M3 Chip Generation - Discussion Megathread

[name99]

The A17 draws 14W (!!!!) every time you open an app, browse the web or do other tasks that need burst of power. I am sorry but this is inexcusable. The Apple Silicon team, before losing most of their best chip designers after the release of the M1, would never release a chip like this. Apple's chip team is now undisciplined and is no longer special compared to other teams at competing teams in the chip industry. They are following the AMD and Intel playbook by drastically increasing wattage to gain small performance improvements and no IPC gains.

When there's smoke, there's fire. I don't see how anyone can deny at this point that Apple has lost most of their best chip designers.

Evidence for this claim? ie evidence by someone who understands who powermetrics works and is capable of creating a graph of the *duration* of this supposed power spike.
Are Apple iPhone batteries even CAPABLE of bursting the current required for a 14W draw -- that's about 4 AMPS! ?
For how long?

 

[name99]

While technically true that Geekerwan managed to get the A17 Pro to draw 14 Watt under load it was also under certain conditions.

1. They increased the thermal envelope by using external cooling
2. The chip used this increased thermal headroom to boost clock speeds

Under normal circumstances the thermal headroom of the design wouldn't allow it and no one else has been able to replicate that under ordinary conditions.

I suspect an additional IMPORTANT variable was that it was obtaining external power, eg through the USB connector.

This is important because Retskrad's entire argument seems to be "it is unacceptable for a PHONE chip to be using as much power as a laptop – even when it is plugged into the wall, and operating as a console".
This appears to be an argument from totem and taboo – if the type category of a device is "phone" then it is unacceptable for it to behave like the type category of "console" even when that makes engineering sense, because reasons.

He has given no reasons WHY this power draw is bad UNDER the conditions of external power and thermal headroom, he just wants to throw out random claims and see what sticks. Anyone else want to try?
Why exactly is it BAD for a phone to use laptop level power, and deliver console level performance IF the power is available and the device remains cool enough?
What EXACTLY is the problem with this? Why EXACTLY are we supposed to be outraged?

 

[MayaUser]

I think that user is looking to make argue between us...i think there are enough replies/comments because Retskrad IMO

 

[leman]

To put it in some perspective, in my experiments (which measured maximal power draw of a CPU) the 85-th percentile of multi-core power consumption for A17 Pro was 7.7 watts. This means that 85% of all measurements consumed 7.7 watts or less (and 50% consumed less than 6.6 watts). The peak power consumption I have measured was 10.3 watts, but only 0.5% of measurements were 10 watts or higher.

I don't really see how these values are inappropriate for a smartphone. From what I've seen Android phones draw more power than that. And didn't our friend claim that Qualcomm has surpassed Apple when it comes to multi-core efficiency?

P.S. The peak power consumption I measured for A13 was 7.6 watts. I thought Apple still had their crack team of engineers back then? How could they make such a bad chip that approaches desktop power consumption?

 

[dgdosen]

From outside looking in, Apple critics, like Me, and people who like to defend Apple, are both guessing. Who knows what’s really going on inside Apple. I’m just putting the pieces together. When Apple does a good job, I praise them. When they stumble and are in a slump, I keep the same energy and say they are doing a poor job. No one has yet argued a compelling reason why the A17 peaks like a laptop/desktop chip. Under no circumstance should a phone chip, with a much smaller battery and thermal envelope, do that. If Qualcomm or Samsung did that, people would heavily criticize them for it.

Such troll:

"Apple's chip team is now undisciplined and is no longer special compared to other teams at competing teams in the chip industry" -- "I guess"

I see a bright future in cable news as yours for the taking.

 

[APCX]

To put it in some perspective, in my experiments (which measured maximal power draw of a CPU) the 85-th percentile of multi-core power consumption for A17 Pro was 7.7 watts. This means that 85% of all measurements consumed 7.7 watts or less (and 50% consumed less than 6.6 watts). The peak power consumption I have measured was 10.3 watts, but only 0.5% of measurements were 10 watts or higher.

I don't really see how these values are inappropriate for a smartphone. From what I've seen Android phones draw more power than that. And didn't our friend claim that Qualcomm has surpassed Apple when it comes to multi-core efficiency?

P.S. The peak power consumption I measured for A13 was 7.6 watts. I thought Apple still had their crack team of engineers back then? How could they make such a bad chip that approaches desktop power consumption?

Mr Williams III was working undercover for Qualcomm

 

[Confused-User]

Two big issues at play there. First and foremost is the limited production of 3nm silicon in comparison to the 5nm silicon used in every current product NOT named iPhone 15 Pro/Pro Max. Apple is not going to cannibalize an already limited supply of 3nm parts in order to bump the specs for Vision Pro, especially when it's likely to sell considerably fewer units given the ~$3000 price point.

I guess you're in write-only mode? I and others have already debunked this ludicrous claim. There is probably no shortage of N3 supply already, and there certainly won't be by the time AVP ships. Further, as someone else pointed out, there's a good chance M3 will be on N3E, not N3B.

Beyond that... Would you care to explain your reasoning as to why, if they had to choose, Apple would forgo making a $3500 product that will define its future, in favor of an $800-1500 product that is subject to extremely elastic demand?

The second issue is the fact that it's not the M2 which handles all the input from the cameras, sensors, microphones, etc. That is the job of the R1 chip:

Focusing on M2 vs. M3 actually overlooks what is doing the bulk of the work inside Vision Pro, and I'd bet the farm that Apple isn't about to replace R1 just a few months before VP actually launches.

That's a good point, and undoubtedly true. It's also not that relevant. The M2/3 will be doing a great deal of work and is almost certainly a big part of the 12ms pipeline, even if the best magic is happening in the R1. Even if it's not, the more CPU, GPU, and power efficiency the AVP has, the more likely it is to be a hit. When you're trying to define the future you put your best foot forward.

Actually, does anyone know how much the M3 is part of the 12ms pipeline? I would expect it does all the final compositing, at least, to prevent added delay on whatever its GPU is doing, but I suppose it's possible it's the other way around. Maybe. I think that if the R1 is doing that, then it must at least have equal access to the RAM, and I don't think it does. But I don't know. A lot is unclear.

As another aside, why does everyone say 12ms? It must be 11.11ms. 12ms would only be good enough for 83Hz displays.

 

[clam zero]

Further, as someone else pointed out, there's a good chance M3 will be on N3E, not N3B.

That would be a pleasant surprise, but I'm not holding my breath. I feel like we would have heard something out of TSMC if all their investment in N3B had yielded just one chip for one company.

 

[k27]

14 Watt? This is probably the source: https://www.notebookcheck.com/Apple...-verbraucht-aber-bis-zu-14-Watt.752622.0.html

 

[name99]

14 Watt? This is probably the source: https://www.notebookcheck.com/Apple...-verbraucht-aber-bis-zu-14-Watt.752622.0.html

Like I said, idiots. Meaning who cares what they claim.

Consider eg the FP case
"Power" (more precisely total energy/total time) will go up by ~17% simply because of the 17% reduced duration. Meanwhile energy goes up by ~21%.
So we have essentially a flat energy-delay product: energy up by 21%, delay down by 17%, net product is up by ~4%. Who cares!
The energy-delay value is essentially the same! ie yes energy went up, but the performance also went up – and by enough that, by the standard metric of these things, the performance increase is worth the energy increase.
And if we go by energy-delay^2, (the other standard metric) we are now substantially ahead...

This is even apart from the issue (pounded on repeatedly) that everything about iPhone is tweaked to optimize interactivity. When you upgrade your phone after two or three years, the delight lies in the fact that it FEELS faster. In other words what the customer cares about is that interactions that used to take .25 seconds now take .15 seconds, and this is noticable everywhere. It makes perfect sense for Apple to optimize for this, even at the cost of a small amount of energy, because this is what customers WANT! I've upgraded my iPhone every two or three years since iPhone 1, and even the most recent upgrade from an A12 to an A15 (whatever the phone names were, I forget) this increased speed was obvious – and a major reason for upgrading.

Of course one also wants battery life. But for the way MOST people use their phones, battery life is not determined by P-cores running SPEC; it's determined by radios, screen, and E-cores; P-core energy usage is just not a large part of the daily energy budget (except maybe for gamers).

 

[PgR7]

Like I said, idiots. Meaning who cares what they claim.

Consider eg the FP case
"Power" (more precisely total energy/total time) will go up by ~17% simply because of the 17% reduced duration. Meanwhile energy goes up by ~21%.
So we have essentially a flat energy-delay product: energy up by 21%, delay down by 17%, net product is up by ~4%. Who cares!
The energy-delay value is essentially the same! ie yes energy went up, but the performance also went up – and by enough that, by the standard metric of these things, the performance increase is worth the energy increase.
And if we go by energy-delay^2, (the other standard metric) we are now substantially ahead...

This is even apart from the issue (pounded on repeatedly) that everything about iPhone is tweaked to optimize interactivity. When you upgrade your phone after two or three years, the delight lies in the fact that it FEELS faster. In other words what the customer cares about is that interactions that used to take .25 seconds now take .15 seconds, and this is noticable everywhere. It makes perfect sense for Apple to optimize for this, even at the cost of a small amount of energy, because this is what customers WANT! I've upgraded my iPhone every two or three years since iPhone 1, and even the most recent upgrade from an A12 to an A15 (whatever the phone names were, I forget) this increased speed was obvious – and a major reason for upgrading.

Of course one also wants battery life. But for the way MOST people use their phones, battery life is not determined by P-cores running SPEC; it's determined by radios, screen, and E-cores; P-core energy usage is just not a large part of the daily energy budget (except maybe for gamers).

Lmao what are those maths, is 21% more power consumption for the same work. They went more voltage to squeeze more frequency, the result is less efficient CPU when you reach those clocks.

 

[Boil]

So if we are seeing increased clock speeds & power draw when increased cooling is added, and if we assume the same for the M3 series of SoCs, I would imagine the Mac Pro (with the best cooling of all Apple silicon products) will see the most performant chips in the series...?

 

[PgR7]

So if we are seeing increased clock speeds & power draw when increased cooling is added, and if we assume the same for the M3 series of SoCs, I would imagine the Mac Pro (with the best cooling of all Apple silicon products) will see the most performant chips in the series...?

Its obvious that is going to be faster than the M2, and we may see a difference in performance between Mac Studio and Mac Pro because of the added cooling

 

[Confused-User]

Lmao what are those maths, is 21% more power consumption for the same work. They went more voltage to squeeze more frequency, the result is less efficient CPU when you reach those clocks.

Um, wow. If you don't understand the difference between, say, watts and watt-hours, you really aren't qualified to be having this conversation. What you're doing makes as much sense as saying "The iPhone 15 Pro weighs 90% of the 14 Pro, so I need to multiply the numbers for the 15Pro by 90% to compare to the 14". No, you don't, that's stupid, because the weight of the phone isn't relevant to the discussion. What you're doing is just as bad.

A Watt is a rate of energy transfer. If electricity were water (it's not, and you can get in trouble thinking it's similar, but this is a useful analogy), it would measure liters or gallons per second. Does that tell you how much water was used? NO! You need to know for how many seconds the water was flowing!

So in the chart posted by @k27, the water was flowing faster for the A17... but the tap was open for less time! So you have to do the sort of math that @name99 did to actually measure how much water (energy) was used.

Also, you have ZERO data about voltages. Neither do I. I'm not aware of any published info about the voltages being used by the A17, but I'm extremely interested.

Now, all that said... see my next post.

 

[Confused-User]

Consider eg the FP case
"Power" (more precisely total energy/total time) will go up by ~17% simply because of the 17% reduced duration. Meanwhile energy goes up by ~21%.

Am I missing something here? Assuming that their published data is correct, it's not 21%, it's 38%. (I think you subtracted 17 twice?) That does not result in a favorable energy-delay product.

Of course, I'm not trusting their data without backup. And if their data is correct, it's likely still a good tradeoff given the bursty nature of most smartphone use.

 

[Confused-User]

That would be a pleasant surprise, but I'm not holding my breath. I feel like we would have heard something out of TSMC if all their investment in N3B had yielded just one chip for one company.

Lol, have you been living in a cave? Some days it feels like that's all we've been hearing in the last month.

In fact, there are no known customers for N3B besides Apple, other than one or two crypto chips that are tiny and being made in low numbers.

 

[PgR7]

Um, wow. If you don't understand the difference between, say, watts and watt-hours, you really aren't qualified to be having this conversation. What you're doing makes as much sense as saying "The iPhone 15 Pro weighs 90% of the 14 Pro, so I need to multiply the numbers for the 15Pro by 90% to compare to the 14". No, you don't, that's stupid, because the weight of the phone isn't relevant to the discussion. What you're doing is just as bad.

A Watt is a rate of energy transfer. If electricity were water (it's not, and you can get in trouble thinking it's similar, but this is a useful analogy), it would measure liters or gallons per second. Does that tell you how much water was used? NO! You need to know for how many seconds the water was flowing!

So in the chart posted by @k27, the water was flowing faster for the A17... but the tap was open for less time! So you have to do the sort of math that @name99 did to actually measure how much water (energy) was used.

Also, you have ZERO data about voltages. Neither do I. I'm not aware of any published info about the voltages being used by the A17, but I'm extremely interested.

Now, all that said... see my next post.

It looks like you are who doesnt know, if the A17 Pro needs 1 hour to do some work, It will need 1*4.4=4.4 wh, the A16 will take longer to complete so It Will need 1.17*3.19=3.7 wh. So the A17 Pro is 17-18% less efficient. I love when you dare to go maths because its so easy to probe you are wrong.

 

[Confused-User]

It looks like you are who doesnt know, if the A17 Pro needs 1 hour to do some work, It will need 1*4.4=4.4 wh, the A16 will take longer to complete so It Will need 1.17*3.19=3.7 wh. So the A17 Pro is 17-18% less efficient. I love when you dare to go maths because its so easy to probe you are wrong.

That would be roughly-sorta correct, if those numbers actually represented a real-world behavior that lasted an hour. (Obviously they don't.) Which is why I said as much in the post immediately after the one you're quoting ("that does not result in a favorable energy-delay product").

But when you say "lmao what are those maths?" in response to @name99's post, then that just sounds like "head hurtz, math too hard, chip slow, fire bad, tree pretty."

 

[quarkysg]

It looks like you are who doesnt know, if the A17 Pro needs 1 hour to do some work, It will need 1*4.4=4.4 wh, the A16 will take longer to complete so It Will need 1.17*3.19=3.7 wh. So the A17 Pro is 17-18% less efficient. I love when you dare to go maths because its so easy to probe you are wrong.

But is the amount of work done in that 1 hours the same for the A17 Pro and A16? Do you know that?

 

[iPadified]

Work done for an hour... People, please stop treating the A17 Pro as a desktop chip to drive home academic points. A17 Pro performance (or lack of) should be discussed in light of the typical work load of an average iPhone user. That being said, the N3 node does not yet seem to be the salvation many hoped for prior to the A17 Pro release.

 

[PgR7]

That would be roughly-sorta correct, if those numbers actually represented a real-world behavior that lasted an hour. (Obviously they don't.) Which is why I said as much in the post immediately after the one you're quoting ("that does not result in a favorable energy-delay product").

But when you say "lmao what are those maths?" in response to @name99's post, then that just sounds like "head hurtz, math too hard, chip slow, fire bad, tree pretty."

The maths are correct if It last 1 hour or 1 second, the A17 core running at 3.78 GHz will be less efficient than the the A16 core running at 3.45 GHz. Well he is wrong, I like the juggling some of you do to defend Apple

 

[PgR7]

But is the amount of work done in that 1 hours the same for the A17 Pro and A16? Do you know that?

They do the same work, the A16 takes longer but consumes less in total

 

[clam zero]

Lol, have you been living in a cave? Some days it feels like that's all we've been hearing in the last month.

In fact, there are no known customers for N3B besides Apple, other than one or two crypto chips that are tiny and being made in low numbers.

That's exactly what I was referring to. We know it's just Apple buying these chips so if they were going solely into A17s I think there's a good chance we would have heard that by now, especially if there was a change of plans from N3B to N3E (which would have been a while ago). Combine that with the fact that M1 and M2 are both built on the same process as the preceding A-series chip and that volume production of N3E didn't begin until a few months before Apple's upcoming event (originally-planned for this month) and it seems safe to assume that M3 will be N3B-based.

 

[quarkysg]

They do the same work, the A16 takes longer but consumes less in total

Ahh ... then it is not the same amount of work done in the 1 hour period. Are you sure you understand the concept?

Are you then not also juggling to diss Apple using the A17 Pro as a weapon?

 

[Confused-User]

They do the same work, the A16 takes longer but consumes less in total

Let's assume that you're correct about this (and I think you are in many cases, though whole-system power is more complex than A17 power, and it's not always going to be that way).

You still seem to be missing completely the point I was making (and that @name99 was making, too). This isn't a bad thing, it's a good thing. Making the system more responsive, "snappier", is generally a win. Pushing it too far, of course, you get overheated chips and poor battery life. So what makes this the right choice?

That answer is subjective, of course. You can argue that for your particular use case and preferences, Apple's choice is suboptimal. You can't however say that they built a bad product.

I suspect that what actually happened was that Apple looked at the whole system and said "we want battery life to be as good as last year, and we'd like better performance, what can do within those constraints?". And the 15Pro is what we got. I am quite happy with their choice, and so are lots of other people. You wanted more battery and are disappointed? Fair, but not a reason to say their product has a problem. They just built something not optimal for you.