Please post your personal review of the Framework Laptop 12th gen here.
Please one post per review will make it clearer to read.
You way want to compare it to the rigg you where using before (maybe it make more sens for people to value the comments)
Received 31st August 2022
Updated 15th September
Received my DIY 1240p after 3 monthes of wait . using ubuntu mate 22
Slaped in the Samsung 980 and everything worked.
I had to unactivate secure boot to be able to boot kernel 5.18 (or 5.19)
Impressions compared to my plastic thinkpad edge 590:
CONS :
-
The screen reflectivity is VERY bad : it is very mirror like (not matte), especially if you use low backlite setting like I do. Inevitabely you need to increase the back light level to compensate even in a dark environement if you have a light behind. I need also to orientat the screen to not see my self sometimes. I am definitely not a lover of those mirror screens. I cannot get use to it spending all the time moving the screen to see something. A review comparing it to a macbook show it is even poorer quality than mac
-
WIFI : needs improvement in stability.
Had issues on a sketchy wifi at coworking and ended up using my phone s tethering … another guy next to me on a Macbook had no issue.
Another day I couldn t view a video on the internet for 30 min and had to reset the computer for it to resume.
The Wifi is even worth than on my E590 and way worse than a Macbook. I hope this comes from the linux kernel and not the hardware itself. -
Brightness buttons don t work OOB
-
CTRL + tarckpad 2 fingers slide up or down doesn t activate the zoom in chrome
PROS :
-
“Feels premium” like a mac or Thinkpad T
-
Stability : after using it 2 weeks it feels as good as the old thinkpad E590 under linux. It solide and Its my daily driver for work, I don t feel like my laptop will fail on me in the middle of a meeting with 10 persons ! Only issue arround for me is kernel 5.18 needing a patch for OBS studio (v4l2loopback/0.12.5).
-
13,5" with 3:2 : I was afraid it was too small coming from a 15"; after a day I can say it is (after scaling the screen) very usable as the height is the same as on 15".
-
The hinge works perfect , no wobble at all.
-
The frame itself is very sturdy feeiling; expect the screen that could be a little more compared to my E590
-
Keyboard feeling as good as thinkpad. I was afraid of the up down arrows, but it workes perfectely
-
Trackpad is better than E590
-
Fan cannot be heard on basic web surfing, on gaming it higher but acceptable (deactivated turboboost and ecores)
-
HIDPI on ubuntu mate is wether 100% = too small OR 200% = way too big => solution hidpi X2 + xrandr --output eDP-1 --auto --scale 1.4x1.4
-
Sound much better than on thinkpad edge 590. But not as powerfull as a friends macbook pro 15" 2018. This is a game changer since I can now really listen to music without starting my sound box. I can watch films without head sets !
-
can do fn lock to change behaviour of top row keys
-
Powerfull and snappy beast with many cores. But I ended up deactivating most of the Ecores since it is not yet optimized to be usefull in the linux kernel. Why would you need efficient cores to consum more power ? (loosing 1Hour work with 1Ă©th gen)
-
The module thing is actually more interesting than I though: if you have your Ă©nd screen on the right, place the HDMI on the right !
-
Played for 2 hours the game metal gear solide V in 1280X800 it is more than 30FPS in medium. The fan kick s in at a reasonable volume but cannot keep the laptop on my lap, it s hot
CPU with 8 Ecore deactivated from BIOS, is arround 30% , still lot of room there
-
Battery drain on suspend is comparable to E590 no complain (like 11th gen had).
Battery unplugged is better than my old computer something like 5-6 hours. Would have been nicer to have a 9 hours to be able to do a full day work
1 hour video = 15%. So 6hours video playback.
-
SSD samsung 980 1TO (was 1600 Mo/s on 2018 E590) the spec of the SSD are 3500 mo /s which is reached here.
-
Brightness keys don t work OOB but can be fixed. I note that I use brightness a lot because it is mirror like screen, so it needs to be much brighter to not see myself in a mirror : 12th gen not sending XF86MonBrightnessUp / Down - #28 by Iann_C
Ready to work :
-
Wifi with good connexion to the router + fibre:
Encoding 2 VP9 videos at the same time doesn t sloww donw the computer that much, it is still quite responsive even at 100% on the Pcores only
Woah! This feedback is incredible. Thank you for sharing
I wrote a full length review here:
https://anarc.at/hardware/laptop/framework-12th-gen/#software-setup
I have only used it for a few days now, updates will come as I find more things. I also made detailed power usage measurements.
Would you consider running additional power usage measurements in single user mode, no graphical DE?
Good to see the type A cards consume so much less on 12th vs 11th gen boards (10mW vs 400mW) however i noticed the HDMI card seems to have increased (again comparing to my test on 11th gen anyway) from 280mW to 500mW. Did you have a display attached when testing?
I was actually surprised to see only 10mW (close to noise) power usage for the USB-A cards, as previous tests I did (during suspend) showed they did use a significant amount of power. So I’m not absolutely sure of those results, in fact.
I’d be very careful at comparing benchmarks in absolute terms like this. My test bench might (and that’s actually a big “if”) be useful to analyze absolute power consumption on the devices on my machine. But as you can see from the raw results, I’ve got quite a bit of noise. Not enough to explain a jump from 280mW to 500mW.
I will point out, however, that the actual results are much less conclusive than what you seem to be concluding. What I actually concluded is this:
Here we add another 440mW on average, 380mW on minimum, and an extra 190mW of standard deviation.
Clearly the HDMI cards do use some power.
The headline is “~500mW” but that’s just rounded up (and badly too, I should probably have made that 400mW). And you’ll observe there’s a solid 190mW of standard deviation here, so 280mW would actually be within the margin of error of the 380mW minimum.
In other words, I’d be very surprised that the HDMI cards would have increased power usage between the two laptop. I suspect it’s exactly the same cards, and that the laptop doesn’t change much to that in terms of power usage. There’s just too much noise in those measurements to reach proper conclusions, I think. We’d need to test your 11th HDMI card on my 12th gen and vice versa to make sure there’s actually a difference in the card.
It’s definitely possible that there is a problem in the 12th gen board, as it is probably responsible for card power management too… but in that case I would assume we would also see differences in the other cards…
So yeah, take all those measurements with a grain of salt. It’s really hard to tell what’s going on here, it’s a really complex system with lots of stuff going on.
sure. here’s a quick test with my current setup unchanged (which is minimum brightness, 2xUSB-C 2xUSB-A cards), right after running systemctl rescue
:
Summary: min 2.44 avg 2.50 stdev 0.06 max 2.69
root@angela:~# powerstat
Running for 300.0 seconds (30 samples at 10.0 second intervals).
Power measurements will start in 180 seconds time.
Time User Nice Sys Idle IO Run Ctxt/s IRQ/s Fork Exec Exit Watts
09:50:02 0.0 0.0 0.0 99.9 0.1 1 94 50 2 0 2 2.53
09:50:12 0.0 0.0 0.0 100.0 0.0 1 108 58 0 0 0 2.55
09:50:22 0.0 0.0 0.1 99.8 0.1 1 386 131 187 110 188 2.61
09:50:32 0.0 0.0 0.0 100.0 0.0 1 74 38 0 0 0 2.55
09:50:42 0.0 0.0 0.0 100.0 0.0 1 67 33 0 0 0 2.51
09:50:52 0.0 0.0 0.0 100.0 0.0 1 70 35 0 0 0 2.48
09:51:02 0.0 0.2 0.1 99.6 0.1 1 165 108 38 33 38 2.69
09:51:12 0.0 0.0 0.0 100.0 0.0 1 70 36 0 0 0 2.64
09:51:22 0.0 0.0 0.0 99.9 0.1 1 115 64 0 0 12 2.61
09:51:32 0.0 0.0 0.0 100.0 0.0 1 68 34 0 0 0 2.51
09:51:42 0.0 0.0 0.0 100.0 0.0 1 69 40 0 0 7 2.47
09:51:52 0.0 0.0 0.0 100.0 0.0 1 68 35 0 0 0 2.44
09:52:02 0.0 0.0 0.0 99.9 0.1 1 80 40 0 0 0 2.47
09:52:12 0.0 0.0 0.0 100.0 0.0 1 62 33 0 0 0 2.47
09:52:22 0.0 0.0 0.0 100.0 0.0 1 84 43 0 0 0 2.51
09:52:32 0.0 0.0 0.0 100.0 0.0 1 80 40 0 0 0 2.48
09:52:42 0.0 0.0 0.0 100.0 0.0 1 65 32 0 0 0 2.45
09:52:52 0.0 0.0 0.0 100.0 0.0 1 86 44 0 0 0 2.50
09:53:02 0.0 0.0 0.0 100.0 0.0 1 64 32 0 0 0 2.45
09:53:12 0.0 0.0 0.0 99.9 0.1 1 65 34 0 0 0 2.51
09:53:22 0.0 0.0 0.0 100.0 0.0 1 73 41 0 0 1 2.50
09:53:32 0.0 0.0 0.0 100.0 0.0 1 59 31 0 0 0 2.47
09:53:42 0.0 0.0 0.0 100.0 0.0 1 64 33 0 0 0 2.44
09:53:52 0.0 0.0 0.0 100.0 0.0 1 81 43 0 0 1 2.47
09:54:02 0.0 0.0 0.0 100.0 0.0 1 75 39 0 0 0 2.47
09:54:12 0.0 0.0 0.0 100.0 0.0 1 109 59 0 0 0 2.48
09:54:22 0.0 0.0 0.0 99.9 0.1 1 107 56 0 0 0 2.50
09:54:32 0.0 0.0 0.0 100.0 0.0 1 107 58 0 0 0 2.47
09:54:42 0.0 0.0 0.0 100.0 0.0 1 75 39 0 0 0 2.45
09:54:52 0.0 0.0 0.0 100.0 0.0 1 95 49 0 0 0 2.45
-------- ----- ----- ----- ----- ----- ---- ------ ------ ---- ---- ---- ------
Average 0.0 0.0 0.0 100.0 0.0 1.0 92.8 47.0 7.6 4.8 8.3 2.50
GeoMean 0.0 0.0 0.0 100.0 0.0 1.0 84.6 43.9 0.0 0.0 0.0 2.50
StdDev 0.0 0.0 0.0 0.1 0.0 0.0 58.6 21.7 34.0 20.4 34.1 0.06
-------- ----- ----- ----- ----- ----- ---- ------ ------ ---- ---- ---- ------
Minimum 0.0 0.0 0.0 99.6 0.0 1.0 58.8 30.7 0.0 0.0 0.0 2.44
Maximum 0.0 0.2 0.1 100.0 0.1 1.0 385.5 131.4 187.0 110.0 188.0 2.69
-------- ----- ----- ----- ----- ----- ---- ------ ------ ---- ---- ---- ------
Summary:
System: 2.50 Watts on average with standard deviation 0.06
and here’s the same thing without anything plugged in, again monitor minimum brightness (not sure how to turn the thing off in console and i don’t want to bother):
Summary: min 2.42 avg 2.49 stdev 0.05 max 2.65
root@angela:~# powerstat
Running for 300.0 seconds (30 samples at 10.0 second intervals).
Power measurements will start in 180 seconds time.
Time User Nice Sys Idle IO Run Ctxt/s IRQ/s Fork Exec Exit Watts
09:58:30 0.0 0.0 0.0 100.0 0.0 1 70 37 0 0 0 2.48
09:58:40 0.0 0.0 0.0 99.9 0.1 1 68 35 0 0 0 2.48
09:58:50 0.0 0.0 0.0 99.9 0.0 1 92 47 0 0 0 2.47
09:59:00 0.0 0.0 0.0 100.0 0.0 1 82 44 0 0 0 2.53
09:59:10 0.0 0.0 0.0 100.0 0.0 1 63 31 0 0 0 2.47
09:59:20 0.0 0.0 0.0 100.0 0.0 1 62 32 0 0 0 2.48
09:59:30 0.0 0.0 0.0 100.0 0.0 1 62 31 0 0 0 2.45
09:59:40 0.0 0.0 0.0 100.0 0.0 1 58 29 0 0 0 2.42
09:59:50 0.0 0.0 0.0 99.9 0.1 1 90 49 0 0 0 2.45
10:00:00 0.0 0.0 0.0 100.0 0.0 1 82 47 0 0 0 2.46
10:00:10 0.0 0.0 0.0 99.9 0.1 1 107 60 0 0 0 2.51
10:00:20 0.0 0.0 0.0 99.9 0.1 1 79 39 1 0 1 2.50
10:00:30 0.0 0.0 0.0 100.0 0.0 1 71 36 0 0 0 2.46
10:00:40 0.0 0.0 0.0 100.0 0.0 1 64 31 0 0 0 2.43
10:00:50 0.0 0.0 0.0 99.9 0.0 1 84 50 1 1 1 2.53
10:01:00 0.0 0.1 0.0 99.8 0.1 1 132 71 1 0 0 2.61
10:01:10 0.0 0.0 0.0 100.0 0.0 1 100 50 0 0 0 2.56
10:01:20 0.0 0.0 0.0 100.0 0.0 1 142 36 0 0 0 2.53
10:01:30 0.0 0.0 0.0 100.0 0.0 1 93 45 0 0 0 2.48
10:01:40 0.0 0.0 0.0 100.0 0.0 1 149 43 0 0 0 2.43
10:01:50 0.0 0.0 0.0 100.0 0.0 1 155 45 0 0 0 2.48
10:02:00 0.0 0.0 0.0 99.9 0.1 1 185 61 0 0 0 2.53
10:02:10 0.0 0.0 0.0 100.0 0.0 1 65 35 0 0 0 2.50
10:02:20 0.0 0.0 0.0 100.0 0.0 1 80 42 0 0 0 2.48
10:02:30 0.0 0.0 0.0 100.0 0.0 1 104 51 0 0 0 2.46
10:02:40 0.0 0.0 0.0 100.0 0.0 1 145 42 0 0 0 2.46
10:02:50 0.0 0.0 0.0 100.0 0.0 1 176 55 0 0 0 2.48
10:03:00 0.0 0.0 0.0 99.9 0.1 1 203 72 0 0 0 2.48
10:03:10 0.0 0.2 0.1 99.7 0.1 1 276 134 38 33 38 2.65
10:03:20 0.0 0.0 0.0 100.0 0.0 1 97 49 0 0 0 2.56
-------- ----- ----- ----- ----- ----- ---- ------ ------ ---- ---- ---- ------
Average 0.0 0.0 0.0 100.0 0.0 1.0 107.9 47.6 1.4 1.1 1.3 2.49
GeoMean 0.0 0.0 0.0 100.0 0.0 1.0 98.7 45.0 0.0 0.0 0.0 2.49
StdDev 0.0 0.0 0.0 0.1 0.0 0.0 50.4 19.5 6.8 5.9 6.8 0.05
-------- ----- ----- ----- ----- ----- ---- ------ ------ ---- ---- ---- ------
Minimum 0.0 0.0 0.0 99.7 0.0 1.0 57.6 29.1 0.0 0.0 0.0 2.42
Maximum 0.0 0.2 0.1 100.0 0.1 1.0 275.8 134.5 38.0 33.0 38.0 2.65
-------- ----- ----- ----- ----- ----- ---- ------ ------ ---- ---- ---- ------
Summary:
System: 2.49 Watts on average with standard deviation 0.05
Note: Power calculated from battery capacity drain, may not be accurate.
One problem with both of those tests is that I still had processes running as my user; turns out systemctl rescue
doesn’t actually kill everything. After killing everything, I get a lower power usage, again single-user mode, everything killed, no expansion cards, minimum brightness:
Summary: min 2.00 average 2.02 stdev 0.02 max 2.06
root@angela:~# powerstat
Running for 300.0 seconds (30 samples at 10.0 second intervals).
Power measurements will start in 180 seconds time.
Time User Nice Sys Idle IO Run Ctxt/s IRQ/s Fork Exec Exit Watts
11:02:06 0.0 0.0 0.0 100.0 0.0 1 40 19 0 0 0 2.03
11:02:16 0.0 0.0 0.0 100.0 0.0 1 39 20 0 0 0 2.02
11:02:26 0.0 0.0 0.0 100.0 0.0 1 44 24 0 0 0 2.02
11:02:36 0.0 0.0 0.0 100.0 0.0 1 36 18 0 0 0 2.00
11:02:46 0.0 0.0 0.0 100.0 0.0 1 42 24 0 0 0 2.02
11:02:56 0.0 0.0 0.0 100.0 0.0 1 46 28 0 0 5 2.06
11:03:06 0.0 0.0 0.0 100.0 0.0 2 44 23 0 0 4 2.03
11:03:16 0.0 0.0 0.0 100.0 0.0 1 43 26 0 0 2 2.03
11:03:26 0.0 0.0 0.0 100.0 0.0 1 45 24 0 0 0 2.05
11:03:36 0.0 0.0 0.0 100.0 0.0 1 36 18 0 0 0 2.02
11:03:46 0.0 0.0 0.0 100.0 0.0 1 44 23 0 0 0 2.02
11:03:56 0.0 0.0 0.0 100.0 0.0 1 41 22 1 0 0 2.02
11:04:06 0.0 0.0 0.0 100.0 0.0 1 39 19 0 0 0 2.00
11:04:16 0.0 0.0 0.0 100.0 0.0 1 35 18 0 0 0 2.00
11:04:26 0.0 0.0 0.0 100.0 0.0 1 52 43 0 0 15 2.02
11:04:36 0.0 0.0 0.0 100.0 0.0 1 38 22 0 0 0 2.00
11:04:46 0.0 0.0 0.0 100.0 0.0 1 39 20 0 0 0 2.00
11:04:56 0.0 0.0 0.0 100.0 0.0 1 39 21 0 0 0 2.02
11:05:06 0.0 0.0 0.0 100.0 0.0 1 39 21 0 0 0 2.02
11:05:16 0.0 0.0 0.0 100.0 0.0 1 36 18 0 0 0 2.02
11:05:26 0.0 0.0 0.0 100.0 0.0 1 47 26 0 0 0 2.03
11:05:36 0.0 0.0 0.0 100.0 0.0 1 36 18 0 0 0 2.02
11:05:46 0.0 0.0 0.0 100.0 0.0 1 39 20 0 0 0 2.00
11:05:56 0.0 0.0 0.0 100.0 0.0 1 36 19 0 0 0 2.00
11:06:06 0.0 0.0 0.0 100.0 0.0 1 39 20 0 0 0 2.00
11:06:16 0.0 0.0 0.0 100.0 0.0 1 36 17 0 0 0 2.00
11:06:26 0.0 0.0 0.0 100.0 0.0 1 41 21 0 0 0 2.00
11:06:36 0.0 0.0 0.0 100.0 0.0 1 36 17 0 0 0 2.00
11:06:46 0.0 0.0 0.0 100.0 0.0 1 40 21 0 0 0 2.00
11:06:56 0.0 0.0 0.0 100.0 0.0 1 38 19 0 0 0 2.00
Average 0.0 0.0 0.0 100.0 0.0 1.0 40.1 21.7 0.0 0.0 0.9 2.02
GeoMean 0.0 0.0 0.0 100.0 0.0 1.0 40.0 21.3 0.0 0.0 0.0 2.01
StdDev 0.0 0.0 0.0 0.0 0.0 0.2 3.9 4.8 0.2 0.0 2.9 0.02
Minimum 0.0 0.0 0.0 100.0 0.0 1.0 35.4 17.4 0.0 0.0 0.0 2.00
Maximum 0.0 0.0 0.0 100.0 0.0 2.0 51.5 43.2 1.0 0.0 15.0 2.06
Summary:
System: 2.02 Watts on average with standard deviation 0.02
And here’s the same, in multi-user mode:
Summary: min 2.33 average 2.49 stdev 0.05 max 2.65
anarcat@angela:~$ sudo powerstat
Running for 300.0 seconds (30 samples at 10.0 second intervals).
Power measurements will start in 180 seconds time.
Time User Nice Sys Idle IO Run Ctxt/s IRQ/s Fork Exec Exit Watts
10:36:22 0.2 0.2 0.2 99.2 0.2 1 2053 612 83 19 80 2.96
10:36:32 0.1 0.2 0.2 99.4 0.1 1 2108 632 81 16 87 3.41
10:36:42 0.2 0.3 0.3 99.2 0.1 1 2229 692 88 23 89 3.68
10:36:52 0.2 0.5 0.3 98.9 0.1 1 2414 763 115 46 119 4.08
10:37:02 0.2 0.0 0.1 99.7 0.0 1 1554 343 79 16 80 3.37
10:37:12 0.1 0.0 0.1 99.8 0.0 1 1283 282 76 16 76 2.92
10:37:22 0.2 0.0 0.1 99.7 0.0 1 1601 414 78 16 86 2.69
10:37:32 0.1 0.0 0.1 99.8 0.0 1 1224 290 76 16 77 2.49
10:37:42 0.2 0.0 0.1 99.6 0.1 1 1523 361 86 23 86 2.52
10:37:52 0.2 0.0 0.1 99.6 0.1 1 1562 345 79 16 80 2.55
10:38:02 0.1 0.0 0.1 99.7 0.1 1 1428 333 79 16 78 2.51
10:38:12 0.2 0.0 0.1 99.7 0.1 1 1361 294 78 16 79 2.42
10:38:22 0.2 0.0 0.1 99.7 0.0 1 1392 334 77 16 77 2.35
10:38:32 0.2 0.0 0.1 99.6 0.0 1 1480 294 80 16 79 2.39
10:38:42 0.2 0.0 0.1 99.7 0.0 1 1451 324 87 23 87 2.34
10:38:52 0.2 0.0 0.1 99.7 0.0 1 1590 348 78 16 78 2.49
10:39:02 0.2 0.0 0.1 99.7 0.1 1 1429 298 79 16 79 2.44
10:39:12 0.2 0.0 0.1 99.7 0.1 1 1391 302 78 16 78 2.42
10:39:22 0.2 0.0 0.1 99.7 0.0 1 1307 315 75 16 75 2.33
10:39:32 0.1 0.0 0.1 99.8 0.0 1 1314 301 75 16 74 2.33
10:39:42 0.2 0.0 0.1 99.6 0.1 1 1388 326 86 23 85 2.39
10:39:52 0.2 0.0 0.1 99.7 0.0 1 1302 311 75 16 75 2.42
10:40:02 0.2 0.0 0.2 99.5 0.1 1 1662 364 157 56 156 2.56
10:40:12 0.2 0.0 0.1 99.6 0.1 1 1441 316 77 16 77 2.52
10:40:22 0.2 0.0 0.1 99.7 0.0 1 1455 358 78 16 78 2.57
10:40:32 0.1 0.0 0.1 99.7 0.0 1 1355 299 77 16 77 2.47
10:40:42 0.1 0.0 0.1 99.7 0.0 1 1377 322 85 23 85 2.43
10:40:52 0.2 0.0 0.1 99.7 0.1 1 1404 302 78 16 78 2.39
10:41:02 0.2 0.0 0.1 99.6 0.1 1 1513 327 79 16 79 2.48
10:41:12 0.2 0.0 0.1 99.7 0.0 1 1534 355 78 16 78 2.54
-------- ----- ----- ----- ----- ----- ---- ------ ------ ---- ---- ---- ------
Average 0.2 0.0 0.1 99.6 0.1 1.0 1537.5 371.9 83.2 19.6 83.7 2.65
GeoMean 0.2 0.0 0.1 99.6 0.0 1.0 1515.5 357.1 82.2 18.5 82.7 2.62
StdDev 0.0 0.1 0.1 0.2 0.0 0.0 283.2 123.8 15.5 8.9 15.6 0.42
-------- ----- ----- ----- ----- ----- ---- ------ ------ ---- ---- ---- ------
Minimum 0.1 0.0 0.1 98.9 0.0 1.0 1224.3 282.1 75.0 16.0 74.0 2.33
Maximum 0.2 0.5 0.3 99.8 0.2 1.0 2414.2 763.0 157.0 56.0 156.0 4.08
-------- ----- ----- ----- ----- ----- ---- ------ ------ ---- ---- ---- ------
Summary:
System: 2.65 Watts on average with standard deviation 0.42
So this tells us a few interesting things:
- single-user mode uses less CPU than normal multi-user mode. I haven’t explicitly checked, but I bet this is my status bar, taking anywhere between 0.1 and 0.2% CPU usage just sitting there doing idle updates, and there are other things waking up using power (docker, syncthing being the worst offenders according to powertop)
- multi-user mode does use more CPU and more power, but not by much, about half a watt, not bad for something running a full GUI! if you actually compare the first two tests that were “wrong” in the sense that they still had multi-user processes running, you actually see it’s using more power than the multi-user mode (min=2.42 vs min=2.33), which is confusing but I think could be explained by the display compositor doing less update to the GPU than the linux console
- we see the USB cards really have minimal power usage. the Min without the cards is 2.42W instead of 2.44W but that’s inside the standard deviation (0.06W)… it does match with the 20mW we’d expect the USB-A cards to take, but still, it’s not enough for me to unplug and plug those cards all the time
I think that’s about all I can extract from this. Curious to hear if others see anything else.
I will not run the whole test suite again in single user, in case that’s what you’re asking. But I’m happy to make other such small measurements.
Let me know!
Thank you for doing that.
I was hoping to see a tighter set of readings / lower standard deviation…expected that there’s less background processes running / firing, and that should have resulted in a flatter set of power measurements / less power consumption fluctuation.
me too! i’m actually quite surprised by this. you’d think with all the garbage I have running on my desktop (which includes stuff like docker and syncthing, which, acording to powertop, wake up all the time), you’d see a massive difference, but the numbers don’t lie…
maybe the linux kernel console really is bad at power saving? i bet this wasn’t as well optimized as the normal “xorg / DRI” code path…
ah. ahem. turns out systemctl rescue
doesn’t do what I think it does and it actually does leave a few processes lying around, including emacs and other stuff started in my systemd --user
session. let me rerun those tests.
update: i’m seeing 2.00 watt outputs here, clearly a change once i pkill -u anarcat
. stay tuned
update2: i updated the original post with normal metrics. single user mode is indeed much less power hungry, about half a watt. but the GUI is not the worst offender!
Nice! So with that, idle of 2W, the Framework Laptop can idle for a full 24 hours if needed. The stdev of 0.02 is really good to see as well.
Yeah well, that’s with everything turned off. You will rarely see such power draw in practice, as you will have things turned on! If you’re looking for 10 hour+ battery life, I think you want auto-suspend or things of that nature, and then you need to tweak that as well, to make sure (say) expansion cards are not drawing power on suspend (because yes, they do!).
@anarcat fair points regarding drawing conclusions and comparing data, the only way I was able to improved reliability/consistency of my data was to extend the duration of the test to to 1 hr or more and repeat them several times and to have the main battery disconnected.
It is a complex system topic and probably one Framework would have the answers to but here we are as users having to figure out what is likely known by the manufacturer.
I also tested power drain w/ USB-A modules plugged and unplugged on 12th gen and found power draw was neglible/within MoE.
Oh, I’ve posted this before elsewhere, but I’ve continued to do some updates (mostly on the software section). Here’s my personal (Arch Linux focused) review: 2022 Framework Laptop DIY Edition 12th Gen Intel Batch 1 · lhl/linuxlaptops Wiki · GitHub
Yeah, I’m not sure anymore. I’m getting weird fluctuations in the standby battery usage… I did manage to go all the way down to 14mA drain, somehow, but now during the last suspend, the drain was all the way up to 76mA.
According to battery-stats it went from 99% down to 68% in less than a day!
That’s with 2xUSB-C and 2xUSB-a so I’m wondering if that might not be the trouble here… It’s really hard to figure out what’s going on during suspend…
Ah, it’s a good point that my tests w/ the USB-A module was difference while active. It could still be doing something fishy with suspend states. I think overall though that even w/ tweaks, currently Framework’s suspend power consumption is so bad that suspend-then-hibernate is the only real option if you’re not going to be plugged in all the time.
Well that’s the thing. If you can reach down to 14mA, then everything is peachy. And I did manage to get there, so it feels like it’s something that’s actually attainable… It’s just that I feel it slipping through my hands everytime I return from sleep.
@anarcat Thank for the incredibly detailed review, et merci d’avoir créé un site si intéressant, camarade!
I have owned my framework for only 3 weeks and I have to say that I love it.