13:18 <wens> dsimic: got another PSU (140w spread over 4 type-c and 2 type-a ports with a power meter). 5v is "boosted" to around 5.15v and seems to let me get away with building the kernel package on all cores on the NVMe

13:18 <wens> haven't hit 3A yet

16:40 <dsimic> wens: oh, going slightly over 5 V at the PSU output is the rather standard approach for overcoming the voltage drops at USB cables, which is actually quite inefficient, but perhaps also cheaper than having more copper in the cables, so the voltage drop on them is lower

16:41 <dsimic> I believe Raspberry Pi was the first to officially go slightly above 5 V with its official USB PSUs

16:43 <dsimic> it could be that your old PSU was actually fine, and just a USB cable with more copper in it (i.e. lower voltage drop) could've solved the instability issues... maybe you could get a 5 A USB-C cable, which must have more copper, at some point, and test it with your old PSU, just for the testing completeness' sake

16:45 <dsimic> also, the built-in power meters are usually too slow to catch power consumption bursts, so you may actually be over 3 A already in short bursts, but the built-in display may be simply too slow to catch and display those

16:47 <dsimic> which goes along with the way NVMe SSDs work during a kernel build, i.e. there are only short bursts of high I/O and the resulting high power consumption, which add up to the pretty much consistent power consumption of the fully loaded CPU cores

16:47 <dsimic> s/add up/adds up/

16:49 <dsimic> with all that in mind, I don't think your old 3 A PSU would be capable of powering the board reliably even with the above-proposed 5 A USB-C cable, but it would still be interesting to test such a scenario :)

16:52 <dsimic> are there any specs of your new PSU, i.e. is it kown precisely what's it actually capable of provoding at its USB-C ports?

16:53 <dsimic> see e.g. https://pine64.com/product/pinepower-v2-165w-desktop-power-supply-eu-version/ for an example of such specs

17:12 <dsimic> maybe your new PSU actually can't go over 3 A at 5 V, at its USB-C ports?

17:17 <wens> looks so. it can do 3A @ 5/9/12/15V or 5A @ 20V

17:18 <dsimic> then it's all about "living on the edge" of the voltage drop present at the cable and the slightly higher voltage at the PSU output

17:26 <dsimic> which also means that your old PSU might work just fine with the above-proposed 5 A USB-C cable

17:27 <wens> I'm also slightly concerned since this PSU is somewhat too warm for just outputting 15w

17:27 <wens> I guess it's because it was quite cheap

17:29 <dsimic> huh, that isn't a good sign, because... what would happen with 100 W+?

17:30 <wens> indeed

17:30 <wens> the box says 140w max

17:32 <dsimic> old PSU designs aren't unknown for getting really warm even with very low power draws, although those are 15+ year old designs

17:32 <dsimic> which shouldn't apply in this case

17:34 <dsimic> e.g. old Nokia chargers were notorious in that regard