Intel Alder Lake 2160p @ 23.976 Hz passthrough HD Audio dropouts (i7-1270p/N100)

Quote from smp

Diff

diff --git a/drivers/gpu/drm/i915/display/intel_hdmi.c b/drivers/gpu/drm/i915/display/intel_hdmi.c
index 9803347..1ada7b4 100644
--- a/drivers/gpu/drm/i915/display/intel_hdmi.c
+++ b/drivers/gpu/drm/i915/display/intel_hdmi.c
@@ -1927,7 +1927,7 @@ static bool intel_hdmi_source_bpc_possible(struct intel_display *display, int bp
{
    switch (bpc) {
    case 12:
-        return !HAS_GMCH(display);
+        return !HAS_GMCH(display) && DISPLAY_VER(display) <= 10;
    case 10:
        return DISPLAY_VER(display) >= 11;
    case 8:

Display More

This would apply to kernel 6.16. The change is harmless but I'd rather not merge it to LE, even as a temporary patch.

Making this change for everyone would indeed be too much. But how much work would it be to keep a parallel build going for the intel systems we could use that has this patch? I assume if we switch kernels, porting this patch isn't much work as it's a simple oneliner?

Quote from sky42

After some more help and this time from jernej . Thank you.

I have now a script to fix the audio problem for any LE11/12 without extra binary. The same script works on LE11 or 12.
wget https://raw.githubusercontent.com/sky42src/LibreELEC.tv/11.80.5-240204/projects/Generic/filesystem/usr/bin/intel-gpu-max-10-bit.sh && chmod +x intel-gpu-max-10-bit.sh && ./intel-gpu-max-10-bit.sh

I opened a pull request to make the helper script part of official LE.
https://github.com/LibreELEC/LibreELEC.tv/pull/8588

Display More

Hi all - I've run this script on my N150 based GMKtec NUC and will be testing. However, I ran it when connected to an office PC monitor, and will need to turn the box off to move it down to the theater to test. Are the changes from this script persistant?

After some free time available I ended up finding a cable (after testing a FEW) and plugging it in to the DP(LSPCON and TrueHD and DTSHD audio work without issue on a intel 11 i5 nuc.

But using this type of cable it says the source is 8bit and / bt.2020

Whereas using the hdmi port it says 12bit

Forgive my ignorance but could someone explain what's going on here?

Setup is:
Intel 11 paki5 NUC
|
Denon x1800h receiver
|
s90c TV (input signal plus)

I get 12bit is probably pointless at this time but 10bit HDR would be better than 8?

I also tried the patch and it doesn't seem to work on this type of nuc/DP/HDMI/Thunderbolt port

Thanks dudes

The audio glitch is likely the outcome of a lack of bandwidth on the connection as reducing bits-per-pixel (bpc) from 12-bit to 10-bit will reduce bandwidth consumed by video and create headroom for audio; and this appears to work around the issue.

Colourspace and Colourdepth are different things. There is also a distinction between RGB and YUV output and these both depend on the HDMI capabilities of the hardware (including any inline converters) and the connected TV and/or AVR.

The DP cable will be mitigating the audio glitch as it forced colourdepth to 8-bit. Combined with BT2020 colourspace this is still valid for HDR, but 10-bit depth would be preferrable to 8-bit as the latter normally results in visible banding due to the lower number of colours available.

Thanks for the reply. Makes sense.

I did run sky42's script again and can confirm that it does limit the depth to 10bit(must of missed that... doh) but unfortunately I still have audio issues with TrueHD etc.

I've ssh into the nuc then stop the kodi systemd service run the script and started it again.

Does it need to be set on boot or something?

Are you using the right HDMI port (looking from back, left if looking from front). This solves problems on Intel NUC 11TNK/TNH

Thanks for replying mate.

https://www.intel.com/content/dam/support/us/en/documents/intel-nuc/NUC11PA_TechProdSpec.pdf

Please see Section 3.1.4

I have paki5 and pahi5 nucs... They only have 1 x hdmi port. It was a hdmi 2.0b which was updated to support HDMI 2.1 in a bios update. Maybe as the original port was 2.0b it's not getting enough voltage... Who knows.

I think I'm boned. I have 3 of them. It's pretty annoying. Stinking intel.

I did pick up: https://www.amazon.com.au/dp/B0DKJ54JWY?…asin_title&th=1 for about $250AUD the other day with an honoured amazon pricing error. So hopefully throwing more money at the problem will fix it