Power curve and efficiency of Apple N3

[PgR7]

I did another test, in Genshin Impact, same settings, same city, not exactly the same character as in the review but I dont think it will matter much. From this screenshot.

15 Pro Max 59.1 fps 4.13W 14.3 fps/W
Snapdragon 870 Android 13 undervolt GPU 45.7 fps 3.6W 12,69 fps/W

Yeah I know some people will say is not fair, that you should have limit the fps to 40, but in reality in lots of games the more performance phone will just perform faster. Its actually impressive that giving more performance it still has better efficiency numbers, and its much better than the Gen 2. And this is why the A17 Pro has better efficiency but will have lower battery life than my Android.

 

[headlessmike]

Do you mean Pentium 4's Willamette and Pentium III's Coppermine?

Intel Pentium 4 1.4GHz & 1.5GHz

www.anandtech.com

Although both were made at 180 nm, Willamette looks like a promising candidate.

It's funny how clueless Intel was at the time.

The future of Intel's manufacturing processes

www.anandtech.com

Yeah. I forgot that the early P4's used the same node size as the late PIII's. But the situation is true even for the later P4's. Intel really bet on the wrong horse there.

That's the theory. In practice we've seen a steady increase in power consumption of CPUs over the yers as the industry pursues higher performance.

When designing a device you usually have a power budget. Increased efficiency elsewhere leaves more on the table for the SoC to use. The gains from shrinking node sizes aren't what they used to be either, so the manufacturers need to keep up the expected generational increases performance anyway they can.

 

[Retskrad]

AMD keeps killing it with new microarchitecture and IPC gains. Meanwhile… Apple has had zero IPC gain the last 5 years.

Alleged AMD Zen 5 Specs Leak: Twice the Cores, 15% Increased IPC Over Ryzen 7000

Zen 5 will be AMD's first major architectural overhaul since Zen 2.

www.tomshardware.com

 

[leman]

AMD keeps killing it with new microarchitecture and IPC gains. Meanwhile… Apple has had zero IPC gain the last 5 years.

Kind of difficult to increase IPC further if you are ahead of the industry by 50-60%...

 

[Xiao_Xi]

AMD keeps killing it with new microarchitecture and IPC gains.

I find it interesting that AMD believes that it will increase IPC less and less. So, the higher the IPC of an SoC, the harder it will be to increase it.

Kind of difficult to increase IPC further if you are ahead of the industry by 50-60%...

It seems reasonable that Apple would have some trouble raising IPC.

 

[APCX]

AMD keeps killing it with new microarchitecture and IPC gains. Meanwhile… Apple has had zero IPC gain the last 5 years.

Alleged AMD Zen 5 Specs Leak: Twice the Cores, 15% Increased IPC Over Ryzen 7000

Zen 5 will be AMD's first major architectural overhaul since Zen 2.

www.tomshardware.com

Yes, this unsubstantiated leak is proof Apple is finished. Well done.

 

[PgR7]

I find it interesting that AMD believes that it will increase IPC less and less. So, the higher the IPC of an SoC, the harder it will be to increase it.

It seems reasonable that Apple would have some trouble raising IPC.

They have a very strong single core, so there is no problem in that, and efficiency is good too

 

[Confused-User]

I did another test, in Genshin Impact, same settings, same city, not exactly the same character as in the review but I dont think it will matter much. From this screenshot.

View attachment 2285222

15 Pro Max 59.1 fps 4.13W 14.3 fps/W
Snapdragon 870 Android 13 undervolt GPU 45.7 fps 3.6W 12,69 fps/W

Yeah I know some people will say is not fair, that you should have limit the fps to 40, but in reality in lots of games the more performance phone will just perform faster. Its actually impressive that giving more performance it still has better efficiency numbers, and its much better than the Gen 2. And this is why the A17 Pro has better efficiency but will have lower battery life than my Android.

That really depends on the applications, doesn't it? I mean, if you could limit fps in iPhone Genshin to "10% better than snapdragon" (and BTW that's just frame rates - the iPhone may be getting substantially better quality, too) you'd have better fps and less energy consumption. It's hardly Apple's fault if certain apps use all the power they can get their hands on. Would you prefer that Apple make it impossible for apps to access that power?

Actually, some people would, and perhaps setting low power mode will do that. @leman, if you have a chance do you think you could run your test in low power mode? If would be interesting to see what if anything that does to max clock. As I wrote earlier, there may be a case for Apple to expose some controls for this.

Beyond that, the SoC isn't the system. I don't think we know yet just how the whole system behaves, compared to the iP14P (much less various androids). There are plenty of measurements out there that say the 15 has slightly better battery. And Apple has just announced that they're shipping a fix for overheating; we won't know until after then what impact if any that will have on battery life.

Lastly... this thread is about the SoC, not the whole system. Your point isn't crazy but it's somewhat tangential.

 

[Confused-User]

The thing is, thats not the only use case in a phone, you load an app, the A17 will boost to 3.78 throwing efficiency away, the A15 boosts to 3.2, it loads the app 2 tenths later and if you track the power consumption the A17 will have consumed more energy for the same load (load an app). If you use a lot of burst loads like app loading, going from app to app, load a webpage, enter in an Instagram or Twitter profile, your A17 will consume more power than the A15 or A16.

You don't know this to be true. Race to idle is a real thing. As I just mentioned in another post, we don't yet have enough data, and Apple's announced fixes, which may or may not have a meaningful effect on performance, may change this story.

It's not impossible that you're right. But drawing conclusions is drastically premature.

Thats why we are reading people with heat and battery problems in the 15 Pro and there are reviews where the 15 PM losses against 14 PM in battery life.

But there are also reviews where the 15 wins. Not Enough Data!

Im sure they increased the voltage a lot to get that 3.78 Ghz frequency.

I would not be confident of that. That's a ~9% max clock boost - not far off what you'd expect iso-power going from N4 to N3. Perhaps the voltage is a little higher... and perhaps it's not. And perhaps leman and I are right: this core is designed to hit the sweet spot for desktops. In that case it's possible it does require somewhat higher voltages. We just don't know yet.

 

[Confused-User]

Pentium 4's were notorious for this. They were less efficient than Pentium III's at a given clock speed, but that was more than just a node shrink.

The Netburst architecture they used was designed to give a high performance at very high clock speeds (5GHz or so). But they were never able to reach those speeds for technical reasons and the performance suffered and it was known for its high power draw. The Core series of CPUs that followed were based on the Pentium III as a consequence.

Sort of. They were based on the Pentium M. The Pentium III was the ancestor of the M, so you're not wrong, but there were generations in between.

 

[PgR7]

if you could limit fps in iPhone Genshin to "10% better than snapdragon"

What you wanna mean by that, if I limit the fps of the Android phone to 40 fps it would be more efficient too, and no, there is some quote of Anandtech running around and it was referring to some games not benchmarks, but this time as you can see in the screenshot, he dialed down some settings in the 15 PM to get the same quality as in the Android version. Lmao and what you think how different the screen power consumption and Wifi modem will be of same modern phones. Its better to not contribute **** to the thread and just guess based on nothing, just a graph that not always translates to real life, right?

You don't know this to be true. Race to idle is a real thing. As I just mentioned in another post, we don't yet have enough data, and Apple's announced fixes, which may or may not have a meaningful effect on performance, may change this story.

It's not impossible that you're right. But drawing conclusions is drastically premature.

But there are also reviews where the 15 wins. Not Enough Data!

I would not be confident of that. That's a ~9% max clock boost - not far off what you'd expect iso-power going from N4 to N3. Perhaps the voltage is a little higher... and perhaps it's not. And perhaps leman and I are right: this core is designed to hit the sweet spot for desktops. In that case it's possible it does require somewhat higher voltages. We just don't know yet.

Yeah sure, Apple will tell you everything about it, if the A17 Pro is more efficient in every scenario then the 15 PM wouldnt lose in the battery life test of every review. I know how people like you and others in this thread are, I can do the most scientific proven test and you still would say its invalid because of some made up excuse.

 

[Confused-User]

AMD keeps killing it with new microarchitecture and IPC gains. Meanwhile… Apple has had zero IPC gain the last 5 years.

Alleged AMD Zen 5 Specs Leak: Twice the Cores, 15% Increased IPC Over Ryzen 7000

Zen 5 will be AMD's first major architectural overhaul since Zen 2.

www.tomshardware.com

AMD is fantastic, they've been firing on all cylinders for years.

Your claim about Apple is complete BS. But worse, it's a total red herring. Among other things, in the last 5 years Apple has added AMX units, massively boosted their NPU, and morphed their E cores into something approaching mid-cores on Android phones, just embarrassing all other ARM E cores. All of these are huge de-facto IPC gains.

But let's say you were referring solely to the P cores. IPC iso-clock isn't that different from five years ago? The A12 P has a GB6 SC score of maybe 1325, at 2490MHz: .532GB6/MHz. The A17 P scores maybe 2940, at 3778MHz: .778GB6/MHz. .778/.532 = 46% improvement. Yes, this is very loosey-goosey, but it's a useful indicator.

Even if there weren't any IPC gains, as I said, your argument is a red herring. Apple's been able to boost clocks by ~52%. Yes of course that requires better nodes, but better nodes don't get you better clocks unless your designs can accommodate running at those higher clocks. The A12 could never have run at 3778MHz. Getting the A17 there is a triumph, and is likely hiding the real capabilities of that core, which in the M3 will probably reach 4.2-4.5GHz.

 

[name99]

AMD keeps killing it with new microarchitecture and IPC gains. Meanwhile… Apple has had zero IPC gain the last 5 years.

Alleged AMD Zen 5 Specs Leak: Twice the Cores, 15% Increased IPC Over Ryzen 7000

Zen 5 will be AMD's first major architectural overhaul since Zen 2.

www.tomshardware.com

Ah yes. The old "my vaporware can beat up your (mismeasured) shipping product" argument.
Throughly convincing to everyone with a brain...

 

[Confused-User]

What you wanna mean by that, if I limit the fps of the Android phone to 40 fps it would be more efficient too, and no, there is some quote of Anandtech running around and it was referring to some games not benchmarks, but this time as you can see in the screenshot, he dialed down some settings in the 15 PM to get the same quality as in the Android version. Lmao and what you think how different the screen power consumption and Wifi modem will be of same modern phones. Its better to not contribute **** to the thread and just guess based on nothing, just a graph that not always translates to real life, right?

Say what?? I'm using the numbers *you* provided. If you scale the iPhone's performance down to be 10% better than the snapdragon, and scale down power usage the same amount (which understates the A17's advantage since scaling down would be better than linear), you get better power usage than the android. There's no guessing here.

I wasn't looking at the graph at all. (Though I think you badly underestimate its value.)

Yeah sure, Apple will tell you everything about it, if the A17 Pro is more efficient in every scenario then the 15 PM wouldnt lose in the battery life test of every review. I know how people like you and others in this thread are, I can do the most scientific proven test and you still would say its invalid because of some made up excuse.

Sorry, but you seem to be accusing me of your own behavior here. I just pointed out something obvious in the numbers *you* provided. If you don't like it, you'll need to explain how your numbers are wrong.

Further, your claim about "every review" is total crap. The reviews have been mixed, some saying better and some saying worse. All the ones I've read say the difference, either way, is small.

 

[Analog Kid]

And this is why the A17 Pro has better efficiency but will have lower battery life than my Android.

Can you help connect the dots for me on this-- what am I not seeing? You said the same thing about app loading, which I was about to reply to, but this time you have what looks like actual benchmark data showing less power consumption than either of the Androids you show and then conclude "that's why battery life is lower than Android".

I honestly can't see how one follows from the other. 30% less power at the same task should mean more battery life, not less, right? Maybe I missed something earlier in the thread...

 

[Analog Kid]

AMD/Intel/Nvidia tend to "overclock" their products to win benchmarks for marketing purposes.

Maybe, but I don't see where Apple ever has...

If that is normal, why does the comparison have to be at equal clock frequency? The merit is to be more efficient at the clock frequency at which SoCs run daily tasks.

The merit is in having a single core than can easily adapt to high efficiency and high performance tasks and a system that holistically optimizes for efficiency through careful design and hardware/software synergies.

 

[jeanlain]

Ah yes. The old "my vaporware can beat up your (mismeasured) shipping product" argument.
Throughly convincing to everyone with a brain...

Even if it's not vaporware, the IPC will still remain behind the A17's.

 

[leman]

@leman, if you have a chance do you think you could run your test in low power mode? If would be interesting to see what if anything that does to max clock. As I wrote earlier, there may be a case for Apple to expose some controls for this.

The data has samples for both high and low power mode! You can see the dots in the lower left area of the graph. What's interesting is that in low power mode the A17 is clocked ridiculously low — at just 1.3Ghz.

 

[Confused-User]

The data has samples for both high and low power mode! You can see the dots in the lower left area of the graph. What's interesting is that in low power mode the A17 is clocked ridiculously low — at just 1.3Ghz.

Oops, sorry I didn't check that before posting.

Hm, I don't think it's ridiculous - but it's also not really useful for the fast-but-not-fastest kind of mode I was thinking about. Oh well.

Still. I wonder what Genshin looks like on an iPhone running in low power mode? And how many fps it can do?

 

[Pressure]

Sort of. They were based on the Pentium M. The Pentium III was the ancestor of the M, so you're not wrong, but there were generations in between.

I thought I was misremembering as well but Pentium M (Banias from the Israeli Haifa team) is absolutely based on the Pentium III (Tualatin design), although heavily modified. I had to double check.

 

[Confused-User]

Can you help connect the dots for me on this-- what am I not seeing? You said the same thing about app loading, which I was about to reply to, but this time you have what looks like actual benchmark data showing less power consumption than either of the Androids you show and then conclude "that's why battery life is lower than Android".

I honestly can't see how one follows from the other. 30% less power at the same task should mean more battery life, not less, right? Maybe I missed something earlier in the thread...

He's talking about his phone, not the one in the benchmark chart he posted. His phone has only ~75% of the framerate while using ~87% of the power (and, while he doesn't mention it, displaying a much lower quality image). But since his absolute power consumption is lower, he argues, his phone's battery will last longer.

This ... is a weird argument. It depends on the phone's raw battery capacity, and it also depends on all the other elements of the system that draw power. (None of which, I'll point out again, have anything to do with the topic of this thread, which is the SoC and not the iPhone.)

There's an obvious reductio ad absurdum here: just throttle your game down to 1 fps on your android, and buy a phone built onto a car battery. It'll last for a week (month?). But who cares? Who would want that?

If the iPhone had such a small battery that it was a problem getting through the day with it, then maybe this would matter. In the real world, it's just silly, because it's inconsequential.

 

[Love-hate 🍏 relationship]

Gerard williams III departure sure hurt apple soc department

 

[APCX]

Gerard williams III departure sure hurt apple soc department

Looks like his arrival sure hurt Qualcomm. How long overdue is their M1 killer?

 

[Love-hate 🍏 relationship]

Looks like his arrival sure hurt Qualcomm. How long overdue is their M1 killer?

time will tell. atm apple holds the crown, but not for too long

 

[APCX]

time will tell. atm apple holds the crown, but not for too long

Might be better to wait until they lose the crown to start boasting/dooming.