Framework 16 Power Usage vs 13

I have both a Framework 16 7940HS and a Framework 13 7840U; The 13 I can get down to somewhere around 4 watts at idle when on battery, but the least I’ve been able to get the 16 down to, with an identical configuration as the 13 (no gpu), is around 8 watts. Everything I do, it seems that the 16 takes more power than the 13.

At first I figured, well the 7940HS is a more powerful processor than the 7840U, but then Phoronix published their recent Core Ultra 7 vs Zen 4 benchmarks, and while the 7840HS does perform slightly better than the 7840U, the differences IMO are negligible yet it still uses more power.

Which leads me to ask the question, why was the HS-series used over the U-series chips in the 16? It seems like just a waste of power. Is there a way to make the 7940HS/7840HS chips stay within the same power envelope as the 13, thus (hopefully) making the power usage similar?

P.S. I understand the tests were done with the 7840HS and I have the 7940HS, but in reality there’s not much difference between the two. Additionally I’m using Linux, not Windows, so I’m looking for Linux suggestions.

The lowest I’ve seen a FW 16 achieve myself at idle is 4.5W, but it’s a difficult state to hold for long. That was a measurement recorded on Gentoo.

During regular usage, my 16 sits around 6.5-7W with spikes higher.

My undestanding is the HS is AMD’s middle of the road chip. It’s not the flagship highest performances, it’s not the energy saving U series, but rather a balance of the two.

Could it be down to the display? The 16’s display will use a lot more power than the 13.

What optimizations do you have applied? There must be more than power-profiles-daemon and powertop autotune.

I tested both with the display around 25%, and with the 16 set to 60Hz. Additionally the keyboard backlights were off on both.

I find the keyboard backlights take about 3W.

That’s all I do. My measurements are taken at 0% brightness, and my cpu usage is at 0%. I’ll sit at the desktop for a minute and wait for it to settle down to a value before I make a measurement as well, gives time for things like the ssd to go into sleep I assume.

I’m only able to get down to ~8 watts with brightness down to its lowest setting, and the screen set to 60hz.

This is with:

  1. The regular expansion caddy rather than the dGPU
  2. 96GB of RAM
  3. No ssd or wifi installed
  4. Webcam off and microphone muted
  5. Screen set to the lowest setting
  6. Screen set to 60hz.
  7. powertop --auto-tune
  8. Sitting at the GNOME desktop with the terminal open and nothing else running.
  9. All expansion cards are USB-C cards, so no extra draw from anything there.
  10. Running off a Rocket Nano usb drive with a power monitor attached; it draws about 0.8 watts on idle which I subtracted from the battery figure.
  11. Updated arch install.

If I let the screen shut off, then I can get it down to 5 watts, but I tested again and I can get the 13 down to 4.5 watts even with the screen around 20% brightness.

Not only that but that 4 watts or so seems to follow me around whatever I do. Typing in LibreOffice? 4 watts or so extra on the 16. Watching a YouTube video? 4 watts or so extra on the 16. That’s somewhat in-line with what Phoronix found where the average power consumption of the 16 was about 8 wattts more than the 13 with extremely similar performance of the CPUs.

I like the 16 because of all the different options it has but given that the U and HS series processors perform basically the same with the U series using less power, I feel that Framework really should have gone with the U series rather than the HS series. As far as I know they’re the same chips just binned differently anyway. That is unless I’m missing some other advantage of the HS CPU…