I have the exact same problem: Seiki 4K 39" monitor, AMD A8-5545M with Radeon HD 8510G APU, Catalyst version 14.20-140522a-172181E
The first problem seems to be bad EDID/DDC data from the monitor causing Xorg to cough up:
EDID timing clock 297.00 exceeds claimed max 115MHz, fixing
which is a bit of a hack in handle_edid_quirks() in interpret_edid.c
So I tried to include setting "IgnoreEDID true" in the xorg.conf, but I found this ing the Xorg.0.log:
(WW) fglrx(0): Option "IgnoreEDID" is not used
Super! This seems to be an Xorg thing to strip out functionality.
And then I created a custom Modeline for 3840x2160 @ 30 Hz:
Modeline "3840x2160_30" 297.0 3840 4016 4104 4400 2160 2168 2178 2250 -hsync +vsync
This modeline is based on the actual EDID data reported by the monitor:
(II) fglrx(0): Supported detailed timing:
(II) fglrx(0): clock: 297.0 MHz Image Size: 854 x 481 mm
(II) fglrx(0): h_active: 3840 h_sync: 4016 h_sync_end 4104 h_blank_end 4400 h_border: 0
(II) fglrx(0): v_active: 2160 v_sync: 2168 v_sync_end 2178 v_blanking: 2250 v_border: 0
And here is where things just get fantastic! In the Xorg.0.log, I find:
(EE) fglrx(0): Add Customize Mode failed: 7
(II) fglrx(0): Not using mode "3840x2160_30" (unknown reason)
Can it get any better than this? Running strings on glrx_drv.so finds the offending strings in AMD's driver.
Add Customize Mode failed: %x
I have searched the web for information on this problem and all I have found were (a) people running into a 330 MHz pixel clock limitation (not a problem at 297 MHz) and (b) crickets.
Does anyone know if it is possible to correct the maximum pixel clock or why Xorg gets it wrong? And is there a problem or upper limit on the driver pixel clock, say for this hardware? And what is failure code "7"? Even the driver doesn't seem to know.