Hi, my Framework 13 AMD 7840U laptop on Fedora lasts only ~ 2 days in sleep. This is the most annoying issue with an otherwise very good machine. I finally let Codex run some experiments, updated the BIOS to the latest version, checked all sleep-related settings, and tested sleep with all the expansion cards out, but to no avail: the minimum draw is over 1W, which gives me no more than 50 hours of sleep. The draw is probably higher due to the upkeep of 64 GB RAM, and nothing can be done about that. Or can it?
As a workaround, I finally setup hibernate and “hibernate after 3 hours of suspend”. I might also try hybrid sleep, where the system is prepared for hibernation, but remains suspended until the battery runs out.
Anyone else who is DDR-rich and has had similar experiences?
probably some 7840 sleep issue. im surprised you got it to sleep in the first place. its nigh impossible to do it on my framework sixteen, with the 7840hs (identical silicon)
Is it actually sleeping for the 50 hour period? Laptops under Linux do sometimes have a bad habit of being woken by a hardware event; the trick presumably is to work out what is doing that, and fixing it. Are any syslogs being written when it should be asleep? Is the laptop ever warm when you take it out of its bag?
Righto. Is your RAM in a 32GB pair? If so, I wonder if you could pull out one stick and re-run the experiment. I know it’s a pain to be without all your memory, but it’d only be temporary, and it would give you some new suspend performance data to see if the RAM is the cause.
(I am making the assumption that each stick has its own power draw, but I am not an EE).
I also recommend checking what is plugged into all the expansion ports. The 7040 series has limitations on which ports are best suitable for USB-A, display, etc. Plugging in a USB-A adaptor into the wrong port can result in high power draw.
I have 96GB of ram in my Framework 13 AMD 7840U with a 55Wh battery from the original 11th gen Intel FW13 laptop. Last night I let my machine sleep for 6 hours and 45 minutes and the battery drained 19% (went from 70% to 51%) which doing some math, that would be 35.5 hours for 100%.
Which, this part is just speculation, but if I scaled that 19% to 64GB (64*0.19/96) then I get 12.66% which comes out to 53 hours (((6*60+45)/(64*.19/96))/60) which is almost exactly your original figure. So it seems sleep duration scales inversely linearly with ram capacity which would imply that the majority of our power consumption during sleep is from the ram.