System details
iMac20,1 (Retina 5K, 27-inch, 2020), Radeon Pro 5300 (1002:7340, Apple 106b:0219), Omarchy 4.0.4, linux-t2 7.2.7 installed, linux-t2 7.2.11 pending in omarchy update
What's wrong?
The linux-t2 7.2.11 kernel that omarchy update pulls today carries two t2linux patches that hit this iMac:
- 6001 holds the GPU core clock at 300 MHz under load.
- 7003 (the 5K patch) brings the panel up as two 2560x2880 monitors,
eDP-1andDP-1, with the same description.
I confirmed both are in the 7.2.11 package's amdgpu.ko. I haven't booted 7.2.11. I built the same pair onto 7.2.7 and measured it under a full-screen OpenGL load:
| amdgpu | GPU clock under load | GPU busy |
|---|---|---|
| 7003 + 6001 (what 7.2.11 ships) | 300 MHz in 60 of 60 samples | 82% |
| 7003, no 6001 | avg 505 MHz, peak 1250 MHz | 56% |
Clock: the fix is dropping 6001, which is my PR on t2linux: t2linux/linux-t2-patches#73. The author of 7003 asked for exactly that setup (7003, no 6001) in t2linux/wiki#743. Until it merges, stock linux-t2 keeps this iMac at 300 MHz.
Two monitors: with 7003, Hyprland treats the two halves as two screens. Aquamarine only folds a tiled panel into one screen when one connector offers the full 5120x2880 mode, and neither half does here. My monitors.lua matches the panel by desc:, so both halves stacked in the same spot and the panel stopped taking clicks. Turning off DP-1 brings clicks back and leaves the right half dark:
hyprctl eval 'hl.monitor({ output = "DP-1", disabled = true })'I haven't tried the stock monitor rule with 7003. Kernel side: t2linux/wiki#766.
Related to #12197: 7003's author says it fixes that SMU timeout. Both of my 7003 boots came up clean (SMU is initialized successfully!).
Data and scripts: https://gist.github.com/Kb2uka/72148e8f7e313e861d6844d9d01b7ef0
Ask: a heads-up for iMac20,1 owners until t2linux/linux-t2-patches#73 lands, and a look at the two-tile case on the Hyprland side. I can test anything on this iMac, with boots and clocks measured.
Activity
0No activity yet.