00:25 Venemo: agd5f: should 'sensors' work with a 7900 XTX?
00:26 Venemo: I mean, is it trustworthy?
19:16 agd5f: Venemo, yes, should be
19:18 Venemo: agd5f: is it normal for a 7900 XTX to consume around 50 W even if it's basically doing nothing?
19:18 agd5f: high refresh rate monitors?
19:19 Venemo: yes, it that the problem?
19:19 agd5f: depending on the timing
19:19 agd5f: https://gitlab.freedesktop.org/drm/amd/-/issues/2315
19:20 Venemo: with both set to 144 Hz: 50W, both set to 60 Hz: also 50 W, one at 60 Hz other at 144 Hz: 90 W
19:20 Venemo: only 1 monitor connected, at 60 Hz: 24 W
19:21 Venemo: all the above at 4K
19:22 Venemo: but resolution doesn't seem to matter, it still consumes around 20 W when I only have 1 monitor at 1080p 60 Hz
19:22 agd5f: it depends on the blanking period lengths. There is a certain latency to switch the mclk and it needs to happen during a blanking period to not affect the display.
19:25 bnieuwenhuizen: agd5f: can't we work around that with the infinity cache now?
19:26 Remco: I remember people fiddling with their modelines to get downclocking to work
19:26 agd5f: bnieuwenhuizen, not sure. question for someone on the display team who knows more about the requirements
19:30 agd5f: bnieuwenhuizen, I know we can turn off the mclk phys when the displays are idle, but only once everything lands in MALL.
19:30 agd5f: er rather desktop is idle
19:31 Venemo: well, my question really is, is this a bug or is it how it's supposed to be?
19:35 agd5f: likely how it's supposed to be depending on your monitors. but I would defer to the display team. There are some additional optimizations as per Aurabindo's comments in 2315 that may improve things.
19:39 Venemo: interesting. thanks for linking that, he actually explains it nicely
19:40 Venemo: would be interesting to see if it gets better if I set my monitors to 120 Hz but they don't seem to support 120 Hz
19:42 Venemo: but judging by how they don't consume less at 60, I don't think 120 would help