J3455-ITX HDMI edid with 4k Display

I am having quite a mysterious problem with new J344 board and Samsung 4k television.

hjhamala wrote: I am having quite a mysterious problem with new J344 board and Samsung 4k television. I cant get picture to the television via hdmi neither in startup, windows or in linux.  Linux dmesg sayes that edid is corrupt. What is quite interesting is that if I connect the same television via hdmi 1.4 capable amplifier and switch quickly hdmi cable straight to tv both OS suddenly can read EDID from the television and I get 4k60p picture. If soft rebooting (not powering off whole board) hdmi output works in startup (bios etc). I amd thingin that this might be some sort of bios problem. Is there any advice what to do? Intel doesnt support yet forcing EDID. Linux supports but I can get it only work with Fedora 25 not with Ubuntu 16.04. Unfortunately I cant get Steam working with hardware accelerated decoding in Fedora.

Your problem may not be very mysterious for several reasons, sorry to say.

One quick spec, your board supports Windows 10, Ubuntu 16.10, and Fedora 25 officially.

I don't see anything in the UEFI/BIOS that would make any difference with your situation.

EDID is a standard used to communicate information from a PC monitor to a graphics source. My point is, do you know if your TV, or any TV, has the EDID data structure available for the video source to read? The output of Linux dmesg that the EDID is corrupt could simply be a generic message that is displayed even when it does not receive a EDID data structure from the TV.

I have a new 4K monitor, and had to install a "driver" file supplied by the monitor manufacture before the video signal would change to 4K on that monitor. If the EDID or E-EDID data was being read or used by the video source, why does that file need to be installed?

Televisions and PC monitors, while sharing some resolutions, are not classified in the same way by PC video source drivers. For example, Nvidia video chip software will list PC and TV monitor resolutions in different lists. If they were both identical in all aspects, why categorize them separately?

Usually, when TVs have problems when used as monitors, they fail or their resolution is incorrect while in the UEFI/BIOS screen. In your case, the ASRock HD UEFI is set to 1080p, and that is all you get with a 4K monitor or TV. My 4K monitor's resolution is not at 4K while using the UEFI.

IMO, your situation where you change the HDMI cable to the HDMI 1.4 amplifier output, which I assume is already transmitting a 4K signal, the EDID data structure is not being read at all by the video source. Did you ever change the resolution of a monitor with the video driver's software? Nvidia and Intel both use the same protocol in their software. You select the new resolution, and then click the Apply button. The monitor goes black while the new resolution is applied, and the EDID data of the monitor is read (which exists in a special storage area and is not included as data in the video signal itself) to verify compatibility before the switch is made.

The video driver being used for Linux will be a problem, since Intel does not directly support Linux video drivers. What Intel video driver are you using? These are the latest Intel video drivers for the Intel HD 500 graphics:

https://downloadcenter.intel.com/product/96553/Intel-HD-Graphics-500

I recently discovered that the resolutions available to a monitor via the video driver's software were not the same as those available via the Windows Display feature in Control Panel. The Windows Display feature actually had more and higher resolutions available. Granted, that was not via an HDMI/digital video interface, but is an example of either the complexity of the situation, or bugs in the video driver software, or both.

The full 4K x 2K @ 60 Hz resolution via HDMI is not provided solely by the Intel HD 500 graphics in the board's CPU. It is provided by hardware acceleration via software, as you know. The newest and best Intel HD graphics, and many video cards that are far more powerful than that, cannot provide 4K x 2K @ 60 Hz over HDMI.

That is really the point of all this. This is a complex situation with multiple variables and several unknowns.

I am quite certain that the television provides EDID because I can read the EDID via HDMI if i make the quick switch after connecting computer via amplifier. I can save the EDID both from Windows and Linux.

Stefan75 wrote: Umm, actually the J3455 can only do DP 1.2 and HDMI 1.4 (like all modern Intel CPUs, even the newest i7). ASRock put a additional chip on the board, it converts from DP 1.2 to HDMI 2.0. I've been chating with Intel about my UHD 60Hz 4:4:4 issue. https://communities.intel.com/thread/111370

Thanks for the link. I read through that.

I don't think the LSPCON converts the processors native HDMI to anything. The J3455 already has native HDMI, so I fail to see a reason to complicate matters in converting a native HDMI signal.

The ASRock J3455-ITX doesn't have a DP port so the J3455 processors native DP is, pardon the wording, tossed to the side. This particular motherboard doesn't have a DP port at the rear I/O panel.


Have you set the TV's to NOT upscale? I believe that may be what you're experiencing there confusion wise. The motherboards rear I/O HDMI Output is direct from the CPU. Since HDMI doesn't support the resolution it has to be the television's internals that are upscaling on the display.



Please, I'm no Engineer, so correct me where you see me going off course in my thinking above.

I really wish Intel had 1 processor with native HDMI 2.0 output, I haven't seen it. It's quite poor, maybe they're waiting for HDMI 2.1 ;)

Edited by Stefan75 - 15 Feb 2017 at 6:27am