Today I could install the new Pro display into my old Framework 13 with AMD 7640U. Install was done in 10 minutes with the provided guide on the Framework site. Here are my initial thoughts on the panel:
The good:
120hz VRR works smoothly under debian with gnome. Everything feels really smooth.
Colors and Contrast seem to be a step up from the initial old panel (with an antiglare screen protector)
using 200% scale is much nicer then scaling fonts or everything to 125/150%
The bad:
Touch did only work after a bios reset (I did the bios update to 3.20 before installing the new panel); this really cost me some time to figure out
Much higher power draw of the new panel
idle power draw up from 4,4 W to 6,5 W (also seeing around +2 W in other workloads like watching a video on youtube)
The panel’s higher power consumption, combined with the AMD platform’s overall inefficiency during idle and video decoding, further reduced battery life.
Do you think it’s possible that the higher power consumption is partly or entirely caused by the panel simply being brighter? I think if you could adjust the brightness to actually match you would probably find the numbers to be much closer.
Actually no, the brightness was roughly set to the same amount and refreshrate was set 60hz.
Afterwards I tested VRR with 60hz and 120hz.
I think that they mentioned in the presentation of the panel, that the old panel was more optimized for lower power consumption and the new one more for pixel response, refresh and touch. But still is it a large amount more power used…
My problem with the original 2.8K display is that the minimal brightness is too bright when using in a dark room. How do you feel about the new one? Can it get dimmer than the original one?
Well, that’s like 60% more pixels, plus 2x refresh rate (although variable) compared to the old display.
Can you set it to use 60Hz and test the draw then?
Ofc things like touch also will add draw. Now, I’m curious.. If you don’t upgrade the BIOS and touch would not work.. would it still draw more power for touch?
I already wrote that I did the initial test at 60hz and roughly same brightness as the old display.
60hz vs 120hz does make difference about 0.5 W. With VRR enabled (30-120hz) I could see idle power draw of 5.5 W (so another 0.5 W down).
It seems comparable to the power draw difference between low power panels and normal panels in Thinkpads, and yes the panel has to drive more pixel I know.
I just installed my display (on the FW13 AMD 7640U) and the BIOS didn’t detect the touchscreen right away either. In my case the Embedded Controller / Power reset worked. That involved:
Shut the laptop down completely.
Disconnect USB-C power.
Remove the input cover.
Disconnect the main battery connector.
Hold the power button for ~15 seconds.
Reconnect the battery.
After that, the touchscreen got picked up and began working. So far I’m liking the new display and it’s a huge upgrade over the OG Matte display.
Very stupid question… but that original 4.4W is for the entire laptop running? i.e., the touchscreen is cutting down battery life by roughly a third? Or is that draw just for the screen specifically and normal laptop operation for something like an hour of web browsing is much higher?
Yes, the entire power draw at idle was measured with upower. So, at idle, it cuts the runtime significantly.
But when using the laptop for normal tasks, the high interrupts from the touchpad and other components push the power usage up to 8–12W, where an extra 2W isn’t that impactful.
I used to watch videos on YouTube with a power usage of 9–12W (that AMD GPU sucks a lot of power even with VA-API/hardware decoding). Now, I get about one hour less runtime in this scenario.
That would’ve been with the new 74Wh battery and the new Panther Lake chips, which (from what I’ve seen) have been praised for their power efficiency. It seems likely that the same setup would do even better with one of the old displays, especially at idle, but at 20h+ it probably wouldn’t be a difference with much practical use.
For anything else (either of the older batteries or older-generation mainboards), I could see the difference being more impactful, again especially at idle (though idle power draw isn’t the best measure as even having a website open will increase that draw a bit).
Still don’t think the screen itself is using anywhere near 2 extra W over the original one at similar brightness. So far we only got 2 data-points (I am aware of), one guy with the new mainboard and old screen getting relatively meh idle power and the one here with old mainboard and new screen getting pretty bad idle power.
I wonder if the touch part is just handled so badly on the amd side that it accounts for a fraction of the extra power, have you ever tried running with it disabled somehow?
When you do research, it is reasonable to expect higher power consumption:
Touchscreen do use 0.5-1 W over a normal panel with the same specs
There are more pixels to drive and a better overall pixel response, both of which result in more power being needed (All low-power IPS displays sacrifice pixel response time to save power; for example, ThinkPad low-power displays smear a lot but save 2–3 W compared to their normal counterparts)
My data points were generated by comparing identical workloads while monitoring battery draw with upower. Although these measurements are not highly sophisticated, real-world laptop usage shows a 15–25% reduction in battery life for me compared to the first gen display released by Framework.
Got mine installed and it looks incredible, 120Hz feels great and it’s noticably brighter than the other displays. However, mine seems to have a problem. It shows artifacts randomly and very dark backgrounds look sparkly (like you might expect for failing VRAM). I’m letting it run memtest86 and will continute to troubleshoot, but figured I should say something in case anyone else sees something similar.
Here’s a photo of one of the artifacts while doing the memtest86 scan. It happens in the UEFI/“BIOS” configuration and I’ve made no other hardware modifications so it’s definitely related to this new panel.
That should be isolatable by disabling the touchscreen
Afaik the low power screens in lenovos got that by using especially transparent lcds so they can use less backlight for the same brightness, the actual lcd turning pixels on and off part of a laptop display does not use a lot of power. Back when I blew up the backlight fuse on my t480s the power difference between displaying 1440p 60 to the backlightless display and displaying the same to an externally powered external display was barely measurable but well below 0.5W and that wasn’t a low power display.
That is a pretty reasonable experiment setup, I am not doubting your data I am just not sure the extra power use is from the display itself. As a 7840u owner I am more glancing at the igpu, that thing gets riled up really easily and seems pretty bad at calming down. Also possible some of the power saving features like psr or abm don’t enable right because the panel is not white-listed or is black.listed. There could be many reasons. With the previous touchpad interrupt issue I would also not be surprised if the touchscreen reading has some issue and is keeping stuff awake it should not.
Would be interresting if you could completely unload the touch driver and see what impact that has, completely disconnecing touch would be pretty hard so that is probably a reasonable middle ground.
