[Guide] Improving Battery Life on Windows [+Enabling Deeper C States]

So I have got a really strange issue with my PM981a and Y540 which possibly reduces battery life to a great extent.

After a clean install of Windows, my NVMe idle temperature while browsing is at 41 degrees which is great and what I would expect.

Then, after copying large files, temperature will rise up to 70 degrees. That's okay as well. The thing is, after that it won't reach these low idle temperatures I had before until I reinstall Windows again.

You would think it's the heat stuck in the chassis, but it isn't. Even after a night of cooldown, the next day my NVMe idle temperature while browsing is at 54 degrees, far higher compared to before stressing the NVMe SSD. I have ruled out ambient temperature and software changes as well, it's exactly the same. Just the idle temperature of the NVMe has changed.

Obviously, more heat also directly translates to power draw, so it worries me that this affects my battery life badly. What could this be? I can't wrap my head around it. Maybe it's a bug where the NVMe SSD is stuck in a certain power level once its stressed? IDK. Is there any way to test this?

What's everyone's NVMe idle temperature while browsing? Maybe I'm not alone with this issue.

 

I am using a Silicon Power P34A80 M.2 NVMe SSD in my Acer Nitro 5 for over 2.5 years now and temps. have been problematic since the beginning. But it's worse nowadays.
In basic workloads, even in gaming it's fine (idles in the early to mid-forties ) but in case of sustained transfers even low bandwidth ones like over MTP limited to USB 2.0 speeds it throttles hard after a few minutes or so. Reaches 80-85°C easily. During that time SSD health in CrystalDiskInfo is shown as "Bad" but afterwards it changes to "Good"!
What's strange is that often the temps. don't fall back to normal even after the workload is finished. Putting the laptop to sleep for a few minutes is the only solution. I suspect poor ventilation. There's no heatsink or thermal pad or anything similar in the mobo for the SSD slot. Heck the M.2 screw wasn't included. The SSD itself didn't ship with any heatsink. However in the time the battery was in good health the SSD didn't seem to impact battery life negatively.
Edit: I flashed a firmware upgrade yesterday and so far throttling seems to be pretty much gone.

 

The minimum brightness of the display in my laptop is a little too high to work comfortably in a dark environment ( background illuminated with an LED light ). So I use an app called Color Veil to reduce the brightness even further. But I recently noticed that it has no effect on the battery life!

 

Usually the minimum LED backlight brightness is defined wtihin your laptop display controller firmware. I'm not sure there's any software that will change the minimum DC dimming or PWM signal on the minimum brightness. What I think Color Veil does is simply make all the colours darker, which doesn't make a difference to batter life unless you're usning an OLED screen. The display backlight, which consumes the majority of display power, still uses the same amount of power, if that makse sense.

 

Yeah. That makes sense.

 

I'll have to dig it out, but IIRC my old VAIO VPCZ1 had a registry setting to increase the maximum brightness beyond what was the normal maximum brightness.

 

The minimum brightness anyhow will be next to know electricity draw by the screen.

As long as you use Intel processor and igzo screen processor is the main cause for battery drain. Apple M1 is the only laptop processor to even out CPU and display right now. Let's hope for big little concept

 

Thanks for linking me this btw, I tried many tricks a few months ago but couldn't improve battery life by a meaningful amount. The slim 7 is already really good out of the box.

 

I don't know how relevant this is to this topic, but it might be a good idea to check this if your battery is getting drained even if the laptop is switched off.
I shut down my laptop but didn't close the lid. When I switched off my bedroom lights prior to going to sleep I noticed a white strip on the screen of my laptop which wasn't visible with the lights on. Apparently the first backlight LED (bottom left of the screen) was not completely switched off. I switched on the laptop and shut it down again, and the problem went away - the screen was totally dark. I've noticed it on several occasions subsequently.
(btw. this is my VAIO VPCZ1 - an 11 year old laptop currently being used for testing Windows 11).

 

So I've switched my boot device from my PM981a NVMe to my SATA SSD and I saw an incredible boost in battery life from that. If you have a SATA SSD drive, you should really do that as well, because Windows constantly accesses the NVMe when its the boot drive and those can get pretty power hungry, especially Samsung.

I saw no drop in responsiveness or anything like that. But battery life on my aging Y540 went from 4:40 hours regular use to around 5:20h. If my battery had its design capacity, it would easily reach 6 hours when browsing the web, watching YouTube videos or doing office stuff, which is pretty great for a gaming laptop.

 

Thanks for posting your findings.

I think a lot of the time it may come down to certain NVMe power saving modes that are not emabled or incorrectly configured in the BIOS. A fair few laptops can run NVMe laptops with no battery drain (SN520, PM961 in my experience).
The converse can apply for SATA SSDs. YMMV

 

Just switch off indexing and other crap. With devsleep enabled my c drive only is active 1/20 of the time. I moved temp files to ramdisk, especially your browser temp directory... And use edge canary with performance mode on....

You can then check it works by looking at time on of the drive in smart data. After 2000 hours windows on my nvme c drive only has 160 hours on time.... But yeah get an efficient laptop with big battery in first place. Best wait for alder lake. It's gonna be way better. Current CPUs are ****ty in light work. They are only efficient at idle and video.

 

Hi there! I've been trying to get the most out of deeper C-States on my i5-8250U Lenovo Yoga C930-13IKB and have ended up with seemingly reasonable numbers on idle, with my system drawing about 3.6-4 mW on average at a 33°-36°C C0% temperature on battery power. [ https://imgur.com/a/N1F7GvC]

I'm curious as to whether there's any way of allowing the system to enter C9 and C10 states, as I know in your guide you mention that enabling "Panel Self-Refresh" allows the system to enter those two states. Is there any known means of tweaking one's system to actually access those two deeper states besides being a togglable setting via the Intel Graphics Command Center / Intel UHD Graphics Control Panel? Apparently the "Panel Self-Refresh" setting is via the "Power" tab within the "System" section in the Intel Graphics Command Center application, but sadly it's not an option on either one of my two Windows 10 devices ( https://imgur.com/a/M2HJP7I).

 

I am not aware of any way to enable PSR unles syour dispaly panel supports it. Unfortunately, it seems to be a requirement for C9/10, so that's not an option for you

 

A little OT maybe: My first laptop was a C2D HP Pavillion with a 48Wh battery. Battery backup was poor at just 1.5hrs (Nvidia dGPU didn't help) but standby was pretty decent at close to a week (I didn't actually test it over a week but extrapolated it after a couple of days use). But in my 8th gen. Acer Nitro 5, when brand new, standby time was actually lower at about 4 days(again only checked over a day or so) with a similarly sized 49Wh battery. Battery backup was vastly better at 7-8.5hrs after being properly optimised following this guide. Could the difference in RAM (3GB in the HP and 20GB in the Acer) alone account for this discrepancy? Significant drop in voltages from DDR2 to DDR4 though.

 

Look for settings in the BIOS and in power options and disable:
1. Wake on LAN. This keeps the LAN port partially alive when the laptop is switched off.
2. Allow powering external USB devices when the laptop is off. This allows the laptop's battery to be used as a power bank.
3. Looking for hotspots/ public wifi.

Personally, I prefer to shut down rather than go on standby. Learned that lesson when I closed the lid and put the laptop in my bag. When I took it out, the screen was black. Something had prevented the laptop from going into standby and the poor thing cooked itself in the bag. Recovered after I let it cool down.

 

1.Wake on LAN was disabled IIRC. I will check again to be sure.
2. There was an Acer utilty to block this but sometime after it stopped having any effect.
3. Connectd Standby isn't supported. So I think after being put to Sleep Wifi is off?
I mostly hibernate the laptop but in the last few months hibernation isn't working as reliably as in the past.

 

Just in case someone is curious about this, I managed to dig out the instructions I had saved in one of my many external drives.
CAUTION: This hack worked in 2015 on Sony Vaio VPCZ1. DO NOT apply it blindly on your system. I'm merely suggesting that we use this information to guide us in doing further research into the maximum brightness issue.

Anyway, here is the post:
http://forum.notebookreview.com/threads/how-to-fix-dim-screen-vpc-z.497093/

Code:

You guys are gonna love this...I've mentioned a few times that when running other os's the screen on my Z runs significantly brighter than in Windows. Well I've found the culprit. Sony limits the display brightness to less than 100% even when you crank it, disable auto-dimming, etc. Please note that there "may" be a reason for this...longer life of the panel, lower heat, etc....but my guess is it's all about battery life. Here's the fix:

HKLM\Software\Wow6432Node\Sony Corporation\Vaio Event Service\Settings\VESAutoDimmer

You will see brightness levels in there - P00 through P08. P08 is used when you set the brightness to max...note it's NOT 255 but it should be! Change the value of P08 to 255 (make sure you select decimal or who knows what will happen) and reboot. Enjoy your significantly brighter screen. 

 

hello everyone, i hate to beg but ive had poor battery life on my Razerblade Stealth late 2019 GTX... ive basically given up at this point... fresh reinstall of windows 11, most recent september 2021 intel drivers and nvidia drivers, disabled services ive had this laptop for 2 years now and its been the same ~4hours of light use. The one thing that has even persisted through the reinstall is this one intel display driver bug in windows update. It basically keeps trying to install a 2019 version of the intel graphics drivers despite having the latest version. Ive tried to uninstall my current one and let windows update it but it doesnt work, i wouldnt think this would cause such heavy battery drain but that the only thing i can think of. Other users have reported 8-10 hours depending on the panel, I have the 4k touch (which was a mistake for a 13inch laptop) but I have a hard time believing that I should be getting only 4 hours. For the entire time ive owned this I always thought it was normal but seeing other peoples wattage below 1 where mine is always at 1.5 i know something can be done. (I dont know how to attach images please help im new lol thanks)

 

Can you send screenshots of your Task Manage rWindow sorted by CPU, Throttlestop main window, as well as batteryinfoview main window please?
Consider uploading to imgur, and sending links to these.

 

Thanks Che for responding, I just have to say your guide definitely improved the idle wattage. Im not sure which website to download batteryinfoview, as ive never used it and theres like 100 different websites for it, could you send a link for the one you used please? Thanks again for the quick response.

 

BatteryInfoView was linked in my first post. Alternatively, http://www.nirsoft.net/utils/battery_information_view.html Scroll down to where it clicks BatteryInfoView

Your Throttlestop main window confuses me. C0% is acceptably low, and you have aC10% working (which is rare) Yet almost 50% of your package is stuck in C2/C3. I am not sure if this a platform configuration issue, which you have little control over.

I've never seen a situation where that's been the case. My best guess is that there is some software in the background that is doing something. Are all Razer software services disabled? Don't just click quit in the bottom right taskbar, but did you follow the services.msc or msconfig guide in the first post? I am seeing NVIDIA Share processes in your Task Manager - do you need this? If not, just disable it

Please go to the Task Manager Processes Tab, Maximise the window, and then click on CPU. It should look something like this. Send a screenshot of it please



Next time could you plesae just copy and paste the URL, so that I do not have to inspect element to look for it

 

Yes thank you for taking time with this, I actually don't use synapse and have it completely uninstalled (its the worst software I've ever used). I have a BIOS mod which essentially just unlocked the thermal limit that is set by razer but doesn't affect anything while on battery. My battery life before and after the mod hasn't changed. I will disable the nvidia stuff like you said. I just today found something really odd which is while im in chrome right now the wattage stays at 1.1 and even just 1 sometimes where if im looking at the desktop its at 1.5-1.7. Is this normal behavior? And also I didnt even notice that my package was staying at C2 C3, I knew C10 was good but I only ever get like 30% in C10 and no more.
https://i.imgur.com/HsIbNlS.png
https://i.imgur.com/yrhduNJ.png

 

ok, for some reason it wont let me edit my reply but ive made a discovery. I can get TS to show this,
https://i.imgur.com/Z99g96h.png
with my C10 being at over 40% and my C0% being way below 1 which ive never seen before. This only happened after closing and opening the lid and then opening chrome and messing with some tabs and going back to the desktop. I also noticed my laptop went to sleep and woke up super quickly. (theres a green led that turns white when its alseep) However before I messed with anything or opened anything it stayed at 3 watts with the C0% being around 1. This was after disabling nvidia in services and msconfig. C2 is better but still there, this is insane I feel like im getting closer to the culprit. There must be some driver issue with nvidia or something what would you recommend? Thanks for dealing with my rambling lmaoooo.

 

Something unrelated but might help is the HP crap you have in the background - don't you have a razer laptop? What is HP software doing in your background? You should disable anything that you do not use in msconfig or services.msc

Your CPU C0% is excellent. A+
Package C state at 95% is also A+
The issue here is something is periodically blocking the package from entering anything below C2 and/or C3

C2 and C3 high active states will leave your package power high. Anything belowe C6 is "good enough" to get down to a package power of 0.4-0.6W IIRC, depending on your CPU.


Driver issues/incompatibilites are extremely difficult to diagnose. If possible, either do a clean Windows reinstallation, or use a razer-provided factory image and see if CPU Package power is reduced.

 

So I did a full clean install again of windows 10 this time instead of 11 and the last 2 times must have been bugged because this time windows is butter smooth. Unfortunately the same characteristics regarding C states remain, nothing changed. I have the latest BIOS and firmware from razer. Everything from services disabled from msconfig too. I cant find a razer factory image but its running extremely well rn on balanced mode and all the same settings. I really dont know how this clean install was different from the last two times but wtvr thats my luck lol. TS shows that my wattage is stuck at 2.7 - 3W and randomly will go way down and then enter C10 when im not looking so what you are saying makes perfect sense. Seeing as this is a straight up fresh install I could only think that some of the factory drivers are bugged. I updated the intel one to the 2021 which had no effect and installed nvidia driver because without it, my fans would be ramped up constantly. Audio drivers, thunderbolt, touchpad and im sure a couple other things got installed through windows so those must be the only things but when I go to search for an update they all say I have the latest ones, what would you suggest? Thanks.

 

I have a newer Razer Blade Stealth 2020 (13") and I can tell you that the C10 state reporting is broken. Sometimes it eventually shows, sometimes it never reports again. Sometimes closing lid or standby, or a full restart works.

However I say that it is bugged because even when C10 doesnt report, my discharge rate is still at around -2.500w, which is the expected drain when C10 is fully kicking in.

My tip, watch the discharge rate when C10 actually reports properly, and assume that is the correct discharge rate when C10 works. If it doesn't report, just check discharge rate anyhow CPU Package power is also equally bugged when C10 doesn't report.

On my desktop PC with modded bios, there are some C10 options as well to fix certain bugs, but for my desktop it isn't relevant as I don't want aggressive power saving there. I don't see such an option on my razer blade bios (which is also exposed to show hidden settings etc).

Anyhow I get about 50% C10 on Idle with custom modded LTSC 2019 and new 21H2 with discharge of -2.500~mw with battery estimate of 22 hours, cpu package at around 0.250mw~ (Brightness and keyboard backlight on lowest)

I am actually chasing another 100mw off my CPU package which I've observed on occasion, but I cannot find out where this comes from. Maybe another calculation bug and I'm chasing a ghost. Either way I think I have achieved the absolute lowest possible battery drain I can achieve.

 

C10 gives around 400mw less. My lg gram 16 will then need around 1.8w overall with 600-800mw package (screen is still quite bright at 50 nits, other laptops go lower). So the cpu package is quite a problem on an otherwise super efficient laptop.
On idle it can run around 50 hours. But in real use depending on display brightness only 11-18 hours. Watching movies gets 25 hours or so, Intel is good at that. Clearly CPU is the main problem, Intel needs to catch up here. For AMD I think it is more stuff like WiFi and Mainboard and so on that is really inefficient. Intel has that dialled. Displays got very efficient too. . Nvme with Dev sleep too.

The biggest change in battery life with Tiger Lake is the browser. Those numbers are with edge and performance mode on, with chrome or Edge without performance mode I can cut away 1/3 of runtime. Somehow edge with performance mode is way easier on the CPU. This difference isn't there nearly as much with older Intel CPUs. Tiger Lake really has a scheduler problem I think. Light load it sucks if not optimized Software.

This guide here is missing in that point. The most important thing is sticking to Windows apps if possible (eg watching movies) and using Edge with performance mode. All other things are peanuts in comparison if the laptop comes well organized from the manufacturer.

 

Did you try working with EPP value and/or exposed (unhide) power plan settings? There's like a dozen options that can be tweaked.

Also in BIOS, assuming you have unlocked bios with exposed options, there are things to tweak about C-states, latency tradeoffs, Demote/promote behavior, clock scaling etc.

Furthermore downvolting various cpu (related) components can obviously help for mid to high load.

 

Yes I used the full arsenal, and the lg gram bios can be opened to a huge setting of arsenals. But it all changes very little.
The main change is the nvme devsleep which was not enables by default and maybe saves 50-60mw on a Samsung 980 (non pro, the pro uses more power).

I even undervolted on bios level but that doesn't change anything on tiger lake U.

But then there must be a reason that the lg gram wins all battery life tests. LG tweaked a lot and must have carefully chosen the components and already changed most windows settings.

So the main thing really is which apps/software you use. And windows inbuilt apps are really low power compared to any other software out there. And the edge performance mode is clearly super aggressive too. I have been using it since 10 months or so on the dev channel, and back then it frequently crashed websites. Its not as efficient anymore but the bugs are gone.

Maybe if your laptop doesn't support c10 then those tips do much more.

Oh yeah another tweak that works well is not using antivirus, the windows antivirus is actually not battery efficient. Other antivirus are less demanding (I switched to the free Panda AV). I have a windows server for compiling stuff. And on my usual compile of 25 hours, without antivirus the time went down to 24 hours. And that was already after I excluded all relevant folders from scanning. Before it was like 25.5 hours (Ryzen 3600 dedicated/64GB Ram/980 pro nvme).
But I don't feel safe without AV on my daily driver. On the server it's different, I don't install anything that I don't absolutely need).

I'm sure actual undervolting would help. Well actually I undervolted the RAM but on lpddr4 this doesn't change a whole lot.
On other laptops the guide helps more. But really if it's not misconfigured or a lot of autostarts and you use the browser a lot, this is the main thing that important

 

DevSlp isn't that for SATA-only devices?

Yea I don't use any AV either and I don't need it. Got a pretty locked down Windows system and I know exactly what I'm doing. Must say though built-in windows defender didn't drain anything at all on latest LTSC 21H2 from early tests and idle consumption was pretty good, but that was before I run my post-install script to disable it entirely

 

I finally got CPU Package power down to 0.5W (0.52W exactly ) in my 8250U laptop. Idle power consumption with display off as reported by Battery Info View was 1.1-1.3W. This is quite surprising with battery wear level at 40%, as when brand new it never went below 2W, 2.5-3W in fact IIRC. Since I had borked my SSD by flashing an unofficial firmware (in the hope of alleviating abnormal thermal throttling ) I was on the HDD and I hadn't installed Bitdefender AV. I had also disabled Windows Defender realtime protection as it was interfering in a Windows PE project. I did install the replacement SSD yesterday and idle power continues to be low. The OS is still running off of the HDD currently.
Edit: Moved the OS to the SSD, reinstalled Bitdefender and idle drain continues to be low albeit in the 1.4-1.7W range.

 

So sorry for the late reply, I really appreciate your insight and it makes a lot more sense now. I too modded my blades bios (mainly to get rid of the stupid synapse junk) and be able to utilize all of the potential of the laptop. Ill be honest im not familiar with the terms you used (LTSC and 21H2). I have gotten battery estimates of like 10 hours at times but always end up getting around 4 in real world use (web-browsing, some very light applications) I am curious to know what real world use you achieved seeing as we have similar speced laptops with the same battery size. Again sorry for late reply and I really appreciate your help.

 

I wanted to add that in normal use I get around -6000w to -7000w discharge and when I close out all applications and lower the brightness I get -5000w at best. Definitely nowhere near the -2,500w you are talking about and would like to know what you did differently as I probably missed something along the way. The only battery related bios options I have unhidden is the C-states option, but im sure there are more that I missed. Thanks again.

 

In my Acer laptop with a U-series CPU all I had to do was to update the drivers and stop unnecessary third-party apps. For the former I made use of a driver updater utility, one from Ashampoo being not too frustrating. With display on and at the minimum brightness idle drain is around 3W.

 

LG Gram 16" with display at lowest brightness is at 1.8w discharge. Disabling some stuff that I rather see essential like antivirus and other services and lowering display resolution to 1920*1200 can even go down to 1.6w.
It depends however how well your laptop is configured in hardware.
I'm sure 1.2w will be possible with alder Lake U15 processors.
More importantly is light work, like browsing. Hard to get below 3.5-4w with Tiger Lake. Apple M1 is way better there. That's why I have hope for the future. It's mainly the CPU package that shoots up for browsing. Watching movies is better...

 

After using the tool only to undervolt XPS 13 7390 following the guide's recommended amount, it is getting kernel-processor-power event 37 constantly, stating one of the processors is being limited by the system firmware.

I reset the BIOS, clean installed OS, same error in the event viewer. Did this permanently damage the laptop?

 

I have come across this message as well. I think it can be safely ignored as at least in my case while overall performance wasn't affected and thermals were appreciably improved. I used Throttlestop to undervolt my previous laptop and it lasted 8+ years without any concern.

 

Thats good to know, I got it down to -3,000w on battery drain however I still run into the issue where if I put it to sleep and wake it back up the C states get messed or something and then my battery drain freaks out and idles at anywhere from 3,500w - 4,000w to 6,000w at seemingly random. I have some unidentified devices that need some driver help but I cant seem to update them idk how. I also dont know which driver update program to use because they all seem fishy and or expensive. If I do a fresh boot and get the consistent -3,000w battery drain i get over 15 hours at idle which would be nice.

 

Try the one from Ashampoo. No adware bundled. Only issue if I remember correctly is that it only allowed one driver to be updated at a time. Anyway after I had finally managed to update all the drivers I happened to stumble upon a command line utility from another forum that backups all installed drivers and later can restore them with just a few clicks. So I won't have to deal with these shady driver updaters in the future.

 

I followed this guide for XPS 13 7390, https://brendangreenley.com/undervo...hermals-battery-life-and-speed/#cpu-undervolt

And I only need to boot EFI/grub from a USB, but XPS 13 7390 has no legacy boot option and does not recognize an EFI only bootable USB. I tried different ways, but it just won't show up on the boot list.

Anyone know how to get around this?

 

Just wanted to update some random people if they ever find this:

The spectre line of hp laptops seems to have issues with having the package enter different c-states after reboot. To fix this, you need to connect a charger.

At complete idle, im sitting at 0.3 W cpu usage (0%) and total system average of 2.8 W on my 2019 spectre x360 15" w/ a dgpu of a 1650 w/ a 4k ips screen.

Another tip is to set speed shift at 160 instead of something higher. I've found that anything higher seems to cause the cpu to spin up more once it is activated causing more power consumption. This dropped my surfing power consumption average by 0.5 watts.

All in all, my only complaints are that it seems the H cpu's require a lot more "oomph" or power to get going for doing basic tasks compared to other lower end u cpus, even though they should be close. Also, it seems like a lot of non-cpu things (ram, southbridge, etc) require the majority of the power at idle.

 

Of course it does. The H CPUs use transistors similar to the desktop ones and are more capable of reaching higher frequencies(and use less power at those higher frequencies). The tradeoff is that the leakage increases, which increases idle power.

I don't think Alderlake is going to be better. It's equal, or even worse. They said absolutely nothing about battery life. The E cores this generation is about more performance. The H series could potentially be better because of moving the PCH to on package. Not a guarantee though.

I think they will redirect the focus to battery life starting with Raptorlake.

However, I see Meteorlake is where it'll really make the difference, since they'll move the PCH effectively on-die with Foveros. Intel's problem is the off-die PCH.

AMD's Ryzen Mobile 6000 "Rembrandt" is going to beat Intel in battery life this generation. They went on-die with the I/O die long time ago, but were behind in power management. But I think they are not behind anymore.

 

Intel announced last year in April that alder Lake will only use half as much power for low power usage cases and idle with alder Lake. Since then it's been quiet however..
We need to wait for u15 and u28 processors to be out to assess this. On the high power end and will be better, but for office use I tend to believe Intel will be at an advantage. So far and hasn't matched Intel there yet even closely. Some Asus Intel based and lg Gram are way lower than anything running on AMD. But then they are still way behind Apple with ARM

 

Intel wanted to get lower power with Alderlake, and I bet they didn't meet that goal.
https://adoredtv.com/exclusive-alder-lake-m-and-p-power-consumption-figures/

The silence around battery life is key. They talked all about Alderlake-H performance gains over the competition and the predecessor and I believe it. In the H space Rembrandt is not much faster and Alderlake-H is.

Contrast that to AMD that claimed significant gains in iGPU and battery life. 30% gains, which is huge! Notice they are claiming this at the laptop-level not SoC. Notice AMD didn't say anything about -H gains, but they said -U will be 20-30% faster.

That tells me:
-Alderlake-H will have the performance leadership
-AMD will have the -U MT, iGPU and battery life leadership

Intel also claimed battery life gains with Tigerlake but it didn't materialize. Either they missed their goals or the feature got pulled. Claims of half the power jives with the adoredtv rumor of where Intel wanted to get to. Also, Intel was likely talking about half the chip-level power. Which might result in maybe 5-10% improvement.

You can tell from earlier pages in this thread that Tigerlake is worse than Comet for battery life. The battery life for Intel will not improve significantly until they get Foveros out to be an effective on-die implementation, which AMD already has.

https://www.notebookcheck.net/LG-Gr...-finally-more-performance.541046.0.html#toc-7

I don't know where you are getting this. Intel has an advantage in video playback, sure but they are slightly behind in web browsing and max load. The LG Gram only does ok because of the massive battery size. Whatever lead they have in video playback will be lost and then some since they are claiming 40% reduction in video playback power use!

 

Intel is doing pretty well at web browsing with newest edge versions and performance mode enabled. Without performance mode or any other browsers it sucks.

Video Intel is ahead too. And yes any high load Intel sucks Vs AMD. Also i7 needs a lot more power than i5 for the same work with Tiger Lake.

But yeah, there has been very little improvement since Intel 8000 series.

Where Intel is usually ahead if AMD is WiFi, and other stuff.

With the LG Gram you can idle at 1.6w per hour. So it's not only the battery that makes the difference. Browsing at 4-4.5w with edge and performance mode at 150 nits screen in the 16". So 16-19 hour battery life in real World use ain't bad. But it could be way better if Intel gets down to arm levels for idle and low power use. Let's see if the u15 and u28 can decrease it a bit. The other alder Lake aren't there and not optimized for it.

That gram review is on default cluttered windows and without performance mode. Decluttering windows including getting rid of antivirus and using edge with performance mode they could have gotten 20-30 percent better runtimes. The display and rest of periphery on the gram is really efficient. It's mainly the CPU that's the culprit. Plus there's the nvme lottery. The non pro Samsung 980 is really good at sipping power. Samsung pro series always neeeds some more juice.

 

The reviews of the ADL-H are coming out and despite the on package PCH the battery life is noticeably worse. Maybe some is due to the Ti Turing.

1.6W is pretty low but that's not the problem. Icelake could get nearly 30 hours with the XPS 13. But in actual workloads it does much worse. Who cares about idle battery life?

The problem is again the off-die PCH. It's too far away so it cannot power down fast enough. Yes you can get idle power down but saving power during web browsing is by being able to go into low power mode in between load scenarios hundreds of times a second. If it's too slow, then it won't be able to.

AMD can also use Intel WiFi, which is the industry leading WiFi chip. Most AMD devices do use them.

Intel CAN get to ARM levels, and they achieved that with the Tablet oriented Atoms. It's not that Atom chips were particularly low power, but it's because they addressed it at the platform level. Despite being 2x as fast the power used was less than half at the platform level compared to dog slow predecessor Atoms. My Venue 8 Pro got 6-8 hours of battery with a 17WHr capacity. It had the same 1.6W idle with screen on but it only needed 2.2-2.5W when browsing the web and watching youtube videos(which weren't accelerated).

 

I have only seen reviews of notebooks with discrete graphics and the core i9 HK.
As I said, we need to wait for u15 and u28 models to get a picture. The H series was never intended for saving battery.
There could be besides higher margins per chip, other reasons why the U series comes later. Intel was never good in switching cores off or especially memory. That's why for a notebook aiming for productivity the i5 was way better than the i7. Same multicore speed, slightly slower single core but 10-20 percent better battery life, while the i7 was only faster in things you don't notice, as heat for most notebooks was the real limitation after a few seconds already (with cooling able to handle 20-30w constant load), while if you wanted the higher speed with better cooling an H series chip was the real alternative.

I am pretty confident, the u15 will reach tiger lake 28w tdp performance at much lower power use. The U28 will be the same efficient on low power things but twice as fast for multicore performance.