Intel Alder Lake vs. Apple M1

[senttoschool]

I think it’s worth emphasizing leman’s point here. Just because two CPUs are capped to not take more than X watts doesn’t mean their consumption are the same in a given scenario.

So to apply that to the chart, a number of the scenarios listed are single threaded scenarios. Excel being one of them. So is that because of higher efficiency, or higher power consumption (i.e. higher boosting that stays under the 45W limit on the one core under load)? 25-30W is 20-50% more power, which would explain the ST benchmarks in the graph. Without power measurements to go along with the benchmarks, you can’t say anything with certainty one way or another.

AMD also boosts on mobile. Often well past its stated limits. For example, their 15w chip will often boost to 30w.

What ADL shows definitively, is that it has a massive lead on Zen3 on mobile in terms of performance, given the same class of laptop. This highly suggests that ADL is more efficient than Zen3 on mobile.

We're not going to get much better than that because every laptop is different in cooling, performance, max wattage. And reviewers don't generally test every benchmark for perf/watt.

Just ask yourself, with these results, would you buy a Zen3 or ADL laptop assuming all other things are equal?

 

[crazy dave]

@senttoschool We're not hating on Intel or making excuses for AMD. That's not what's happening. It may be what you're reading, but it isn't what we're writing. Personally, I'm just going to leave it here as I said I would since I'm clearly not able to get the point across and I don't think there's anything really novel left to say in this discussion.

 

[crazy dave]

That configuration had one of the highest-end Intel consumer CPUs of its generation with the lowest-end discrete AMD GPU of its generation. I would not call it very balanced.

Sure and the M1 Pro is a mere 4.6 Tflop GPU paired with a 10-core CPU which is far faster than that highest end Intel consumer CPU you're quoting. It's almost the same balance, slightly more GPU horsepower (it should be noted that the equivalent AMD GPU of this generation is slightly less powerful than the 14 core M1 Pro but it's the same ballpark). Apple basically took that balance and made it its base model.

And the 128 GB is definitely a consumer level build, because it's in a consumer device powered by a consumer CPU.

You cannot say that if a computer cost more than $5000 then the CPU isn't a consumer build then turn around and say spending $12,000 on RAM is a consumer build because the CPU is an i-series. Your definitions are all over the place and incredibly inconsistent. Further it's important to note that the i-series vs xeon divide is not likely to translate to Apple silicon. So I'm just not really sure where you are going with your definitions. What is a consumer level chip to you? What is not a consumer level chip and how does that apply to the as of yet unreleased higher core count M1 chips?

I expect something like $6k for a 10-core Mac Pro, $10k for a 20-core model, and $15k for a 40-core model, assuming reasonable RAM/SSD options and including a decent monitor and other peripherals. The entry-level model will probably have an M1 Max or equivalent, just like the current entry-level model is comparable to a much cheaper iMac.

What are you including in this $6K machine? You can get a 10-core MacBook Pro for $2200. It's extremely likely that you'll be able to get a 10-core Mac Mini for less. Is this an M1 Pro/Max chip in this $600 machine? Again, are those by your definition consumer level chips or not? Would the rumored 12-core variant be a consumer level chip? I mean we'll find out for sure by the end of the year, but I'm not sure how you are pricing things.

Also it's important to note that a 20-core M1 could be ~50% faster than an i9 Alder Lake and a 40-core one would be about double.

It's more about planning than constraints. Apple has reserved a certain amount of capacity from TSMC and paid for it, and they expect to sell everything TSMC is able to produce. They wouldn't like it if they end up with large amounts of unused capacity or huge piles of devices nobody wants to buy. Because the M1 Max is 4x bigger than the M1, Apple has to choose whether it would rather sell four M1 devices or one M1 Max device. They choose the ratio between the devices and their prices in a way that they expect to maximize the profits.

The reality is obviously more complex, but that's the basic idea. Big businesses are centrally planned economies that would make the USSR jealous. Spare capacity was seen as a waste of money that was optimized away, which is why the supply chain crisis hit so hard.

Yeah ... that's my argument to you which nullifies your argument that they can choose to sell 4 M1s for every M1 Max and therefore any SOC with more than that will be priced to the moon. You're repeating my argument that Apple plans its production to meet the demand that it thinks it is going to get all the way up the product stack. Yes they could choose to *make* 4 M1 dies for every M1 Max die, but that's not what they would be able to *sell*. Apple has a lot of data about what its customers want and how much of each product tier to make and what to price things at in order to move them.

In this context, a high core count is a relative concept. It's about adding more cores than people are used to, in an attempt to gain competitive advantage. 10-core consumer chips have been common for a few years, and both Intel and AMD have been toying with higher core counts for a while.

Yes and Apple has a 10 core consumer level chip, I think even by your definition. Why would a higher core count Apple made chip that is more than 10 cores not be consumer grade? Again, what is your definition of consumer grade? I am so very confused and we seem to be miles away from the initial statement that Apple could and would ship SOCs with CPUs with more than 10 cores which you seem to think will be prohibitively expensive. Again, it should be noted that @EntropyQ3 confirmed that industry watchers have concluded that Apple is paying significantly less now that they are manufacturing their own chips, not more as you seem to be insistent that they must be.

 

[leman]

But the ADL scored higher....

And that's the 12900K, which is pushed far above its optimal range. We know all along that mobile ADL is where it will shine more because of it's "big.Medium" (according to you) architecture and the fact that Intel won't push it as hard where efficiency matters more.

Sure, it scored 20% higher in FP tests at 50% (estimated) higher power consumption… that’s hardly more efficient.

So on desktop, where efficiency matters less, and where people generally buy Intel K processors for gaming, ADL is more efficient than Zen 3 for its main purpose.



And on laptops, where efficiency matters the most, ADL is more efficient than Zen3.

What your slides show is that ADL is faster, not really more efficient. Interestingly enough, the situation is a bit fuzzy on desktop because Zen3 has very high idle power draw due its multi-chip architecture (the IO die/fabric is a bit if a power hog). But they don’t have this problem on mobile.

As we discussed before, in multicore stuff ADL does end up more efficient because, as already stated, it has more cores and can clock them lower. That doesn’t mean that Sunny Cove is more efficient than Zen3. It simply means that Intel packaged their stuff with throughout in mind.

 

[leman]

When are we going to stop making excuses for AMD and actually give ADL its due as the king of x86 consumer? Look, I get it. Apple Silicon has conditioned everyone here to hate on Intel. And the fact that Intel was stuck on 14nm for so long made it worse. But ADL is clearly superior to Zen3 in just about everywhere. Let's move on.

Oh, I’m not hating on Intel. And I definitely don’t see Zen3 as superior. Frankly, I have always been very puzzled by the amount of fanfare AMD has been getting with Zen, their only achievement was getting parity with Intel in per-core performance. And I certainly don’t hate ADL. Intel did a great job turning the situation around and coming back after a long period of stagnation. Their marketing strategy is sound and they do excellent work with what they have. The thing is, I just don’t find any of these products interesting or compelling. I am a bit if a tech nerd, and that is not technology that makes my heart beat faster, because it’s crude. Of course you will be on top of CB charts if you pack your machine full of slower low-power cores. And of course you’ll top single-core charts if you run your cores at 30W+. Where is the sophistication, the “smart” solution? This is just brute forcing your way through, putting on a show. That is not engineering I personally am interested in and that’s why I am happy that Apple chose their own way, with really interesting products.

 

[senttoschool]

Sure, it scored 20% higher in FP tests at 50% (estimated) higher power consumption… that’s hardly more efficient.

Anandtech reviews for Zen3 and Golden Cove. ADL uses 25-30W in single-core SPEC. Zen3 maxes out at 20W per core.

The Intel 12th Gen Core i9-12900K Review: Hybrid Performance Brings Hybrid Complexity

www.anandtech.com

AMD Zen 3 Ryzen Deep Dive Review: 5950X, 5900X, 5800X and 5600X Tested

www.anandtech.com

@senttoschool We're not hating on Intel or making excuses for AMD. That's not what's happening. It may be what you're reading, but it isn't what we're writing. Personally, I'm just going to leave it here as I said I would since I'm clearly not able to get the point across and I don't think there's anything really novel left to say in this discussion.

It seems like you've all been backed into a corner by saying Zen3 is more efficient than ADL and are now just trying to find anything to back it up despite all evidence pointing otherwise.

@leman Anandtech's reviews shows that ADL's ST is much more efficient. Not less. You need to look at package power.

First, Anandtech's package power for ADL ST is 25.3W and 29.2W:

In SPEC, in terms of package power, the P-cores averaged 25.3W in the integer suite and 29.2W in the FP suite, in contrast to respectively 10.7W and 11.5W for the E-cores, both under single-threaded scenarios. Idle package power ran in at 1.9W.

Next, AMD's 5950x in total package power is actually 49w. You made the mistake of only looking at 1 core power, which you correctly stated was at around 20.6w.

This actually makes ADL both faster and more efficient.

In addition, we know that ADL's 12900k, which is what Anandtech reviewed, is pushed well beyond its efficiency sweet spot. This is why I've said all along that laptop ADL should lead Zen3 even more, which we're starting to see.

 

[leman]

Next, AMD's 5950x in total package power is actually 49w. You made the mistake of only looking at 1 core power, which you correctly stated was at around 20.6w.

This is true. I did mention it in my previous post - Zen3 desktop has a significant package power overhead due to its multi-chip organization. I would really like to compare core efficiency (as this is what we are talking about), not package efficiency. I mean, it is a significant drawback for AMD on desktop, but not on mobile, as they are using a monolithic core there.

You are correct though, the numbers Anandtech published for ADL are package power. I thought it was core power. It’s a shame that we don’t have published figures for that… one would think that this is basic information customers should know.

 

[Fomalhaut]

Next, AMD's 5950x in total package power is actually 49w. You made the mistake of only looking at 1 core power, which you correctly stated was at around 20.6w.

Doesn't the table show the package power going up to 136W? Where do you get 49W from?

 

[senttoschool]

Doesn't the table show the package power going up to 136W? Where do you get 49W from?

Just use the package power for column "1". We're measuring ST package power.

 

[crazy dave]

It seems like you've all been backed into a corner by saying Zen3 is more efficient than ADL and are now just trying to find anything to back it up despite all evidence pointing otherwise.

@leman Anandtech's reviews shows that ADL's ST is much more efficient. Not less. You need to look at package power.

First, Anandtech's package power for ADL ST is 25.3W and 29.2W:



Next, AMD's 5950x in total package power is actually 49w. You made the mistake of only looking at 1 core power, which you correctly stated was at around 20.6w.

This actually makes ADL both faster and more efficient.

In addition, we know that ADL's 12900k, which is what Anandtech reviewed, is pushed well beyond its efficiency sweet spot. This is why I've said all along that laptop ADL should lead Zen3 even more, which we're starting to see.

@leman @Fomalhaut
Quick point: Actually here's the same graph for Alder Lake:

The Intel 12th Gen Core i9-12900K Review: Hybrid Performance Brings Hybrid Complexity

www.anandtech.com

Here you can see package power in ADL jump to 78W with a single thread active. But under Linux it claims 30W per core. The per core measurements were broken in their tests on Windows.

 

[senttoschool]

Frankly, I have always been very puzzled by the amount of fanfare AMD has been getting with Zen, their only achievement was getting parity with Intel in per-core performance.

It's because Intel was milking its dual and quad cores for a decade until AMD changed the game with 8-core Zen1 at an affordable price. Then Intel severely messed up with the 14nm delay so they couldn't advance their per-core performance, allowing AMD to catch up with Zen2 and Zen3.

And AMD is the only competitor to Nvidia a the moment and internet people hates Nvidia.

This garnered AMD a lot of fanboys and people think that AMD is the "savior". I'm sure there are a ton of AMD-fanboys on this forum too but I'm not going to call them out.

 

[leman]

Here you can see package power in ADL jump to 78W with a single thread active. But under Linux it claims 30W per core. The per core measurements were broken in their tests on Windows.

That’s a different test… they are talking just POVray vs the average power running the entire spec suite. The wording is a bit unfortunate.

In an ideal world, we would get tables with various stats for all the tests. Unfortunatel, not even the best reviewers bother…

 

[senttoschool]

@leman @Fomalhaut
Quick point: Actually here's the same graph for Alder Lake:

The Intel 12th Gen Core i9-12900K Review: Hybrid Performance Brings Hybrid Complexity

www.anandtech.com

View attachment 1955433

Here you can see package power in ADL jump to 78W with a single thread active. But under Linux it claims 30W per core. The per core measurements were broken in their tests on Windows.

First, the 25-30w claim was for SPEC. This is POV-Ray.

Second, Anandtech didn't do a good job explaining the differences. I believe POV-Ray has AVX-512, which Anandtech enabled on ADL for review. AVX-512 increases both power consumption and performance greatly for applications that make use of it.

 

[crazy dave]

That’s a different test… they are talking just POVray vs the average power running the entire spec suite. The wording is a bit unfortunate.

In an ideal world, we would get tables with various stats for all the tests. Unfortunatel, not even the best reviewers bother…

No its more complicated than that even, it's how they each report package power. And the AMD results were on POV ray too.

 

[crazy dave]

First, the 25-30w claim was for SPEC. This is POV-Ray.

Second, Anandtech didn't do a good job explaining the differences. I believe POV-Ray has AVX-512, which Anandtech enabled on ADL for review. AVX-512 increases both power consumption and performance greatly for applications that make use of it.

POV-ray is the SPEC subtest 511 - however it is possible that this was built differently. They didn't make that explicit however.

 

[senttoschool]

POV-ray is the SPEC subtest

And Anandtech's SPEC package power uses average package power.

Regardless, my point about AVX-512 still stands.

Intel Rocket Lake (14nm) Review: Core i9-11900K, Core i7-11700K, and Core i5-11600K

www.anandtech.com

 

[senttoschool]

@crazy dave I'm done with this. It's taken too much time and it isn't going anymore. It doesn't seem like all the data points are getting to you. The goalpost continues to move. All evidence points to ADL being both faster and more efficient than Zen3 at each market's primary use case: gaming and laptops.

I think it will be even more obvious when the thin and light laptops come out and suddenly ADL is both significantly faster and offers more battery life than Zen3 without a doubt.

Regardless, I'm very happy with my M1 Pro 16" and neither ADL nor Zen3/4 excite me like @leman said. All I've been trying to do is bust some anti-Intel and pro-AMD myths around here.

 

[crazy dave]

And Anandtech's SPEC package power uses average package power.

Regardless, my point about AVX-512 still stands.

Intel Rocket Lake (14nm) Review: Core i9-11900K, Core i7-11700K, and Core i5-11600K

www.anandtech.com

Yes but they compiled SPEC with AVX2, the numbers for AVX 512 are from the previous page. They couldn't use AVX-512 because they did mixed testing with POV-Ray on P and P+E cores. POV-Ray is an intense test however, out of all the SPEC tests. I'll grant you that. However, per-core loading in the AMD graph is also POV-ray so it is consistent:

AMD Zen 3 Ryzen Deep Dive Review: 5950X, 5900X, 5800X and 5600X Tested

www.anandtech.com

A number of these tests have been requested by our readers, and we’ve split our tests into a few more categories than normal as our readers have been requesting specific focal tests for their workloads. A recent run on a Core i5-10600K, just for the CPU tests alone, took around 20 hours to complete.

Power​

• Peak Power (y-Cruncher using latest AVX)
• Per-Core Loading Power using POV-Ray

Zen 3 uses less per-core power and package power than ADL Golden Cove for the same test. That doesn't change which is a better buy for a consumer right now (its ADL I would say for most of the product stack).

 

[crazy dave]

@crazy dave I'm done with this. It's taken too much time and it isn't going anymore. It doesn't seem like all the data points are getting to you. The goalpost continues to move. All evidence points to ADL being both faster and more efficient than Zen3 at each market's primary use case: gaming and laptops.

I think it will be even more obvious when the thin and light laptops come out and suddenly ADL is both significantly faster and offers more battery life than Zen3 without a doubt.

Regardless, I'm very happy with my M1 Pro 16" and neither ADL nor Zen3/4 excite me like @leman said. All I've been trying to do is bust some anti-Intel and pro-AMD myths around here.

Funny I feel the same way about you. Have a good night/day.

 

[leman]

I think it will be even more obvious when the thin and light laptops come out and suddenly ADL is both significantly faster and offers more battery life than Zen3 without a doubt.

In the end, one part of the problem is that we simply don't have enough data. From reading all these tests my impression is that Zen3 still has some efficiency edge over ADL, but looking at the Anadntech stuff again I might very well be mistaken.

I hope that we will see more tests (inclusding detailed power reports) that will be able to clear this up.

 

[crazy dave]

In the end, one part of the problem is that we simply don't have enough data. From reading all these tests my impression is that Zen3 still has some efficiency edge over ADL, but looking at the Anadntech stuff again I might very well be mistaken.

I hope that we will see more tests (inclusding detailed power reports) that will be able to clear this up.

The AMD results from the Anandtech graph @senttoschool linked to are also from POV-Ray - not an average SPEC score and it doesn't seem to be AVX-512 related.

Intel Alder Lake vs. Apple M1

Yes but they compiled SPEC with AVX2, the numbers for AVX 512 are from the previous page. They couldn't use AVX-512 because they did mixed testing with POV-Ray on P and P+E cores. POV-Ray is an intense test however, out of all the SPEC tests. I'll grant you that. However, per-core loading in the...

forums.macrumors.com

So yes, I would say that Zen 3 cores still have an efficiency edge.

 

[ian87w]

The proof will be in the pudding. M1 Macbooks have been out for more than a year, while Alder lake is still coming soon. We have seen a fanless M1 Macbook Air being tested through and through with flying colors. Until Alder Lake can match that, if I'm in the market of a $1000+ laptop, my wallet will go to Apple. Anything less, my wallet will go to Ryzen based laptops as the OEMs will mostly offer more bang for the buck (eg. more RAM and storage) for a Ryzen model than an intel model, with better performance compared to an equivalent 10th/11th gen i3/i5.

If intel can do better, let them prove it on the market. Bring on the laptops. I'm still seeing OEMs selling 10th gen intel, so to me, Alder lake is still vaporware. And no point in only showing the i7/i9 as they will be priced ridiculously expensive.

 

[Fomalhaut]

The proof will be in the pudding. M1 Macbooks have been out for more than a year, while Alder lake is still coming soon. We have seen a fanless M1 Macbook Air being tested through and through with flying colors. Until Alder Lake can match that, if I'm in the market of a $1000+ laptop, my wallet will go to Apple. Anything less, my wallet will go to Ryzen based laptops as the OEMs will mostly offer more bang for the buck (eg. more RAM and storage) for a Ryzen model than an intel model, with better performance compared to an equivalent 10th/11th gen i3/i5.

If intel can do better, let them prove it on the market. Bring on the laptops. I'm still seeing OEMs selling 10th gen intel, so to me, Alder lake is still vaporware. And no point in only showing the i7/i9 as they will be priced ridiculously expensive.

Absolutely; while it is interesting to hypothesise what eventual performance will be, at the end of a day we buy a computer, not a CPU. The overall performance of the machine is what will determine its value.

For me it's academic. I've just bought an M1 Max, and won't be buying an Intel laptop in the near future unless I have some pressing need to run Windows or Linux on a powerful laptop (I already have a couple of desktop machines).

Competition is good for innovation and only wish Intel and AMD well in their endeavours. I find the fan-boyism of blindly supporting a vendor with cherry-picked results to be rather pointless. Buy whatever is available and best suits your needs. It's only a machine, not a religion.

 

[Xiao_Xi]

Tom's Hardware considers that "In judging AMD vs Intel CPU performance per watt, [...] [t]he latest Ryzen processors consume less power on a performance-vs-power basis, which equates to less heat generation."

Intel vs AMD: Which CPUs Are Better in 2024?

We put Intel vs AMD in a battle of processor prowess.

www.tomshardware.com

 

[crazy dave]

In an ideal world, we would get tables with various stats for all the tests. Unfortunatel, not even the best reviewers bother…

The problem is a lack of time. Most reviewers get very little time before the deadline to run their tests even once never mind all over again with measuring power and then troubleshooting results and then writing it all up. Review units have to be sent back afterwards and even if that’s a decent amount of time the financial incentives of page views means there is very little benefit to going back. For product reviews your first impression is often your only one.

I recently read a rather morose article on a hardware blog (a big one I think but can’t remember which) that basically predicted the death of the traditional review model employed by sites like Anandtech. It used to be that those kinds of sites and magazines were a crucial part of a product launch to move units. But now companies are restricting time on reviews further and further with the embargo now dropping only on or after launch. There are simply so many other easier ways to get your product out there through “influencers” and social media and there’s less risk of you, you know, getting those pesky bad reviews or even nuanced reviews that say your product might not quite be as good as you say it is. It was a depressing piece.