Video: Outputs - Signal Format and Color Depth
Steps
The EDID decides what an output can do. This page decides what it does — the colour format on the wire, the bit depth, the dynamic range, and the depth of the Windows desktop behind it. Each is set per output.
If you want a 10-bit signal, Output depth has to say 10. That is the single most consequential setting on the page. A 10-bit EDID makes 10-bit possible; it does not make it happen.
Before you start
-
Set the EDID first. Output depth defaults to what the EDID advertises, so an 8-bit EDID gives you an 8-bit starting point no matter what the display can do.
-
Know what the display or processor actually accepts. The page validates against it on Apply, not before.
-
Sign in as Admin. The GPU section is admin-only.
1. Open the Outputs page
In the Hub, select the server, click System Management, then Configure → GPU → Outputs.
Every output on every GPU is listed with its resolution and refresh rate, and a Current: line reporting what it is running right now — format, bit depth, range and desktop depth. That line is the read-back; the controls below it are the request.
2. Choose the signal format
Format is the colour encoding on the wire. Over DisplayPort, RGB is the normal answer and the one to keep.
Some HDMI links cannot carry full RGB at the resolution and rate you are asking for — that is where 4:2:2 or 4:2:0 come in. They subsample colour to fit the signal into the available bandwidth, so reach for them because a link demands it, not by preference.
3. Set the output depth
Output depth is the bit depth of the signal. It defaults to what the loaded EDID advertises — a 10-bit EDID gives 10 bpc, an 8-bit EDID gives 8.
Apply commits one output. Apply to similar pushes the same settings to outputs of the same kind, and Apply to all to every output on the GPU.
4. Expect Apply to be the real test
Some of these settings cannot be validated until they are applied. Asking for 12 bpc on a display that only supports 10 looks perfectly reasonable in the dropdown — the page accepts the selection and only reports that the connected display does not support it when you press Apply.
A rejected setting is information, not a fault. The refusal tells you exactly what the connected device will not take. Set the value back to one it accepts and apply again — in the video, 12 back down to 10.
5. Set the range where it matters
Range selects limited or full dynamic range. It is an HDMI concern in practice: a mismatch between what the server sends and what the display expects shows up as crushed blacks or washed-out whites rather than as an error.
6. Leave Desktop depth alone unless you are doing HDR
Desktop is the Windows desktop's own buffer format, not the signal. The page states the important part outright: desktop depth is 8 bpc by default even when the output runs 10 or 12 bpc, and for ordinary work that is fine.
It matters for one case. If the Backstage output is configured for an HDR colourspace, set 16 bpc float HDR here — which additionally needs HDR support from the connected device.
HDR is a two-ended setting. Changing Desktop depth here is only half of it — the HDR output has to be configured in Backstage as well. Doing one without the other produces a signal that is wrong at one end.
Done when
Each output's Current: line reads back the format, depth and range you asked for — and specifically says 10 bpc if a 10-bit signal is what you are after.
What this depends on
|
Depends on |
Why |
|---|---|
|
The loaded EDID |
Sets the default and the ceiling for Output depth. |
|
Display / processor capability |
Validated on Apply — an unsupported depth is refused. |
|
Link type and bandwidth |
Decides whether RGB fits or a subsampled format is needed. |
|
Backstage HDR configuration |
Desktop depth of 16 bpc float only makes sense alongside an HDR output in Backstage. |
