News Rocket Lake i7 Loses to Apple M1 in Single-Threaded Performance

atomicWAR

Glorious
Ambassador
Yeah this is not what I would call a "fair" result. Yes apple wins (save the i9 11900k) in single thread performance but your also not using the whole core on the x86 side since hyperthreading/symmetric multithreading is not being used for this test. I WOULD be more impressed if these scores counted the hyperthreading for the core being taxed which typically adds about 20% more performance. Until then it is an apples to oranges comparison that doesn't mean near as much as the article would like you to believe. Pun intended...
 
Last edited:

salgado18

Distinguished
Feb 12, 2007
928
373
19,370
I'd say it's a bit irrelevant, since you can only get a new Mac with an M1, and can't use the M1 outside a Mac. Nice to see the chip is fast, but until they compete for the same space, doesn't mean much.

Also, it's a brand new chip, while Rocket Lake is built upon older architectures and a worse node.
 

Brian28

Distinguished
Jan 28, 2016
48
15
18,565
Yeah this is not what I would call a "fair" result. Yes apple wins (save the i9 11900k) in single thread performance but your also not using the whole core on the x86 side since hyperthreading/symmetric multithreading is not being used for this test. I WOULD only be impressed if these scores counted the hyperthreading for the core being taxed which typically adds about 20% more performance. Until then it is an apples to oranges comparison that doesn't mean near as much as the article would like you to believe. Pun intended...
It's not to be taken in isolation, as the multi-core performance is very relevant. But it's absolutely fair, as there are still some workloads that are not designed to be multithreaded at all. That means they can't use HT, either. The test is called single-thread, not single-core.
 

atomicWAR

Glorious
Ambassador
It's not to be taken in isolation, as the multi-core performance is very relevant. But it's absolutely fair, as there are still some workloads that are not designed to be multithreaded at all. That means they can't use HT, either. The test is called single-thread, not single-core.

I am not saying your argument isn't without merit but when you are comparing threads and only using only part of one CPU core for the test and the whole core on the other CPU, the test is really incomplete in my eyes. Heck imagine throwing an IBM Power 10 CPU where there are 8 threads per core and only counting one. Again it would not be without merit but it would paint a very incomplete picture.
 
Last edited:
Yeah this is not what I would call a "fair" result. Yes apple wins (save the i9 11900k) in single thread performance but your also not using the whole core on the x86 side since hyperthreading/symmetric multithreading is not being used for this test. I WOULD only be impressed if these scores counted the hyperthreading for the core being taxed which typically adds about 20% more performance. Until then it is an apples to oranges comparison that doesn't mean near as much as the article would like you to believe. Pun intended...
Hyperthreading only kicks in if a single thread doesn't saturate the execution resources of a core and if another thread can fill in the gaps. If the test can saturate the execution resources of the core, then Hyperthreading is irrelevant.

EDIT: On another note, Hyperthreading is irrelevant anyway since if a single thread is spawned for this test, it's not going to trigger Hyperthreading anyway.
 
Straight from the passmark site:
The sample rate for the 11th gen is too small for accuracy.
It's impressive that the M1 runs at lower Mhz and is still up there.
11th gen is still severely nerfed if they are actually running at the stated Hz, 3.5 instead of 5.3 is huuuuge.
 

andrep74

Distinguished
Apr 30, 2008
11
13
18,515
Doesn't matter if it's fair or not. If at the end of the day someone gets their job done on a machine that costs 1/2 as much, on a battery that lasts twice as long, and runs just as fast, they're not going to lament the lack of hyperthreading.

What Intel and AMD really need to worry about is M1X or M2. It's amazing that these chips emulate older architectures and worse nodes at or near the same performance level as the current-gen x86 chips. One has to wonder if Apple's clout has led to their current monopoly on 5nm and whether that has the most significant impact on overall performance, and, therefore, their future.
 
What Intel and AMD really need to worry about is M1X or M2. It's amazing that these chips emulate older architectures and worse nodes at or near the same performance level as the current-gen x86 chips.
Apple isn't using emulation. They're using what's called Binary Translation. If you understand the ISA well enough, which Apple has plenty of experience in that regard, then you can effectively "recompile" the executable in real-time and run with minimal performance loss.

It's the same principle as how .NET and Java based apps work.
 

spongiemaster

Admirable
Dec 12, 2019
2,273
1,277
7,560
Doesn't matter if it's fair or not. If at the end of the day someone gets their job done on a machine that costs 1/2 as much, on a battery that lasts twice as long, and runs just as fast, they're not going to lament the lack of hyperthreading.

What Intel and AMD really need to worry about is M1X or M2. It's amazing that these chips emulate older architectures and worse nodes at or near the same performance level as the current-gen x86 chips. One has to wonder if Apple's clout has led to their current monopoly on 5nm and whether that has the most significant impact on overall performance, and, therefore, their future.
Holy lurker, Batman. You've been here almost 13 years and that was your first post?
 
  • Like
Reactions: Makaveli

Lorien Silmaril

Distinguished
Jul 18, 2014
36
32
18,570
Yeah this is not what I would call a "fair" result. Yes apple wins (save the i9 11900k) in single thread performance but your also not using the whole core on the x86 side since hyperthreading/symmetric multithreading is not being used for this test. I WOULD only be impressed if these scores counted the hyperthreading for the core being taxed which typically adds about 20% more performance. Until then it is an apples to oranges comparison that doesn't mean near as much as the article would like you to believe. Pun intended...

I don't feel like you understand what "single thread" testing means. Intel can't boast about single thread performance then lament not being able to hyperthread because, you know, there's only one thread.

the comparison was totally fair.
 
I don't feel like you understand what "single thread" testing means. Intel can't boast about single thread performance then lament not being able to hyperthread because, you know, there's only one thread.

the comparison was totally fair.
Who buys an 8 core but then turns off everything running to only run a single thread?!
I'm saying that because for intel you can maintain the absolute highest single core boost on one core even with all other cores loaded, can the M1 do the same?!
 

watzupken

Reputable
Mar 16, 2020
1,007
505
6,070
Who buys an 8 core but then turns off everything running to only run a single thread?!
I'm saying that because for intel you can maintain the absolute highest single core boost on one core even with all other cores loaded, can the M1 do the same?!
People don't deliberately disable cores to run just on a single core, but applications out there may not always be optimized for multicores well. So if an app can only utilize 2 or 4 cores, the other 4 cores will be at idle, and performance will be purely based on how fast the 2 or 4 cores can deliver.

Also, you should consider the power discrepancy between the 2 chips. It is true that Rocket Lake can boost single core clockspeed a lot more, but at the expense of power. If you are looking from a power efficiency standpoint, Apple's M1 on 5nm will absolutely tounce Intel's ancient 14nm.

In the first place, I would not even compare these 2 chips since one is more suited for small and mobile devices like the Mac Mini and Macbooks, while the other is clearly confined to desktop given the high power consumption and heat output. A better product from Intel to compare would probably be Tiger Lake.
 

anonymousdude

Distinguished
People don't deliberately disable cores to run just on a single core, but applications out there may not always be optimized for multicores well. So if an app can only utilize 2 or 4 cores, the other 4 cores will be at idle, and performance will be purely based on how fast the 2 or 4 cores can deliver.

Also, you should consider the power discrepancy between the 2 chips. It is true that Rocket Lake can boost single core clockspeed a lot more, but at the expense of power. If you are looking from a power efficiency standpoint, Apple's M1 on 5nm will absolutely tounce Intel's ancient 14nm.

In the first place, I would not even compare these 2 chips since one is more suited for small and mobile devices like the Mac Mini and Macbooks, while the other is clearly confined to desktop given the high power consumption and heat output. A better product from Intel to compare would probably be Tiger Lake.

Agreed. This comparison is both incomplete and largely irrelevant. They're gunning for different markets at least for now. Who knows what Apple will decide to do with the future chips though. All we can draw from this single benchmark is that Rocket Lake and the M1 have similar performance in PassMark. Whether or not this really means anything in the real world remains to be seen.
 

usiname

Prominent
BANNED
Feb 27, 2020
92
20
535
Comparing Intel vs AMD with single thread benchmarks is fine, because both have multithreading. Comparing m1 vs both of them in single thread is wrong, because you disable on of the biggest advantage of x86, and in real world m1 is not even close. 15w 5800U destroyed m1 in cinebench, with less power and 7nm. ARM can't compete with x86 in heavy multithreading tasks and won't in near future, it could be faster is specific tasks but x86 on 7nm is still faster and if AMD was on 5nm then m1 will not just lose in benchmarks like now, but in battery life too, because even now amd is equivalent.
 
Last edited:
People don't deliberately disable cores to run just on a single core, but applications out there may not always be optimized for multicores well. So if an app can only utilize 2 or 4 cores, the other 4 cores will be at idle, and performance will be purely based on how fast the 2 or 4 cores can deliver.
I'm talking about turning off things that are running, stop using all other software to run your single threaded app all alone by itself.
 
Apple isn't using emulation. They're using what's called Binary Translation. If you understand the ISA well enough, which Apple has plenty of experience in that regard, then you can effectively "recompile" the executable in real-time and run with minimal performance loss.

It's the same principle as how .NET and Java based apps work.

Yes and no. With il like .net and java all the commands are high level and not machine specific.

To hsl on gpu's and ucode instruction sets on cpu are broken down into a series of micro ops. Now there is a tremendous amount of ucode which is extremely hardware dependent. Ie: addressing modes and virtual memory. Or avx instructions, machine specific registers like interrupts or error code trap recoveries. There may be no equivalent hardware and emulating it can be both time consuming or near impossible.

Think of it akin to.emulating floating point math in FORTRAN before math coops became standard. It took computer's forever to do floating point. Same thing can be said about many mimd ops.
 

atomicWAR

Glorious
Ambassador
Hyperthreading only kicks in if a single thread doesn't saturate the execution resources of a core and if another thread can fill in the gaps. If the test can saturate the execution resources of the core, then Hyperthreading is irrelevant.

EDIT: On another note, Hyperthreading is irrelevant anyway since if a single thread is spawned for this test, it's not going to trigger Hyperthreading anyway.

I understand how the allocation of resources work and when hyperthreading kicks in, or doesn't. And your right if a core is truly saturated there is nothing left for the hyperthread. All I am saying is frequently workloads, even benchmarks, don't tend to actually use all of those resources as the cores are designed in such a fashion to help ensure they have some resources left for the hyperthread. But your not wrong on what you said as you stated it.

On your edit... that was the whole point of my post to begin with. Using a test that only spawns a single thread off a hyperthreaded/symmetric multithreading CPU core is by its nature is potentially not using the the entirety of the CPU core's resources/compute power. There is no question that the M1 has extremely potent single threaded grunt. It was never my intent to imply otherwise. My point was I didn't feel the comparison was an apples to apples one and therefore not entirely a fair comparison to make. I would just like to see some benchmarks where the M1 goes against an x86 CPU's entire core, both threads. ST absolutely matters but so does using the whole core. I think having a comparison of both ST and total core performance would be more revealing about the position Apple is actually in, and x86 by extension. That is not an unreasonable stance to take.
 

atomicWAR

Glorious
Ambassador
I don't feel like you understand what "single thread" testing means. Intel can't boast about single thread performance then lament not being able to hyperthread because, you know, there's only one thread.

the comparison was totally fair.

I very much understand what single threaded means, I have been a tech for 22+ years. I have worked on everything from mainframes, military hardware and all the way down to personal computers and most everything in between. So I think it is totally reasonable to want to see a total core usage when comparing CPUs. And yes I do understand that means going beyond one thread...you opinion is valid but it doesn't make mine without merit either. The two need not be mutually exclusive. ST counts for a lot but so does the total work a core can do as well. I guess my point is I would like to see a full x86 core (thread 0 and 1) be measured against Apples single threaded core as well as the already listed ST performance. The picture is very incomplete without that comparison. So while you see the comparison as fair, I see it as skewed in favor of the M1 due to the testing periameters. Let me be clear I am no fan of Intel. And If I am being honest, I would love to see them get smoked in both benchmarking scenarios so we will see increased competition in the CPU spacce. I don't, sadly, see the M1 winning against as many x86 cpus on a full core level but until we get some benchmarks that show these kinds of results we won't know and I'll be left feeling like the comparison isn't totally fair to either platform in the long run.
 

JayNor

Reputable
May 31, 2019
426
85
4,760
"Especially given that the M1’s pro-level successor, the M1X, is already in the works. '

Perhaps the right comparison would be vs TGL. Rocket Lake is a desktop chip.

Intel's Alder Lake chips are also "in the works".
 

Lorien Silmaril

Distinguished
Jul 18, 2014
36
32
18,570
I very much understand what single threaded means, I have been a tech for 22+ years. I have worked on everything from mainframes, military hardware and all the way down to personal computers and most everything in between. So I think it is totally reasonable to want to see a total core usage when comparing CPUs. And yes I do understand that means going beyond one thread...you opinion is valid but it doesn't make mine without merit either. The two need not be mutually exclusive. ST counts for a lot but so does the total work a core can do as well. I guess my point is I would like to see a full x86 core (thread 0 and 1) be measured against Apples single threaded core as well as the already listed ST performance. The picture is very incomplete without that comparison. So while you see the comparison as fair, I see it as skewed in favor of the M1 due to the testing periameters. Let me be clear I am no fan of Intel. And If I am being honest, I would love to see them get smoked in both benchmarking scenarios so we will see increased competition in the CPU spacce. I don't, sadly, see the M1 winning against as many x86 cpus on a full core level but until we get some benchmarks that show these kinds of results we won't know and I'll be left feeling like the comparison isn't totally fair to either platform in the long run.
Sorry didn't mean to be rude.

My point is that x86 currently relies on SMT to saturate a core, but for applications that is single-threaded, it's not gonna take advantage of SMT anyway.

Yes you could argue that's not using the core to its "maximum potential" but it's more of a drawback of x86 itself, not RISC's fault for not having as much overhead; realistically if you are trying to find out how fast single-threaded workload can run on competing processors, this sort of comparison isn't unfair.

It sounds like you are looking for a core-to-core comparison (regardless of threads) and that's been done before, but you likely have read about it already:
 
  • Like
Reactions: atomicWAR

atomicWAR

Glorious
Ambassador
Sorry didn't mean to be rude.

My point is that x86 currently relies on SMT to saturate a core, but for applications that is single-threaded, it's not gonna take advantage of SMT anyway.

Yes you could argue that's not using the core to its "maximum potential" but it's more of a drawback of x86 itself, not RISC's fault for not having as much overhead; realistically if you are trying to find out how fast single-threaded workload can run on competing processors, this sort of comparison isn't unfair.

It sounds like you are looking for a core-to-core comparison (regardless of threads) and that's been done before, but you likely have read about it already:

No worries, it happens. Text doesn't always show intention as well as we would like at times. So I mean this truly, thank you. Sorry If i was overly defensive but I got a couple comments on this subject where folks either completely missed my point and/or they just exclaimed I don't understand single threaded when they clearly missed what I was getting at.

Back to our debate, I guess unfair wasn't the best wording on my part. I just feel like it paints a very incomplete picture of a core's capabilities. Just like it isn't RISC's fault it doesn't have as much over head...you can turn it around as well...it's not x86's fault they have learned to execute extra processes with unused resources/overhead. When they spend additional die space and extra power for an additional thread its reasonable it be included in comparisons. Especially when wattage is being compared. Though we all know who wins in wattage at the end of the day I guess. Hint its not an x86 CPU lol.

I am not saying ST doesn't have it's place, we all know it does. Plenty of apps only use one thread. Though I'd also argue your getting two single threaded apps, code allowing, to have work done on them at the same time on a dual threaded CPU. This is something that should not go unrecognized. Regardless when comparing CPU architectures, I generally don't think it is the best practice to hobble a core to compare it to another one. ST absolutely deserves mentioning, but so does the fully threaded core IMHO. I wouldn't want ST to be not be mentioned for example and only hear about the per core performance to turn things around. Anyways thats just me. I like a little more info, with my info. If that makes sense.

Almost forgot, yes I did see that WccfTech article a ways back, I would like to see more bits like it from more reputable sites like Toms. WccfTech is more of a Tech tabloids guilty pleasure....ie I don't trust very much I hear unless I have very good reason to...that said as you know it paints a different perfomance picture from the x86 CPUs. They don't get a huge boost but it is enough to make them more competitive than when they are hobbled. ST is great but at the end of the day, everyone runs more than one ST app at any given time. So the MT a single core should not be ignored when comparing cores which is my ultimate point.
 
Last edited:

kaalus

Distinguished
Apr 23, 2008
73
47
18,560
How can you even compare these chips - one is a laptop CPU with TDP of ~30W, the other is a desktop behemoth with TDP of 125W and actual use around 250W. Yet the little guy trounces the goliath. This is a complete wipeout by Apple. If M1 was allowed to use 250W, it would be much, much faster leaving all current desktop offerings in the dust. It's such a shame we cannot get a PC powered by M1, only a Mac.