Issue with AMD 6800xt + Level1 DP 1.4 KVM single monitor

I’ve been using this (awesome) KVM for some months now and, apart some strange behavior under Linux with gaming mouse and keyboard (Corsair) - I just had to cherry pick the usb hub where the Kvm is connected - all is fantastic.

Not sure how, but I was able to put my hands on a XFX 6800xt MERC319 and I am not able to make it like the awesome kvm in any way and I have no idea on how to troubleshoot the issue. I can provide just a list of basic tests I did to try and get my head around it:

  • I’m using the cable from the level1 store. They have been working with the previous card (Vega 64), they work if connected directly to the monitor
  • the kvm DP ports look fine as if I connect anything else (laptops, other machines) all works perfectly
  • the problem is the same under Linux (wayland and X ) and Windows
  • If I connect via rdp (while the kvm is connected) I see like there is no monitor connected at all

I suspect that, since my monitor is only DP 1.2 compatible, there is something wrong going on with the protocol negotiation, but I would not know how to troubleshoot it.

Any suggestion on where to look (preferably on Linux)?

Anyone has already tested this HW combination?

Just to share another data point.

I am running a 6900 XT in Linux. I have a L1 KVM. It is hooked up to an HP ZR30w monitor which I believe is DP 1.2. I seem to remember having problems with some DisplayPort cables back in 2011 when the monitor was new, because 2560x1600 @60Hz @10-bit is definitely more difficult than 1920x1080.

After that data point a suggestion: make sure the cable from the KVM to the monitor is also high quality.

1 Like

Since you said this worked with a Vega64 one difference I can think of is PCIe4. I have been reading about all kinds of interference problems being caused by this. It’s been blamed for USB failures, WiFi interference and sound issues.

It’s just very high frequency and higher power levels too.

If you are running PCIe-4 then maybe, just to rule it out, set your BIOS to limit to PCIe-3.

I have a mother board that It’s still on PCIe3 only, so this should not be the problem.

I share the same doubts you have about the cable KVM->Monitor, I’ll try to swap it as a next move. I tried using L1 cables both from source to KVM and to KVM to the monitor. They both work taken singularly to connect directly to the monitor, but it’s also true that chaining them adds up quite a lot to the total length.

Any chance to get a suggestion on how to get a good cable? Any particular brand/model?

Just an update for posterity: I got a new short cable, supposedly high quality from a German brand ( KabelDirekt), certified DP 1.4 (who knows…for sure it feels high quality) for the KVM->monitor segment and the behavior is unchanged.

Anecdotally, using vnc I verified that both Linux and Windows fall back to a very low resolution after being connected to the KVM and keep that resolution after plugging back the monitor directly. Probably meaningless, they just think to have nothing attached.

My monitor is an Asus MG279Q (just in case someone with the same HW combination will stumble on this and report his/her experience)
If I’ll change the monitor in the future, I’ll try again.

I’m just very sad to lose functionality of this amazing KVM just few months after buying it. I’ll keep using it for USB switching and do the video part manually.

You may need to hard power cycle the KVm.

Low res is indications for bad negotiation which may be more rembered than you realize. Did you get one or two new cables? Remember the KVm has no repeater so each cable might work individually but not together as from the video cards point of view it has to push the signal twice as far

1 Like

I got one cable. I’ll get a couple of new ones as short as possible - hoping that they are good - and try again

Thanks, I’ll keep this posted if I have some new info

I tried with 2 more cables (from the same vendor), supposedly high-quality, no change.

Not sure if this means something: I parsed the edid data from /sys using edid-decode both whith the KVM in the middle and without it (direct cable) and they are identical…does this means that the negotiation is successful or is this just a part of what’s being negotiated?

edid_kvm.txt (5.2 KB)

yeah, probably not actually high quality. There is no repeater in the KVM. Each cable can work fine individually but understand when the KVM is in the mix you’re asking two cables together to also still be fine.

Digital signals are great until the signal-to-noise ratio is too high… then they aren’t. Is club3d an option available to you?

I get TrippLite “8k” cables all day from Amazon. It’s one brand that is almost universally decidedly not 8k < 10% actually work at 8k.

Many other “8k” cables have a success rate between 70-90% when you are truly pushing them to their limits.

Hence the importance of trying other brands :slight_smile:

I’m checking Amazon right now and it seems like it is an option. Thanks, I’ll try them too and let you know

So, I got 2 club3d 1mt cables (DP 1.4) and the result is still the same. I tested all the possible combinations and the KVM is fine. It works with 2 different laptops, it works with my old vega64 card. It always works, just not with this particular 6800xt.

I’m inclined to just live with it at this point. I’ll try again if I’ll change the monitor or the video card and see what happens

Very weird. There is a gui drive strength option in windows Radeon drivers on the right side under custom

Do you mean the “preemphasis”? Never noticed it before. Tried up to the max, still does not work (tried also all the combinations between that and “voltage swing”).

Now I am curious, if nothing else, for “scientific” reasons, about what’s going on inside those cables. I guess that there must be some kind of oscilloscope for dp testing. For sure, crazy expensive

Today’s update: I got a new monitor ( Benq EX3415R) and the problem is completely gone, everything works as expected with the KVM. I retested all the DP cables I got, and they all work. It’s basically impossible to reproduce the problem with the new display.

This does not explain what was wrong with the previous monitor, I guess it will remain a mystery. At least if someone has the same problem, could use this experience as a reference.

This topic was automatically closed 273 days after the last reply. New replies are no longer allowed.