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

Spectre x360 15t here. I can't seem to find a way to get my package to drop to a lower c state. It is hanging in c2 most of the time, and c0 even more. I feel like I'm down to the bare minimum here without removing stuff that is actually useful on the laptop.

 

Icelake isn't affected by Plundervolt. The new bugs that are popping up seems to be hitting Skylake-based cores.

 

have the same cpu as you I5 8250u whats a good Mw usage for netflix i am currently seeing 4200 lowest loads and up to 8000 flipping web pages. I was wondering if its possible to get the 8250u down to around 3000 mw while browsing the web....any advice and also throttlestop seems to be interfering with power saving settings.

 

@Spartan any ideas bro

 

How many times do people have to ask for screenshots? The TS Main Window, C Package States, FIVR windows, Task Manager, and HWInfo sensor screenshots should be standard bare minimum.

If you can't get into Package C states, it seems to be like you are the minority. Not many other x360 users seem to be reporting the same issue. Drivers? Power Configuration?

Your thousands of millwatts seem to be reporting total system power draw from the battery. This includes things like the PCH, display, RAM, HDD, etc. There is actually very little you can do to modify non-CPU power consumption, short of enabling PSR, DevSleep, etc.

I have never known a laptop to be able to actively browse the web with a total system draw of 3W. Keep in mind that the standard laptop display uses at a minimum 1W. I'd expect the SSD, RAM, and onboard circuirty to use another half a watt. That leaves just 1W for the CPU. A good way to measure your CPU's impact is to create a baseline - that's your laptop idle with display at min brightness and at the desktop with ZERO background programs.

It is impossible to ascertain as to what you mean with your last sentence - interfering with what power saving settings? Can you give an example, please?

 

2019 Razer Blade Pro 17

After following the majority of these instructions (I pushed my UV to -0.150Mv Core/Cache, 270 Nits Brightness, keyboard light on), I was able to successfully increase my battery life by 1.5 - 2.0 hours. With normal use (web browsers, ebooks, music etc), I get a total of 5 to 6 hours of usage. Note that the 2019 Razer Blade Pro 17 comes with a 70Wh battery and is known for being strictly a "desktop replacement" machine. I have been able to work on this machine while on travel thanks to this guide!

 

I'm so glad I came across this thread. Thank you so much for the all the work you've done on this issue.

I just purchased the brand new Aero 15 OLED with a i9-10980HK and 2080 Super Max-Q and was only getting 1-2 hours of battery life. I was so close to sending it back even though I loved everything else about the laptop. In fact, I even had the vendor give me an RMA# and return shipping label.

I followed all the instructions but it only increased my battery life by about 30 minutes. Then I saw the advice to download a driver updater utility. I used that and it found 9 drivers that were out of date despite Microsoft telling me everything was already up to date in Device Manager. After installing those updates I'm now sitting at 7 hours of battery life! Actually now it says 11 hours remaining from a full charge!

My discharge rate dropped significantly from -45wh an hour to 8-12wh an hour. My CPU finally enters a deeper C-State and now idles at 0.5-1w

Thank you!!

 

Hi everyone,

I have a problem with OneDrive. If I want to open a OneDrive folder that is synced to the windows explorer, it just freezes the whole windows explorer, and maybe after several minutes it actually opens the folder. If I want open the next OneDrive folder, everything freezes again.
If I want to open any normal forlder that is not part of one drive, everything works fine. Also there are no other issues with my SSD I can think of.
I reinstalled OneDrive and the issue seemed at first gone. Once I could access the folder through the OneDrive icon in the icon tray. That lasted unfortunatly only for that session. Now the problem is here again.

Why am I writing this in this thread?
In the past I never had this issue on my Laptop. After I accessed the advanced power settings on my SSD "AHCI Power Link Management HIPM / DIPM", that I discovered thx to this thread, somewhat after this point the issue first appeared.
Thats the only idea I have to what might be the source of the problem.
But: the issue persists, whether I change the AHCI Power Manangement from Lowest to Active or Not.

So maybe if someone has idea, or had similiar problem, any suggestions would be great.

 

Hi, I'm not aware of anything that might be causing this issue. If the issue is persisting after restoring the ACHI Power Management setting, it miht be something else that is causing this. In Task Manager, when opening up OneDrive, do you see any CPU, Network, or Disk activity?

 

Cant check right now, because I reinstalled OneDrive again and for now its working, like last time. I will see how long it holds.
I wasnt able to find something related to my issue in the internet aswell...

 

Thanks for the nice guide, I'd like to add a bit to it:

- In case, your PC is stuck at C2, C3 and 'powercfg -energy' report says there are USB devices without selective suspend - you can force them to use it:
either by VID/PID or simply:
find 'SelectiveSuspendEnabled' under HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Enum\HID and HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Enum\USB\ subkeys, changing 0 to 1 (or 00 to 01) forces it on.
After you forced Selective suspend, check Device Manager and enable "Windows can power down this device" for USB devices.
I can even force it for my mouse and it forces mouse off after about of 2 secs of inactivity.
I forced it on for problematic devices (touchpad in my case), and can achieve C6/C7 package state now with 0.3W package power. (I7-6700HQ)

- Standard Windows AHCI storage driver will not allow for C6/C7 unless you enable HIPM+DIPM, installing Intel RST driver enables C6/C7 regardless of DIPM (probably forces it silently).

- Regarding CPU parking - turning it on helps CPU to achieve higher Turbo Boost clocks for 1-2 core load. For Turbo Boost it is necessary for cores to be truly idle to reach peak performance for active cores. Without CPU parking, Windows stupidly tosses tasks across cores and average peak performance of 1-2 thread benchmark is reduced.

But since Skylake CPUs, you can enable Hardware Duty Cycling, which offloads managing inactive cores to CPU itself, just like SpeedStep for P-states. It is disabled by default, you need to unhide these to parameters and enable them:

Enables Hardware Duty Cycling for Intel(R) Dynamic Platform and Thermal Framework
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Power\PowerSettings\48df9d60-4f68-11dc-8314-0800200c9a66\e6902942-b0cf-41f2-9225-20839490eb8c

Specify whether the processor may use duty cycling
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Power\PowerSettings\54533251-82be-4824-96c1-47b60b740d00\4e4450b3-6179-4e91-b8f1-5bb9938f81a1

Having this on, without CPU parking, I can see, my CPU can achieve almost full 3480 (~3500) Mhz for single-core load.

 

Is anyone else running into an issue - especially on Lenovo notebooks - where the laptop is able to reach C6+ states while idle, and then after the next power source change (either DC-->AC or AC-->DC) it's no longer able to get past C3?

 

Does the dGPU have any effect on the package power? I have an i5 8250U+MX150 laptop and package power is in the 1.8-2.3W range. That too after tweaking a bunch of things mentioned in the guide.
The boot drive is a 512GB NVMe SSD (Sillicon Power P34A80). Also I have 20GB RAM installed. Intel Power Gadget shows DRAM power consumption varying from 0.8-0.95W usually. Does DRAM count as package power? I just can't get it any lower. CPU package doesn't enter C6 or lower. Its usually averages around 85-90% mostly in C3 state. I have 2 USB hub always connected but removing them doesn't seem to have any effect.

 

This is a known BIOS issue with some notebooks. I know that some Dell XPS have these issues. You may need to update yoru BIOS, if available.

When off, the dGPU does not consume any power. It does not affect CPU Package Power. Your issue is that your CPU Package is stuck in the C3 state - it should be at least C7 or higher. You will need to find the root of this problem. I used to have an 8250U, which idled at 0.3-0.4W. Can you verify that your dedicated GPU is off? HWInfo and monitoring programs of the likes should report 0.0C for the GPU temperature if it is off.

As far as I understand, yes DRAM power does contribute to CPU Package Power. Have you updated your NVMe driver? Does Silicon Power provide a custom NVMe driver? Can you veryf that DevSleep is working?

 

Yes the dGPU is not active. TS shows GPU temp. as - - °C.
AFAICT, SP doesn't provide any custom NVMe driver. The generic MS driver is installed.
About a year ago when the boot drive was still the HDD and only 8GB RAM was installed the package power was around 0.8W. Also the CPU Package would enter into C6/C7. When Windows 10 2004 is released I will try with a clean flash.
Edit: HIPM, DIPM etc are longer available in Advanced Power Settings. Maybe 1909 removed it. I don't think DevSleep is working.
Edit2: I ran one those driver updater tools. After updating Realtek Card Reader min. package power fell to 1.3W and averages around 1.6W. Only 2 updates are allowed daily for free users. Will check in a few days. Manually searching is tedious.
Edit3: I followed the steps outlined in the guide again and HIPM/DIPM settings are available once again.

 

I'm desperate for help I've searched the entire internet with no prevail. I own a predator g3-571 I'm trying to enhance the battery life since I get max one and a half-hour of battery tops, I followed this guide partially to the letter but got stucked with c state always at c3 and the advanced power settings (I can't access it completely), also I noticed that my package power is higher than what I've seen online it's always above 13w
Links to throttlestop settings and battery report
https://ibb.co/K99CjFc
https://ibb.co/NjVXVT7
https://ibb.co/chK2VmS
https://ibb.co/wh2Zdd3

 

Got it down to 0.6W. Fluctuates between 0.7-1.3W. CPU Package enters C6/7/8 though it could be better I feel. Updated Realtek Ethernet and a couple of ones related to Intel (SMBus and iHDCP). A few more still remain to be updated. Thanks for the guide.

 

Are you sure your laptop is idle and doing almost nothing when doing those tests? Your package power will be much higher when under load - that's natural.

 

Thanks for the screenshots. Unfortunately, it doesn't look like your laptop anywhere near idling at all. In Task Manager, you should see less than 0.5% of CPU utilization in the 'Processes tab'. Your C0% is also waayy too high to be idle.

Once again, you need to find out why your CPU is stuck at C3. Just looking around this thread alone (and what I wrote in the guide), other people have fixed it by updating the drivers.

I wrote this post not intending for it to give massive gains in 5 minutes. You simply can't expect to press a few buttons and magically double your own battery life. You need to do your own research for system-specific info and experiment here and there.

Great to hear. Post screenshots of your TS main window and C states. The i5-8250U should be able to idle at 0.4W idle package power, and at about 85+% in the C7/8 package state.

 

the C-state is always at C3 and I couldn't get it back with any method and the package power as I told you something is consuming power and I don't know what, that's my problem ill have to intense research for this but if you have any idea would be appreciated.

Update: Success after a little tinkering a little the problem was a service starting on booting the Nvidia Control Service managed to get PKG power from 15w to 2.3~3.0w, I think I can do more but that'll have to do for now power estimates skyrocketed from a merely 1 hour 30 mins to 3 hours 50 mins 100% battery. ill continue with the guide further to see the problem with my C state

 

Battery backup while browsing the web has seen the biggest gains. BatteryInfoView suggests 7-8 hours. Previous best was ~6 hours. And this was when the laptop was brand new, nearly 2 years ago. Surprisingly old Edge was only a tiny bit better than latest Chrome in this regard. Will try next with Chromium Edge. And I never thought merely scrolling web pages could trigger the iGPU so much. In some cases 60-70% usage.
Video playback with lower power renderers like EVR-CP, MPC etc was estimated to increase by 1-1.5 hrs. With MadVR which taxes the CPU and in particular the GPU a bit more the increase was more modest ~45 mins. But the stark difference in image quality makes the tradeoff worth it.
Idle life got a big boost as well at an estimated 16 hours. Its true that Intel notebooks are good at doing nothing!
This could translate into longer runtimes while downloading/updating stuff.
Benchmark scores(CB R15) are seemingly a little higher.
I only checked for a couple of hours or so. Will get back after more testing.

 

Did you disable all non-Microsoft services as mentioned in the guide? In my case this brought CPU usage down to 1% or so. Previously it would fluctuate between 2-5%.
Also update the drivers. This finally allowed CPU Package to enter C6/7/8.

 

yep I updated my post the problem was with Nvidia control service (standard not LS) when I disabled it the pwr package decrease to 2.5~3.5 w from a previous 15w the CPU still fluctuates but that's as a result of using chrome always and leaving it open but my main problem is still entering the C6/7/8 states I'm still stuck with C3 and cant go deeper. also my advance battery settings is not working for some reason so i don't know if i can go deeper than c3 on the Helios 300

 

I wanted to share my experience following this guide:

Laptop: Eluktronics W650KK1

The laptop was purchased April 2018 as a barebones kit. It has a desktop 1151 socket which now hold a locked i7-7700. There are two ram slots with 2400 ram running in dual channel. It has a 1050ti dGPU and uses optimus. The battery is rated at 62.16 Wh. I have an aftermarket 120 Hz display installed. There are two 500 gb SSDs installed.​

For the last two years I've used this laptop chained to a charger. Generally battery life was 2-3 hours. This was when the battery was new. Recently I have been working in more remote locations which didn't have an outlet available. Having invested about $700 into this laptop I didn't want to give up on it so (prior to finding this guide) I downloaded throttlestop and applied a -100.6mV undervolt and set the speed shift EPP to 220. This allowed me to squeeze out MAYBE 4 hours of run time before hitting critical battery level. The battery is now degraded to about 85% allowing a full charge capacity of ~ 53 Wh.

After following this guide I identified several issues, intel drivers were a big one. However no matter what I did I couldn't get the computer to drop more than 60% into c7. At that point c0 usage was between 1-1.5% and the cpu package was pulling 0.5-0.8 watts.

Pulled my hair out for a bit, ran an energy report which showed the realtek audio driver preventing sleep. I figured it was another driver issue as I had all non MS services disabled and nothing running. Updated drivers, rolled back drivers, couldn't get the error on the energy report to go away and sometimes the cpu wouldn't drop lower than c3.

Finally found a relevant post on MS support forum about realtek preventing sleep. Someone suggested checking Cortana, well after disabling cortana's microphone access we are looking really good.

-c0 now hovers around 0.5%
-cpu package power between 0.3 - 0.5 watts
- >80% in c8! instead of 0-40% in c7.





tl;dr: do steps in post #1, run energy report, chase errors. If realtek audio prevents sleep check cortana's access.

Edit: Update on some more issues I ran into.

Yesterday I was happily using my notebook on the couch, enjoying the great battery life, when I set it down to cook dinner. I came back to it about 45 minutes later and resumed light usage but noticed TS was reporting the cpu was no longer accessing the c7/8 state. Idle power draw was much higher than just an hour earlier. I'd seen this behavior a few times earlier this week in the early stages of optimization and ignored it because there were bigger fish to fry, and it was always corrected with a restart.

I figured some service must have started so I checked task scheduler to see if anything had activated and there were several "check for update" tasks that showed being recently run. I went through and enabled the options in these tasks to "run only on AC" and kill them if on battery. I don't want to completely gimp everything as this is my primary work machine and I don't want to have to manually check for updates so I think this will be a happy medium. Testing continues but I am still extremely pleased with the results.

Also, side note, I have throttle stop starting at log in with task scheduler, but I noticed it defaults to the green icon which indicated it isn't running. I have to open TS and click "turn on" to turn the icon red. Am I correct in assuming TS does nothing until I turn it on?

tl;dr#2: c-states no longer being used efficiently after some time, checked scheduled tasks for culprit, disabled them.

 

Okay, one more update, I've finally hit a point where I'm not sure what direction to go next.

Problem: When put into sleep and woken up notebook only access c3 and lower states (c0-c3).

Workaround: A reboot or hibernation cycle resets the behavior allowing notebook to access c8 state (c0-c8)

This is 100% reproducible.

Edit, this behavior is intermittent: I also noticed that when entering sleep the power plan changes to power saver, at this point if I click on the battery there is no power slider to change it out of power saver, I must do it from the mobility center or from the power plan page.

So from a usability POV I can control the power usage by monitoring this and cycling in and out of hibernation (if the computer is coming out of sleep) or I can disable sleep in lieu of hibernation, neither approach is ideal.
Can anyone offer any insight into what I can check to curb this? I'm leaning towards a driver but I have no idea which one could be causing this and I've updated all the drivers that were out of date.

Screenshots of c-states below, I didn't include task manager, etc. because there is no difference in c0% or c-state%. I'm puzzled.

 

A commonly heard of issue on many laptops, most noteably some of the XPS lineup. I remember reading about it somewhere on these forums, and apparantly it was due to a Thunderbolt re-initialisation (from sleep) failure, so something would loop. As far as I know, this issue is only solveable by a BIOS update. Or, IIRC somebody fixed it by plugging something in and out. Infuriating, I know. You're either stuck with keeping your lid closed or being forced to use hibernate. As a student, I couldn't live without sleep. You shouldn't have to monitor your computer for little things like that. You should be able to use it and expect it 100% of the time to perform as expected. I would say, return the computer (probably not possible with the age) or upgrade/downgrade the BIOS.

With regards to the power plan change, it's probably some program you have installed which is reverting a power plan. MS tried to remove power plans from the battery flyout, but you're smart enough to go to the Mobility Centre.

The dev of TS knows better than me @unclewebb but I think turning TS "On" and "Off" is irrelevant for most modern CPUs, for which TS is basically used for undervolting and modifying power limits, which are applied regardless if TS is on or off. I think only the settings on the left are applied. If you want TS to start in "On" mode, I would say to turn it on, and then press the save button.


https://www.reddit.com/r/Dell/comments/91313h/xps_15_9570_c_state_bug_after_s3_sleep_and_modern/
Have a read of the above thread.

 

https://www.reddit.com/r/Dell/comments/91313h/xps_15_9570_c_state_bug_after_s3_sleep_and_modern/

I'll check it out. There hasn't been a bios update released from the manufacturer that I can find but I'll give them a call tomorrow and see what they think. The device is VERY out of warranty, and I've replaced the screen myself, but from my experience the guys at eluktronics are pretty awesome about answering questions so I'll see what I can find out.

I poked around in event viewer after a sleep resume and hibernate resume but couldn't find anything out of the ordinary.

Thanks again!

Edit: I'll try a bios upgrade/downgrade based on what I can find, if that doesn't work it'll be a reinstallation of windows. I'll report back at some point but I have to travel for work tomorrow and I don't feel like performing either of those actions the night before I need my notebook all week.

 

FWIW, I'm having the exact same issue on Lenovo P1 2nd Gen. All drivers are up to date, BIOS is up to date... It's been months and I still haven't figured out how to force it to start using C8 after coming out of sleep mode or being disconnected from TB3.

 

Interesting, my notebook doesn't have a TB port, or any USB 3.1 ports. So if that is an issue specifically related to TB then we may be chasing different issues here. I've reached out to the manufacturer so hopefully they are kind to me even though my laptop is out of the warranty/support period. I couldn't help myself and I fresh installed windows from USB media last night and it made no difference.

I'm kind of thinking this issue may be more wide spread than we know of and that many people just don't notice/care because 8-12 hours on their ultrabooks is good enough. If I'm able to push 7-8 hours on this beast with a desktop 7700, then those little ultrabooks should be lasting much longer.

 

Hi all, left field question, but I'm hoping you can help me troubleshoot.

I'm running Windows 10 via Bootcamp on two models of Macbook Air (please don't throw stones at me, it's for work).

I've found this guide extremely useful in increasing the battery life of my 2019 Macbook Air, with an 8th gen Kaby Lake i5-8210Y processor. I'll post the screenshots below, but by using the guide I was able to drop the PKG Power to 0.5w and the C0% to 1.1 at idle.

I've just received the 2020 Macbook Air, with a 10th gen Ice Lake i3-1000NG4 processor. For the life of me, I cannot get the PKG Power below 2.3w, and the C0% fluctuates between 1.7 and 2.5 at idle.

I've tried:

- Updating the Intel drivers to non-Apple ones (as Apple's drivers are often out of date and crap)
- Disabling every device I could think of in Device Manager to see if it had any effect on PKG Power or C0% - no real luck there.
- Looking for rogue Services in msconfig and Task Manager - nothing jumps out.

At this stage I'm completely stumped - the battery on both is 49.9W-h. The 2019 is able to achieve 10 hours battery after following this guide, the 2020 barely 6.

If anybody could give me some advice on things to try and have a look at the screenshot below to see if anything jumps out at you, that'd be greatly appreciate. Thanks for reading and any leads you might be able to provide

 

As @Che0063 requested in another post some extra info if people post here, here's HWInfo and Task Manager - sorry I didn't post them initially, new in town.

All taken at idle with no apps in the foreground.

 

Okay, I think I'm making a little progress in tracking down what the issue is. Still stumped on how to fix it.

I went back and checked the C-states of the 2019 Macbook Air (the i5-8210Y powered machine) and the package load looks pretty good. For some reason, C7, C8, C9 and C10 aren't available, but the whole load is sitting at C6 on idle.

C States of the 2019 Macbook Air - C2 mostly free


Going back and checking back against the 2020 Macbook Air (with the 10th gen Ice Lake i3-1000), holy moly. Everything's sitting in C2, C6 doesn't even seem to be available.

2020 Macbook Air - C2 Overload



Playing around with it last night (trying to install new drivers, tweaking advanced power settings, etc), somehow or other I was able to shift 90% of the load into C3, but C6 still wasn't available.

Does anybody have any ideas of where to from here? Thanks again for reading!

 

No, thank you for posting so many screenshots and system info. It makes it so much easier for us to give you tips, and makes me more willing to help you.

I would start by checking if this issue also occurs in Mac OS - I'm not aware of any software that can measure C state residency in MacOS, but I think Intel Power Gadget should report Package Power Consumption. If the Package Power is significantly lower than the 3.3W in Windows, it would suggest that your C states are indeed working.

I note in your HWInfo screenshot that EIST is disabled - Maybe it doesn't matter because you have SpeedShift, but it's isnteresting because I have never seen that on a notebook.

In a few of your screenshots, your Package C state has a tiny bit of residency in C3 - that indicates that the state is working, but something is blocking it, leading me to think it is a driver issue.
Can you download RW-Everything > MSR > User and add in these registers?

#define MSR_PKG_C2_RESIDENCY 0x60D
#define MSR_PKG_C3_RESIDENCY 0x3F8
#define MSR_PKG_C6_RESIDENCY 0x3F9
#define MSR_PKG_C7_RESIDENCY 0x3FA
#define MSR_PKG_C8_RESIDENCY 0x630
#define MSR_PKG_C9_RESIDENCY 0x631
#define MSR_PKG_C10_RESIDENCY 0x632



Make sure to add them one by one (Type register in the box, press add, press done, then repeat. Otherwise your system might crash)

EDIT: Sorry, I wasn't clear. You only type in e.g. "0x60D" without the quotes.


The numbers are counters that count up from the moment you boot up your system, and TS calculates state residency based on that I think.

Also, can you try validating the C state interrupt limit in TS > C state > Valid, and increase both the Time Unit and Time Limit?

 

Not a worry at all - I just want to be able to provide as much info as possible, especially because this stuff (if/when solved) will be Googled by others in the same position. Nothing more frustrating than finding your exact problem and not getting an update on how it went.

I'm just heading out now but when I get home I'll try your suggestions (RW-Everything info and TS C-state valid box checking, plus time increases) and let you know my results. I've never used RW-Everything before so I might screw it up initially, but let me know if I've done something incorrectly and I'll rectify it.

 

Alrighty, so I downloaded RW-Everything and added all of the values you posted. Results here - I'm not sure what to make of them:



Now as far as checking Valid on the C-states and fiddling around with Time Unit and Time Limit, I tried incremental increases all the way up to maximum, with no change on C-state load distribution. The only change I was able to make was negative, in the 'Package C-State Request' field. If I change that to C6 (the C-state I actually want everything shifted to), C3 becomes completely unavailable and everything moves into C2. When I change it back to C-10, everything goes back to normal - a small percentage in C3, but the vast majority in C2.



Just trying a boot into MacOS to check the power consumption in Intel Power Gadget, see what's what. Back in a sec.

 

@Che0063 - just checked on MacOS with Intel Power Gadget. At idle the pkg power load in watts is around 0.8 - 1.1, when web browsing etc it's around 2.4. From what you've said, this suggests that the C-states are working and are somehow being blocked in Windows.

After rebooting into Windows, for whatever reason, most of the load is sitting in C3 (jumping between 60-80%), with C6 still showing 0.0 and not moving at all.

 

Thanks for the tip about MS Films and TV app. Although picture quality is clearly inferior to madVR, power efficiency is unbeatable. Package power mostly around 1.2-1.4W and 2-2.2W while playing 1080p/2160p H.265 10-bit medium bitrate respectively(idle package power fluctuates between 0.7-1.1W) . Battery runtime for 1080p video accordingly is ~8.5 hours. EVR-CP and MPC video renderers deliver a bit better picture quality but battery life tops out at a bit over 6 hours(with madVR its 5 hrs).
Battery life with Chromium Edge is nearly an hour-and-half longer compared to Google Chrome, at ~7.7 hours.
I had disabled Turbo Boost while on battery. Maybe an occasional lag while browsing but no issue in video playback. Brightness while web browsing was set much lower than while playing video. Outdoors battery life ofc would be significantly lower.
Right now battery life is the best it has ever been by far.
P.S. As an added bonus after updating the Realtek card reader its once again recognizing SD cards. It had stopped working sometime around the upgrade to 1909 build.

 

@UnspecifiedName How did you check for driver updates? In my case Windows didn't offer any new ones and ofc the manufacturer hadn't updated any in over a year. I had to download a driver updater program (DriverMax) for this. Mostly the Intel related ones and Realtek drivers fixed the CPU Package not entering C7/8.

 

I only mentioned Thunderbolt because that's how I connect my laptop to AC power (via TB dock), but I don't think TB itself is relevant. I've always suspected that the issue is triggered by changes in power state. In my experience deep C-states always work fine when Windows is fully booted while the laptop is on battery power, but if I start it on AC power and then switch to DC - that's when I typically run into the problem of not being able to reach deep C-states.

I think you're right about that. After all, what percentage of laptop users would even read a guide like this one, let alone play with ThrottleStop or other CPU state management software? I would bet it's less than 0.5% of all laptop users.

 

FWIW, here are screenshots that demonstrate the C-state issue I'm having.

If I boot Windows while on battery power, I'm able to reach C8 states and CPU package power consumption drops below 1W while idle (0.4W is lowest I've seen it go).

However, if I start Windows while on AC power and then switch my laptop to battery, I can't get past C3 and CPU package power stays above 2.5W while idle.

It's frustrating because it can make a 3-4 hour difference in battery life, and so anytime I unplug from AC power I pretty much have to also restart my laptop if I want to get optimal battery time.

 

Hey man, thanks for your help - based on your posts in this thread, I did check for updates. I used the free version of Driver Booster and that found 11 drivers in need of update, so I did that. I just tried DriverMax and it found 3 more (I did the two free ones I could do today), but still can't access C6.

Given DriverMax gave me a new Apple USB Controller driver, I ran a powercfg to check if the three USB Selective Suspend errors I've always gotten (on this laptop and the 2019 I'm comparing it to) were still there. They are, unfortunately - the device ID is Intel(R) USB 3.0 eXtensible Host Controller, and I've updated the drivers for that both from Intel and from Driver Booster.



Additionally, there are these errors (again, not new - they're on the 2019 Macbook Air that can access C6 as well as this 2020 Macbook Air that can't).



I just opened the 2019 Macbook Air and checked the C-states on Throttlestop to make sure I wasn't going crazy - interestingly, C3 isn't utilised at all. 10% of the load is sitting in C2, 90% is sitting in C6.

I still haven't even been able to enable C6 on this machine at all. Yesterday I sat down with Device Manager (again) and disabled everything with Throttlestop and the C-states window open to see if that made a difference, going so far as to disable the GPU, trackpad and keyboard - no change.

Given this is a 10th gen shiny Ice Lake CPU, my idle PKG power is a joke. I'm very determined to keep trying for C6, I just can't think of anything else to try.

 

It's frustrating because it can make a 3-4 hour difference in battery life, and so anytime I unplug from AC power I pretty much have to also restart my laptop if I want to get optimal battery time.​

Now that's interesting. Our issues are similar enough that I'm hoping if one of us figures it out it will work for the other's problem.

I found a bios update but no notes for what was changed and it made no difference for my issue. I'm at the point where I have basically a vanilla windows install and the problem persists.

Edit: couldn't figure out how to quote on mobile until it was too late

 

@UnspecifiedName Even before USB Selective Suspend was enabled CPU Package was entering C7/8. Doesn't seem to make much difference either way. I have a USB Mouse connected all the time and 2 USB hubs.
You mentioned checking for rogue services but did you disable ALL non-Microsoft ones? Prior to that C0 would be fluctuating between 2-5% IIRC.

 

Thanks for your reply, really appreciate it. In Msconfig I did yeah - disabled everything that wasn't Microsoft. I didn't go through services.msc and disabled everything because the page one guide now doesn't recommend that

 

If nothing works and if this isn't a fresh install of Windows maybe check after a Windows reinstall and manufacturer provided drivers. Apple provides the Windows drivers? I can't think of anything else if you have followed all the relevant steps outlined in the guide.
Edit: @UnspecifiedName Any update? I booted into Lubuntu and right out-of-the-box the CPU Package was entering C8 and C0 was a mere 0.3-0.4% compared to Windows where it would fluctuate between 0.6-0.8%. I used the Powertop utility by Intel to check.

 

Yeah I'm running out of ideas too. I did a clean install of Windows this morning and let Microsoft provide generic drivers for all the devices - still no C6. I then installed the Apple drivers which are provided, still no C6.

Whilst testing it against the 2019 Macbook Air (the 8th gen Y-series i5 model), I was able to figure out that plugging in any USB device (mouse, keyboard, etc) blocks C6 entirely, leading to the same idle PKG power and C state package loading as the 2020 Ice Lake model.

I thought I was on to something here, so went about disabling every USB device and controller in device manager - unfortunately, still no C6, and also no working keyboard or trackpad until restart

 

Just FYI, being able to solve all the issues shown by the powercfg command can further improve battery life. Yes, the C-states won't be affected much, but system power use will drop.

Unfortunately that's a much harder nut to crack. PCI Express suspend not working is one that's often a problem in laptops but it requires things like installing Linux to enable a command or something.

In Apple's case they simply do not care about the Windows side. Next year they'll abandon it for their own ARM-based processors anyway. For others they are probably too focused in getting the device out by some time period that they don't do thorough enough work.

 

Yeah I know what you mean. As I just wrote over in the Throttlestop thread, Apple has very little incentive to care about Windows as it's not part of their business model.

But for Windows-focused OEMs to make this stuff hard for consumers is just insane. Someone who buys a modern Windows notebook off the shelf should not have to play whack-a-mole with drivers, hardware and firmware to come up with the magic solution that actually grants them the advertised battery life/performance.

 

That would be really nice.

However Windows PC manufacturers already have low margins and sellers employ every tactic to improve it. There's a reason why they tend to bundle so much third party software into the machines, which makes the whole thing worse. It's so low that HP, the world's largest PC manufacturer, wanted to get out of the market altogether. Sure the CEO was pretty much nuts, but to think there was an incentive to move away from the main bread and butter is amazing.

I don't think anything can be done about it. It's just how it is.

At least on the Windows side we can fix most of the issues.

 

I prefer to work in the Windows environment as well, that's why I'm so disappointed that something so integral to the notebook's functionality (enabling the built in functionality of the damn processor) has been neglected.

For anyone googling this post in the future, I've tried every tip and suggestion in this thread (and the Throttlestop guide) to enable deeper C states for the processor package of the 2020 Macbook Air, and have not been successful.

If a future user is able to figure out a solution, or if Apple issues a driver update enabling C-states lower than C2 and C3, please post in this thread - it'll help a lot of people forced to use Macbooks achieve much better battery life when working in Windows.