Disable dithering - AMD

JTL

KM Can you help me find out which bytes to modify? I could then try to load the modified EDID as a kernel parameter and test if the changes fixed the problem. This is all assuming that xrandr displays correct EDID data, and that passed EDID kernel parameters are actually respected by the driver.

I have reason to believe in some situations (not sure about amdgpu driver) that custom EDID binaries are only loaded after initialization of the display output, which may use the existing EDID received from the sink device.

JTL

jordan FRC = monitor side dithering

JTL I have reason to believe in some situations (not sure about amdgpu driver) that custom EDID binaries are only loaded after initialization of the display output, which may use the existing EDID received from the sink device.

Also the fact dithering can be enabled in the VBIOS of the card, meaning that it occurs before an OS is even relevant.

smilem

Tried this on a AMD Radeon HD 3850 AGP GPU driver 8.970.100.0 (Catalyst 14.4)

Monitor 6bit SyncMaster 205BW - DVI - CCFL
Polarization 45°
Gray RGB 195 195 195
Windows Server 2000 R2 SP2

The dithering is not disabled. visible with slow motion camera. this GPU - monitor combo dithers also in BIOS.

jordan

smilem Wouldnt that mean its FRC since it also happens in bios?

jordan

JTL I know but if its happening across multiple systems maybe its the monitor

JTL

jordan GPU dithering+FRC is an "either/or" situation. Doesn't make this easier, of course.

smilem

I tested BL2410 from benq, should be 8bit by display specifications web site https://www.displayspecifications.com/en/model/18e1739

However using amd registry dithering disabling for dvi, i can confirm it still not working, strange but in presentation mode the bl2410 monitor dithers from the darker shades, when all other modes from white.

Seems display specifications cant be trusted, 8bit panel cant dither by itself and it was tested on multiple cards and especially Radeon HD 3850 AGP

Changing display modes changes dithering behaviour, any 24' 27' monitor from benq that is really 8 bit?

416159303

amd锐龙系列核显的Temporal dithering功能有一个bug，就是不能关闭。一直处于开启状态。

我在多年前，做过一个hdmi接口的采集卡。但是采集的数据，和系统的截图不一样。后来咨询amd的工程师，说这是一个bug。就放弃了使用amd cpu，没想到这么多年还没改变。

我的英文不好，你们翻译着看吧。

simplex

416159303

Give us more details - which GPU you tested, what you found, what was AMD answer

416159303

simplex 我当时测试了2200G和 a8 3870，

用这个方法 https://anyware.hp.com/knowledge/how-do-i-turn-off-temporal-dithering-in-an-amd-graphics-card

a8 3870可以正常开启和关闭的，但是2200g 不行。

当时amd是在电话中告知的，具体记不清楚了，只记得锐龙系列核显都是这个问题

simplex

416159303

I suppose, what we call "dithering", is internal grapfic pipeline post-processing ( temporal noise reduction or blending or something else ). It change pixel values each frame, in very low range ( for example +1 .. - 1 value ), and it give eye-strain. All new nvidia/amd/intel GPUs has it.

We can control output dithering via ditherig, ColorControl, or novideo_sRGB. We can check output dithering state reading registers values (check my thread on this forum). But new GPUs did extra image processing

416159303

simplex 好的，我只看了开始的部分，以为你们要关闭它。

simplex

416159303

Yep, 1st message -> https://ledstrain.org/d/2807-safe-hardware-pc-builds-list

« Previous Page