I told people: create your own and better IPC table. Show me how better you are. None. Silence. Obviously they like to hate however smart enough knowing that their own IPC table full of these crazy results would discredit them, not me. So they stick with hate.
The problem is you are cherry picking clock speeds to what best suits your narrative. For example, in your chart, you assume the 3950X is pegged at 4.6GHz for the entire test, which is almost certainly not true. And then for these results that you do not like, you're assuming they're pegged at base clocks the entire time so they'd have ridiculously high numbers that you could just throw out - which is also not true. The Epyc 7742 has a 3.4 GHz max boost speed - yielding 321.7 pts/ghz - and that's assuming it stayed at that 3.4 GHz boost the entire time. The Renoir chips you also tossed out have a 4.2 Ghz max boost, yielding 298.5 pts/ghz - still better than the 3950X result you chose.

The 3950X result you chose is showing a 4% IPC increase over Zen 1 in your own chart. How you can even seriously entertain that as accurate?
 
Last edited:

NTMBK

Diamond Member
Nov 14, 2011
8,841
1,916
136
That is really not true at all. Core design has a HUGE impact on clocks. Look back at AMD's Winchester vs. Orleans cores. Same process - but one maxed out at 2200 MHz, the other at 2600 MHz at the same power. Or how about we compare two different architectures on the same node - AMD Zen1 vs. AMD Polaris. Polaris struggles to achieve much more than ~1.3 GHz no matter how much power you give it.

Just because an Apple A-whatever can run at 2.6 GHz, doesn't means it'll hit 4 GHz even with all the power headroom in the world. If that were true, we'd already have 4-5 GHz "fast" ARM desktop CPUs. But we don't.
Wait, I thought Winchester was 90nm and Orleans was 65nm?

But you make a good point- just look at Conroe. Same 65nm process as the Pentium D that it replaced, but it targeted higher IPC and lower clocks, so it never clocked as high- despite the higher performance.
 

ASK THE COMMUNITY