Questions and experiences with Intel 1xxH / Meteor Lake

I need some advice and experiences about Intel 1xxH / Meteor Lake. I was wondering if it is more power efficient in idle, light task and video decode (watching media) compared to my amd 7640u, because the amd system seems to be pretty bad in this regard. Atm I comparing it to an Intel 11 gen system (x1 yoga) which sucks much less power in light tasks and also media consumption.

I also have the problem, that using the touchpad leads to a big power usage jump on the amd system (about 2-3 W monitored with upower or powertop). I also see about 2000 interrupts on the intel gen 11 x1 yoga, but this doesnt lead to that of big power usage. Maybe someone could test this scenario on a framework with Meteor Lake.

Also any reports about real world usage, battery usage in sleep and stability overall would also be highly appreciated :slight_smile:

Device:

  • framework 13, intel core ultra 125H, CPU with self applied amazon “pho PTM 7950” (i.e. likely not from Honeywell).
  • 2256 x 1504 60Hz 2.2K matte display,
  • samsung 990 pro 1TB,
  • Crucial 32GB (2x16GB) 5600MHz DDR5
  • Expansion cards: two usb c, 1 usb a (for intel I don’t think it matters what port this one is in, but it might matter if it’s present or not - I’ve never tested without it) and the mini sdcard reader (not in use).
  • Direct line of site connection to a 5 GHz wifi AP about 4 meters distant with the framework stock Intel WiFi 6E AX210.

OS: kubuntu 24.04 lts, kernel 7.0.0-14-generic, X11

I made about ~10 “casual” spot observations over a 5 min period from upower while running off battery (which was about ~50%) with a “balanced” power profile. upower dose not seem to update very frequently (even when I used watch -n 1) - say about every 5 seconds.

“Idle”

  • CPU (from btop): mostly at 400 MHz, occasionally bouncing up to 1.5 GHz.
  • System temp: ~40 °C, fan: 0 RPM
  • Firefox with 3 tabs open and few terminals running in the background but obviously not doing much.
  • Screen brightness at 15% (I’m indoors with plenty of natural light so this is very usable for me).
  • lowest observed power draw was ~2.5 W
  • typical observed power draw was ~5.5 W
  • a few ~9 W observations

FWIW, it seemed there was higher power draw (5.5 W and above) with trackpad usage. But then CPU usage was also higher when using the trackpad so interpret that any way you like. Also, I’m using keylightd for keyboard backlight control - keyboard backlight and power button led are on while typing or using the trackpad, off after a few seconds of idle.

Watching youtube, CPU 1.5-2 GHz, cpu temp ~ 58°C, fan at ~2300 RPM, I never saw power draw above 15W (most observations were ~14 W).

I suspect the biggest contributions to keeping my power draw low are

  1. using a PTM 7950 like material on the cpu, - like others on the forum, I’ve seen a 10-15 °C cpu temp. drop after switching (under load - no significant difference at “idle”)
  2. using linux &
  3. setting the screen brightness low.

HTH.

Thank you a lot for your detailed analysis.

5.5 W is analogous to my observation that just using the touchpad in idle, in general, raises the power usage of a Framework 13 device by around 2–3 W. Maybe it is a problem with the I2C system in general on all Framework 13 mainboards.

14 W is really high for an Intel system when watching YouTube. Can you check with intel_gpu_top if maybe the browser doesn’t use VA-API/video decode and the media engine altogether, and is using the CPU instead? I read that the media engine on this APU is directly on the LPE core chiplet outside of the GPU, so the GPU should go down to a low-power state even when watching media (maybe the power profile must be set to low-power/battery saver for that)

Watching more closely may not have helped:

intel-gpu-top: Intel Meteorlake (Gen12) @ /dev/dri/card1 -  104/ 707 MHz;  64% RC6
        358 irqs/s

         ENGINES     BUSY                                                 MI_SEMA MI_WAIT
       Render/3D   10.82% |████▉                                        |      0%      0%
         Blitter    0.00% |                                             |      0%      0%
           Video    1.23% |▋                                            |      0%      0%
    VideoEnhance    0.00% |                                             |      0%      0%
         Compute    0.00% |                                             |      0%      0%

   PID   Render/3D     Blitter       Video    VideoEnhance    Compute    NAME            
  2035 |▌          ||           ||           ||           ||           | kwin_x11        
  1519 |▍          ||           ||           ||           ||           | Xorg            
152707 |▍          ||           ||           ||           ||           | firefox         
152827 |           ||           ||▎          ||           ||           | RDD Process     
168706 |           ||           ||           ||           ||           | konsole         
1457

“Video” busy bounced between 0% to 6% with youtube running. However, I saw power draw at 16-17 W (I was also switching between terminals). My system temperature also was slightly warmer at 62 °C - i didn’t check the fan so see if it was running at a higher RPM.

Thanks again. I suppose you’re using the Snap version of Firefox on Kubuntu; I never could get VA-API to run reliably on that thing. The thing is, I only see an extra 4 W above idle on my X1 Yoga with an 11th Gen Intel chip when running 1080p YouTube videos.

Either way, if this is the normal power usage of that chip when using the media engine, then it would not be any better in that regard than my AMD 7640U.

1 Like

not using snap (it’s removed).

Can’t speak to what is normal, but I suspect you will find it within whatever the range for “typical” is. I have not taken any steps to optimize power use so you may be able to do better. Unfortunately, going to the next generation(s) is likely your best bet for efficiency.

@sn6526 So I had a new install with ubuntu on an intel system. It seems on intel hardware libva doesnt work out of the box. With vainfo you see no codec support at all. This changes after installing the intel-media-va-driver-non-free and the power usage decoding videos goes down alot.

hmm, i’ll give the non-free version a try - it will be interesting to see if it makes a difference. Thank you for the post.

Currently the (free?) intel-media-va-driver package is installed on my system and seems to be functional.

vainfo output
sudo vainfo
Place your right index finger on the fingerprint reader
error: XDG_RUNTIME_DIR is invalid or not set in the environment.
libva info: VA-API version 1.20.0
libva info: Trying to open /usr/lib/x86_64-linux-gnu/dri/iHD_drv_video.so
libva info: Found init function __vaDriverInit_1_20
libva info: va_openDriver() returns 0
vainfo: VA-API version: 1.20 (libva 2.12.0)
vainfo: Driver version: Intel iHD driver for Intel(R) Gen Graphics - 24.1.0 ()
vainfo: Supported profile and entrypoints
      VAProfileNone                   : VAEntrypointVideoProc
      VAProfileMPEG2Simple            : VAEntrypointVLD
      VAProfileMPEG2Main              : VAEntrypointVLD
      VAProfileH264Main               : VAEntrypointVLD
      VAProfileH264Main               : VAEntrypointEncSlice
      VAProfileH264High               : VAEntrypointVLD
      VAProfileH264High               : VAEntrypointEncSlice
      VAProfileJPEGBaseline           : VAEntrypointVLD
      VAProfileJPEGBaseline           : VAEntrypointEncPicture
      VAProfileH264ConstrainedBaseline: VAEntrypointVLD
      VAProfileH264ConstrainedBaseline: VAEntrypointEncSlice
      VAProfileVP8Version0_3          : VAEntrypointVLD
      VAProfileHEVCMain               : VAEntrypointVLD
      VAProfileHEVCMain               : VAEntrypointEncSlice
      VAProfileHEVCMain10             : VAEntrypointVLD
      VAProfileHEVCMain10             : VAEntrypointEncSlice
      VAProfileVP9Profile0            : VAEntrypointVLD
      VAProfileVP9Profile0            : VAEntrypointEncSlice
      VAProfileVP9Profile1            : VAEntrypointVLD
      VAProfileVP9Profile1            : VAEntrypointEncSlice
      VAProfileVP9Profile2            : VAEntrypointVLD
      VAProfileVP9Profile2            : VAEntrypointEncSlice
      VAProfileVP9Profile3            : VAEntrypointVLD
      VAProfileVP9Profile3            : VAEntrypointEncSlice
      VAProfileHEVCMain12             : VAEntrypointVLD
      VAProfileHEVCMain422_10         : VAEntrypointVLD
      VAProfileHEVCMain422_12         : VAEntrypointVLD
      VAProfileHEVCMain444            : VAEntrypointVLD
      VAProfileHEVCMain444            : VAEntrypointEncSlice
      VAProfileHEVCMain444_10         : VAEntrypointVLD
      VAProfileHEVCMain444_10         : VAEntrypointEncSlice
      VAProfileHEVCMain444_12         : VAEntrypointVLD
      VAProfileHEVCSccMain            : VAEntrypointVLD
      VAProfileHEVCSccMain            : VAEntrypointEncSlice
      VAProfileHEVCSccMain10          : VAEntrypointVLD
      VAProfileHEVCSccMain10          : VAEntrypointEncSlice
      VAProfileHEVCSccMain444         : VAEntrypointVLD
      VAProfileHEVCSccMain444         : VAEntrypointEncSlice
      VAProfileAV1Profile0            : VAEntrypointVLD
      VAProfileAV1Profile0            : VAEntrypointEncSlice
      VAProfileHEVCSccMain444_10      : VAEntrypointVLD
      VAProfileHEVCSccMain444_10      : VAEntrypointEncSlice

EDIT: i just tried the non-free version. I get the same vainfo output with it as above. It might be more power efficient - I’m getting 8-10 W with youtube video. However, my fan RPM is 0 atm. That might be because the igpu is running cooler (more efficient) or because my ambient temp is a bit cooler right now or both. Any event, hope this works better for you.

EDIT 1: I don’t think the non free driver makes any difference for me. With a slightly higher ambient temp (~2 °C) my fan runs when using youtube and power usage goes up. For my most recent observations, RPM was running close to ~3000 RPM and power usage was 15-20W. I think this also highlights the variability inherent in my “casual” observations. If your intel drivers where not functioning correctly before and now they are, then I expect you made a positive change so keep going with that. I don’t think your going to get the kind of improvement you want changing to intel core series 1 from your current amd system. Maybe the newer series 3 intel - maybe.