Surface Laptop (Intel and AMD) vs M1 MacBook Air. Oh my...

[leadeater]

3 hours ago, HenrySalayne said:

Limiting the power draw independent of the battery level to a worst case scenario (battery almost drained and even including ageing) is unnecessary and just a design choice. Nothing else.

How is it just a "design choice" when the engineers may have evaluated and deemed it a requirement for the device to safely operate? I think we have different interpretations of what is a design choice and what is a technical limitation of a component that factors in to design is.

 

I do not consider it unnecessary to ensure the usability of a device over it's life, that's more a necessity.

 

If you will only ever consider this as "a design choice", then I think we can just move on as there's no actual discussion to be had. I'm not going to tell you your opinion on that matter is wrong.

 

And I still disagree with your notion that the issue only applies when almost drained, voltage drop is across the entire discharge curve so at any point this voltage loss could pass between acceptable to not acceptable for the required load.

 

3 hours ago, HenrySalayne said:

No, they are not. At least to my maybe a little bit rusty understanding of the electrochemical processes in these cells. But if you have additional information about this topic, I would love to expand my knowledge

It is because charge and discharge curves are very different and so are the tolerated C rates. You can have a look at the site I linked you to as it has a wealth of information on it. But just because you can ram 30W/40W/60W of charge power in to a single cell does not mean it's capable of supplying 30W/40W/60W of discharge at all or for any meaningful amount of time.

 

Edit: Sorry appears I originally forgot to supply the source link earlier, will provide it below along with some extra information for you.

 

And a lot if not most battery packs are blended with NMC

https://batteryuniversity.com/learn/article/types_of_lithium_ion

 

And charging graph

https://batteryuniversity.com/learn/article/charging_lithium_ion_batteries

 

Charging a single cell battery with 30W/40W/60W is a bad idea, yea it's possible and devices allow it but RIP your battery and it's short life. The same is true for high current discharging as well, you want to pull more than 1C out of it to supply the device with enough power to boost, well your are shortening battery life doing that too.

/Edit:

 

3 hours ago, HenrySalayne said:

Funnily enough this is by definition arbitrary. Why 14? Just a wild guess? I certainly never mentioned more than the two already in place.

Because it was supposed to be arbitrary (I chose a high number to make that very point), just as your graph was because it also bears no resemblance at all to how CPU and their power boost. Without active modification it's either in allowed boost state or not and the allowed boost state is by default 30W. Either you are proposing that this can be changed over the remaining charge or you are not. If you are then I'd like you to actually answer the issue of how this should be done and to a way that is actually reliable.

 

Yes there are two, you're still assuming that one of them, 30W, is actually possible to use on battery. So I propose to you the question, what if it is not with the battery in the Surface? Your opinion is far as I can tell that it is, but what if you are wrong?

[HenrySalayne]

21 minutes ago, leadeater said:

How is it just a "design choice" when the engineers may have evaluated and deemed it a requirement for the device to safely operate?

Have they? Or is it just a choice they made? The evidence points to the latter, but maybe some batteries don't follow the rule and are just weird. I'm not a battery expert.

22 minutes ago, leadeater said:

It is because charge and discharge curves at very different.

Yes, they are, but that's a completely different topic. The maximum current you can push into a cell and the maximum current you can pull out of it are somewhat related. There has to be an actual physical flow of charged particles.

 

38 minutes ago, leadeater said:

Because it was supposed to be arbitrary (I chose a high number to make that very point), just as your graph was because it also bears no resemblance at all to how CPU and their power boost.

Just a quick recap why I drew that graph:

Quote

That doesn't matter as the issue is under load, batteries do not recover load voltage when the load reduces.

But they do. As I clearly marked on my drawing.

 

As this discussion is moving nowhere and I haven't found a notebook battery datasheet supporting or disproving my claims (and you probably haven't found anything either), I would propose to call it a day until someone finds some concrete evidence. I still think batteries are rarely a bottleneck and any throttling in "battery mode" has completely different reasons. Mobile devices tend to have a lot more constraints especially in regards to cooling and battery life. Maybe throttling is also in place to prevent some private parts getting burnt. Who knows.

[leadeater]

5 minutes ago, HenrySalayne said:

But they do. As I clearly marked on my drawing.

You still missed the problem with what you drew, boost state is 30W so what you drew has nothing to do with that at all. Like I mentioned before draw your nice red line again with it returning to 2C discharge rate or your drawing isn't related to the discussion we were having.

 

At 2C draw the voltage will never be higher than a previous time along the discharge curve, that is impossible.

[leadeater]

@HenrySalayne

What you drew

 

What it would actually look like.

 

And yes I know my line is worse, a lot worse lol. That aside when the CPU boosts again at some later time the load voltage has not recovered as you tried to say it does. It will only ever decrease, it's impossible to recover without charging.

[HenrySalayne]

20 minutes ago, leadeater said:

And yes I know my line is worse, a lot worse lol. That aside when the CPU boosts again at some later time the load voltage has not recovered as you tried to say it does. It will only ever decrease, it's impossible to recover without charging.

I didn't say that at all. You said "batteries do not recover load voltage when the load reduces". But they do. Reduced load --> more voltage. That's the point here. It's quite obvious a discharged battery will always have a lower voltage for a given load compared to a fully charged one. That's basically the core of my entire argument.

And just for the sake of completeness: Reducing the load will actually allow the battery to recover a little bit. But the effect is marginal and not worth to take into consideration.

[leadeater]

54 minutes ago, HenrySalayne said:

You said "batteries do not recover load voltage when the load reduces". But they do. Reduced load --> more voltage

Have a read again what what I actually said, the second sentence.

On 4/30/2021 at 8:38 AM, leadeater said:

That doesn't matter as the issue is under load, batteries do not recover load voltage when the load reduces. Unloaded voltage will be higher but this along with load voltage always goes down as a battery discharges, it will never go up.

 

Though to be fair I should have phrased it as reduced or lower load not unloaded as unloaded more implies no load at all. Yes if you reduce the load the voltage will recover but at the same load (the load) i.e. as in your return back to the load again it will be lower than it was before.

 

It's a wording problem, maybe you didn't understand the point but the load in the discussion is a fixed amount, something we actually know (30W). So this is the load hence the load voltage does not recover by reducing to some other load as that is not the load voltage we are talking about i.e. that 30W. If the cell voltage needs to be say 3.5V, per cell, to deliver the total device power safely for the CPU in 30W boost then that is the technical limitation that has to be designed around. (not claiming it's actually that voltage).

 

X load != Y load

X load voltage doesn't go up by transitioning to Y load.

 

I think I've made my point more than clear now, load (a specific load) voltage does not recover by reducing load.

 

Edit:

And just to reiterate my original point so it's as clear as possible. For a given known load amount a battery pack of a known configuration needs to be above a certain cell voltage to be capable of supplying the device the required power for that load. This load is not variable, when the battery cell voltage goes below the threshold it will no longer have enough output wattage for the device. Voltage loss for Energy Cells is fairly quick so I would hope the battery is sized to take that in to account however the alternative solution is to just not allow as higher load while on battery, more than one way to address such a limitation.

[TheReal1980]

5 hours ago, leadeater said:

Have a read again what what I actually said, the second sentence.

 

Though to be fair I should have phrased it as reduced or lower load not unloaded as unloaded more implies no load at all. Yes if you reduce the load the voltage will recover but at the same load (the load) i.e. as in your return back to the load again it will be lower than it was before.

 

It's a wording problem, maybe you didn't understand the point but the load in the discussion is a fixed amount, something we actually know (30W). So this is the load hence the load voltage does not recover by reducing to some other load as that is not the load voltage we are talking about i.e. that 30W. If the cell voltage needs to be say 3.5V, per cell, to deliver the total device power safely for the CPU in 30W boost then that is the technical limitation that has to be designed around. (not claiming it's actually that voltage).

 

X load != Y load

X load voltage doesn't go up by transitioning to Y load.

 

I think I've made my point more than clear now, load (a specific load) voltage does not recover by reducing load.

 

Edit:

And just to reiterate my original point so it's as clear as possible. For a given known load amount a battery pack of a known configuration needs to be above a certain cell voltage to be capable of supplying the device the required power for that load. This load is not variable, when the battery cell voltage goes below the threshold it will no longer have enough output wattage for the device. Voltage loss for Energy Cells is fairly quick so I would hope the battery is sized to take that in to account however the alternative solution is to just not allow as higher load while on battery, more than one way to address such a limitation.

I am no battery expert but could you please tell me why the Intel version is so much better than AMD, performance wise, when running on battery ONLY?

In one test it went down about 50% in performance which is totally unacceptable.

[leadeater]

1 hour ago, TheReal1980 said:

I am no battery expert but could you please tell me why the Intel version is so much better than AMD, performance wise, when running on battery ONLY?

In one test it went down about 50% in performance which is totally unacceptable.

That issue itself is a little complicated because it depends a lot on how long each test pass actually is and the overall benchmark length. I personally don't run GB so I don't know how long it takes but the shorter the benchmark and the more it can stay the entire time in higher power boost state the more impact you'll see in score by restricting boost power or not allowing it at all.

 

Some OEMs have also put in delays on their Ryzen Mobile devices for boost when on battery so if a test is too short it'll never actually boost.

 

Also the Intel CPU is actually more power efficient than the AMD CPU is, 10nm 11th Gen Tiger lake is actually a decent bit more power efficient. I couldn't even begin to tell you what the boost power configuration is on the Intel Surface Laptop 4 is because almost every value related to that is configurable and Intel also has other tools which can actually change this in real time.

 

Quote

Intel’s Adaptix is a suite of technologies that includes Dynamic Tuning 2.0, which implements DVFS feedback loops on top of supposedly AI-trained algorithms to help the system deliver power to the parts of the processor that need it most, such as CPU, GPU, interconnect, or accelerators. In reality, what we mostly see is that it reduces frequency in line with memory access stalls, keeping utilization high but reducing power, prolonging turbo modes.

 

Since as far as I remember we only see the 50% performance drop in GB I would say that is an issue with the benchmark tool or how it's being used more than anything else.

 

Anyway here is a power usage graph from Anandtech in their Tiger Lake review. The 4800U system is a Lenovo Yoga Slim 7,  1065G7 Microsoft Surface Laptop 3, 1185G7 Intel Reference Design system.

https://www.anandtech.com/show/16084/intel-tiger-lake-review-deep-dive-core-11th-gen/7

 

AMD CPUs/APUs far as I've seen boost for a lot longer and that is more demanding on a battery than it is to have a very short high draw then dropping down to a much lower value.

 

Surface Laptop 3: 45.8WH (2 Cells) @ 7.58V

Surface Laptop 4: 47.4WH  (2 Cells) @ likely similar voltage to above but slightly higher which gives the WH capacity increase

Lenovo Yoga Slim 7: 60.7WH (4 Cells) @ 15.36V

Note: HP equivalent (~45WH) WH batteries are 3 Cells

 

Far as I know the Surface Laptop devices use a rather low battery nominal voltage compared to other OEMs/ODMs and this means that for the same power the current is a lot higher and as such is a lot harder on the battery. The battery in the Yoga Slim 7 is not only able to supply higher currents it is able to supply the same current as the Surface Laptop battery for longer but due to the higher nominal voltage it only requires half the current for the same amount of power. The battery in the Yoga Slim 7 is simply better suited to the task and is much less likely to impact decisions around CPU boost parameters.

 

The battery in the Surface Laptop's are pretty much at their upper limits of current output for either CPU, it's actually a litter higher demand on the Intel but for a lot shorter. The current draw is actually above the generally recommended 2C per cell where as on the Yoga Slim 7 battery it's well below 2C, like around ~0.5C.

 

The rated current output for the Surface Laptop 4  battery is 6A btw, so if the battery gets down to 6V (3V per cell, willing to bet MS runs it below 3V) it's only able to provide 36W and if the CPU alone can draw 30W for a sustained length I do not believe the remaining 6W is enough to run the screen and everything else in the laptop. So therefore it is my opinion that the voltage required to allow the CPU to boost to 30W is above 3V per cell and could be high enough that Microsoft choose to not allow full boosting at all on battery due to this.

 

I don't know I'm simply offering a technical reason for why Microsoft could have put this power limit in place, if they actually have. The supporting math for it is borderline so it's still entirely possible the battery has nothing to do with it at all, I can certainly see how it could but could is not the same as is.

[TheReal1980]

1 hour ago, leadeater said:

That issue itself is a little complicated because it depends a lot on how long each test pass actually is and the overall benchmark length. I personally don't run GB so I don't know how long it takes but the shorter the benchmark and the more it can stay the entire time in higher power boost state the more impact you'll see in score by restricting boost power or not allowing it at all.

 

Some OEMs have also put in delays on their Ryzen Mobile devices for boost when on battery so if a test is too short it'll never actually boost.

 

Also the Intel CPU is actually more power efficient than the AMD CPU is, 10nm 11th Gen Tiger lake is actually a decent bit more power efficient. I couldn't even begin to tell you what the boost power configuration is on the Intel Surface Laptop 4 is because almost every value related to that is configurable and Intel also has other tools which can actually change this in real time.

 

 

Since as far as I remember we only see the 50% performance drop in GB I would say that is an issue with the benchmark tool or how it's being used more than anything else.

 

Anyway here is a power usage graph from Anandtech in their Tiger Lake review. The 4800U system is a Lenovo Yoga Slim 7,  1065G7 Microsoft Surface Laptop 3, 1185G7 Intel Reference Design system.

https://www.anandtech.com/show/16084/intel-tiger-lake-review-deep-dive-core-11th-gen/7

 

AMD CPUs/APUs far as I've seen boost for a lot longer and that is more demanding on a battery than it is to have a very short high draw then dropping down to a much lower value.

 

Surface Laptop 3: 45.8WH (2 Cells) @ 7.58V

Surface Laptop 4: 47.4WH  (2 Cells) @ likely similar voltage to above but slightly higher which gives the WH capacity increase

Lenovo Yoga Slim 7: 60.7WH (4 Cells) @ 15.36V

Note: HP equivalent (~45WH) WH batteries are 3 Cells

 

Far as I know the Surface Laptop devices use a rather low battery nominal voltage compared to other OEMs/ODMs and this means that for the same power the current is a lot higher and as such is a lot harder on the battery. The battery in the Yoga Slim 7 is not only able to supply higher currents it is able to supply the same current as the Surface Laptop battery for longer but due to the higher nominal voltage it only requires half the current for the same amount of power. The battery in the Yoga Slim 7 is simply better suited to the task and is much less likely to impact decisions around CPU boost parameters.

 

The battery in the Surface Laptop's are pretty much at their upper limits of current output for either CPU, it's actually a litter higher demand on the Intel but for a lot shorter. The current draw is actually above the generally recommended 2C per cell where as on the Yoga Slim 7 battery it's well below 2C, like around ~0.5C.

 

The rated current output for the Surface Laptop 4  battery is 6A btw, so if the battery gets down to 6V (3V per cell, willing to bet MS runs it below 3V) it's only able to provide 36W and if the CPU alone can draw 30W for a sustained length I do not believe the remaining 6W is enough to run the screen and everything else in the laptop. So therefore it is my opinion that the voltage required to allow the CPU to boost to 30W is above 3V per cell and could be high enough that Microsoft choose to not allow full boosting at all on battery due to this.

 

I don't know I'm simply offering a technical reason for why Microsoft could have put this power limit in place, if they actually have. The supporting math for it is borderline so it's still entirely possible the battery has nothing to do with it at all, I can certainly see how it could but could is not the same as is.

I looked for a comparison between the M1 and the newest AMD chipset on a laptop and found this.

 

They say it can not even access turbo mode in "battery only" mode.

 

Have Linus talked about this (the difference in performance between run on battery/power assistance)? I know AMD has great new chips on the market but in laptops they seem to be rather limited, even compared to Intel chips.

[leadeater]

2 minutes ago, TheReal1980 said:

They say it can not even access turbo mode in "battery only" mode.

Do you have a rough time stamp for this?

[TheReal1980]

6 minutes ago, leadeater said:

Do you have a rough time stamp for this?

Yes please listen from 5:14 and he says it a couple of seconds later.

[leadeater]

16 minutes ago, TheReal1980 said:

Yes please listen from 5:14 and he says it a couple of seconds later.

I still suspect it's issue with the tests being done, probably all short enough to be affected by things like this: https://arstechnica.com/gadgets/2020/11/amd-laptops-have-a-hidden-10-second-performance-delay-heres-why/

 

I highly doubt it's not allowed to boost, it's more likely another system with the configured ~12 second boost delay and every benchmark he's using is short enough that 12 seconds is a significant portion of it. Geekbench in total only take a few minutes and it's comprised of multiple different sub test

 

Turbo mode just sounds like a power profile to me, not a statement saying the CPU cannot actually turbo.

 

Edit:

Also you can remove this boost delay on battery: 

 

However it's pretty crap this is a thing in the first place.

 

Edit2:

10% lower on battery performance for a longer benchmark, CB R23.

 

[TheReal1980]

10 minutes ago, leadeater said:

I still suspect it's issue with the tests being done, probably all short enough to be affected by things like this: https://arstechnica.com/gadgets/2020/11/amd-laptops-have-a-hidden-10-second-performance-delay-heres-why/

 

I highly doubt it's not allowed to boost, it's more likely another system with the configured ~12 second boost delay and every benchmark he's using is short enough that 12 seconds is a significant portion of it. Geekbench in total only take a few minutes and it's comprised of multiple different sub test

 

Turbo mode just sounds like a power profile to me, not a statement saying the CPU cannot actually turbo.

Ok, let's say you are correct.

This means that people who use laptops with AMD chipsets have a much worse experience doing daily, normal stuff on a computer that only require short bursts of performance. How is that acceptable?

 

Linus should check these AMD laptops more in depth I think, this could be good for Intel.

[leadeater]

5 minutes ago, TheReal1980 said:

This means that people who use laptops with AMD chipsets have a much worse experience doing daily, normal stuff on a computer that only require short bursts of performance. How is that acceptable?

Yep it'll be worse, not used one so don't know how noticeable it actually is. Also check my second edit, pretty much confirms the boost delay and on battery CB R23 is only 10% lower than on AC.

 

This is why Intel tried to make a big deal out of it, problem is nobody really took any notice. I'd like to see a day to day usage comparison between two nearly identical laptops, AMD vs Intel, and get feedback/comments about experience. LTT has enough people on staff to get a good sample. I think this is one of those times where stepping outside of benchmarks and getting real world experience is key.

 

Edit:

Hmm why not, maybe it is worth looking at enough? @LinusTech

[TheReal1980]

3 minutes ago, leadeater said:

Yep it'll be worse, not used one so don't know how noticeable it actually is. Also check my second edit, pretty much confirms the boost delay and on battery CB R23 is only 10% lower than on AC.

 

This is why Intel tried to make a big deal out of it, problem is nobody really took any notice. I'd like to see a day to day usage comparison between two nearly identical laptops, AMD vs Intel, and get feedback/comments about experience. LTT has enough people on staff to get a good sample. I think this is one of those times where stepping outside of benchmarks and getting real world experience is key.

 

Edit:

Hmm why not, maybe it is worth looking at enough? @LinusTech

I saw your edit, thanks.

Yeah, a video about this would be interesting.

[CarlBar]

3 hours ago, leadeater said:

AMD CPUs/APUs far as I've seen boost for a lot longer and that is more demanding on a battery than it is to have a very short high draw then dropping down to a much lower value.

 

Surface Laptop 3: 45.8WH (2 Cells) @ 7.58V

Surface Laptop 4: 47.4WH  (2 Cells) @ likely similar voltage to above but slightly higher which gives the WH capacity increase

Lenovo Yoga Slim 7: 60.7WH (4 Cells) @ 15.36V

Note: HP equivalent (~45WH) WH batteries are 3 Cells

 

Far as I know the Surface Laptop devices use a rather low battery nominal voltage compared to other OEMs/ODMs and this means that for the same power the current is a lot higher and as such is a lot harder on the battery. The battery in the Yoga Slim 7 is not only able to supply higher currents it is able to supply the same current as the Surface Laptop battery for longer but due to the higher nominal voltage it only requires half the current for the same amount of power. The battery in the Yoga Slim 7 is simply better suited to the task and is much less likely to impact decisions around CPU boost parameters.

 

The battery in the Surface Laptop's are pretty much at their upper limits of current output for either CPU, it's actually a litter higher demand on the Intel but for a lot shorter. The current draw is actually above the generally recommended 2C per cell where as on the Yoga Slim 7 battery it's well below 2C, like around ~0.5C.

 

The rated current output for the Surface Laptop 4  battery is 6A btw, so if the battery gets down to 6V (3V per cell, willing to bet MS runs it below 3V) it's only able to provide 36W and if the CPU alone can draw 30W for a sustained length I do not believe the remaining 6W is enough to run the screen and everything else in the laptop. So therefore it is my opinion that the voltage required to allow the CPU to boost to 30W is above 3V per cell and could be high enough that Microsoft choose to not allow full boosting at all on battery due to this.

 

I don't know I'm simply offering a technical reason for why Microsoft could have put this power limit in place, if they actually have. The supporting math for it is borderline so it's still entirely possible the battery has nothing to do with it at all, I can certainly see how it could but could is not the same as is.

 

Nice digging on the battery details. And yes i doubt very much it can boost at 30w off the battery even with a full charge.

 

That said even if it could, we come back to the fact that to avoid overloading the battery you need some way of monitoring the actual degredation of the battery in each specific laptop.so it knows which profile to use. If anything goes wrong with that you've got a recipe for a battery fire. And thats litigation city right there. Companies aren't going to screw around with that unless they absolutely have to.

[HenrySalayne]

4 hours ago, leadeater said:

Far as I know the Surface Laptop devices use a rather low battery nominal voltage compared to other OEMs/ODMs and this means that for the same power the current is a lot higher and as such is a lot harder on the battery

I'm sorry I have to get into this discussion once again. Lower voltage doesn't necessarily mean more stress on the battery. You can divide every battery into small equal units. Higher voltage just means, these units are connected in series and not in parallel. If you take a 40 Wh battery with 7.2 V and a 40 Wh battery with 14.4 V and you pull 40 W of power, the load is identical. You can split my example batteries into 4 pieces of 10 Wh and 3.6 V each. If you do the math of a loaded 2 in series, 2 in parallel configuration comparing it to a 4 in series configuration the current draw for each piece is the same (2.8 A).

For the same battery technology capacity and current capabilities are proportional. If you double the capacity, the current capabilities also double.

4 hours ago, leadeater said:

AMD CPUs/APUs far as I've seen boost for a lot longer and that is more demanding on a battery than it is to have a very short high draw then dropping down to a much lower value.

I'm a little puzzled. Didn't we just discuss this in length and came to the conclusion there is basically no difference in short peak loads and sustained loads, because the battery doesn't recover in any meaningful form?

 

Edit:
I don't think Microsoft is a good example to talk about batteries. As far as my experience goes, they are not good at it. If you want to see some decomposing bloated Microsoft batteries, check this video out.

 

[leadeater]

3 hours ago, HenrySalayne said:

I'm sorry I have to get into this discussion once again. Lower voltage doesn't necessarily mean more stress on the battery.

True that is fair point however that depends on the actual configuration of the battery as you point out after this. If you can find a 2020/2021 device in this class that has 4 cells and is ~45WH battery then I'm all ears.

 

Other problem is we are at the lower end of the WH capacities with modern cells so an increase in voltage is nearly always going to increase WH capacity or you have to use older generation/manufacturing cells to achieve that higher voltage at the same WH capacity.

 

4 cell parallel & series ~45WH batteries are in Intel 8th Gen products

2/3 cell series ~45WH batteries are in Intel 10th/11th Gen products

 

It's not actually possible to raise/ lower~45WH battery voltage without increasing/decreasing WH unless utilizing older (lower capacity) battery cells that probably aren't in production anymore. Unless there is something I am not thinking of? 2 cells in series the minimum possible as single cell voltage is too low.

 

Surface Laptop 3/4: 2 cell 45.8WH/47.4WH (7.58V, 2S)

Lenovo Yoga Slim 7: 4 cell 60.7Wh (15.36V, 4S)

HP Envy x360: 3 cell 51WH (11.55V, 3S)

Dell XPS 13: 4 cell 52WH (7.6V, 2S2P)

Acer Swift 3: 4 cell 56WH (15.4V, 4S)

Dell Latitude 13: 4 cell 63WH (15.2V, 4S)

MSI Modern 14: 3 cell 52WH (?, 3S)

 

I can't think of or find any laptops in this similar device class in the current model year 2020/2021 that uses both a 2 cell battery and/or as low capacity as the Surface Laptop. Even the M1 MacBook Air has a 49.9WH (11.4V, 3S) battery. From what I can tell only Dell is actively using smaller capacity cells in a current product, so proof edge cases exist but the trend I can see is higher voltage means higher capacity, due to increased cell count.

 

3 hours ago, HenrySalayne said:

I'm a little puzzled. Didn't we just discuss this in length and came to the conclusion there is basically no difference in short peak loads and sustained loads, because the battery doesn't recover in any meaningful form?

If a CPU is in a higher power boost period for significantly longer does logic not dictate that it would have used more power meaning it is more stressful on the battery as it has drained it more and increased it's heat more?

 

Quote

Tiger Lake: 241 seconds for 4082 joules, averaging 17.0 W
Renoir: 234 seconds for 5386 joules, averaging 23.0 W

The same task took similar amount of time but the 4800U used more energy, 30% more, to do the same work.

 

The AMD Ryzen Mobile CPU only uses less energy when the workload is longer and utilizes more threads so is able to finish the task sooner. But that does mean the Intel device can run further in to the discharge voltage, also think Intel has much better device power controls and features than AMD does.

 

If I had to guess it's probably why that 10 second boost delay on a lot of Ryzen Mobile laptops exist, 30% more energy usage on shorter/lighter loads probably something you do want to prevent. I had forgotten this was even a thing until today, I had known about it before but didn't really look in to it so forgot.

 

3 hours ago, HenrySalayne said:

I don't think Microsoft is a good example to talk about batteries. As far as my experience goes, they are not good at it. If you want to see some decomposing bloated Microsoft batteries, check this video out.

Well Microsoft doesn't actually make the batteries and they are supplied by the same manufactures of them as everyone else. My old HP work laptop had the same battery swelling problem as well, cause by overcharging due to using a dock. Probably the same cause here, overcharging of the battery. You'd think something as basic and known to be critically important wouldn't happen, but it does 

[TheReal1980]

I would like Linus and the gang to inspect this much further with real life situations.

How many times can you render a specific movie clip for example by only using the battery? Does each rendered movie clip take longer with time? How long did it take to render it 20 times et.c.