[joevt]

I will send an e-mail to [email protected]

One note:

The latest edid-decode can dump the AMD Vendor Specific Data Block block now.
The bit change in the AMD VSDB is part of its version number.
The Cable Matters and CAC-1080 have this as v1.1.
The CAC-1085 have this as v1.0.
I don’t know why the display would give different adapters different version numbers for the AMD VSDB. The version doesn’t seem to affect HDR working or not since both the CAC-1080 and the Cable Matters show the same version.

[joevt]

MST hubs are daisy-chainable (you can connect a MST Hub to another MST Hub) so you shouldn’t have to disable the MST feature. That seems like a bug with the AMD driver or Windows. But you said it works if the MST Hub is not connected during startup but connected later. That could make it a difficult bug to solve (or at least reduce AMD’s motivation to fix it).

[joevt]

Do you have an Intel Mac that you can use AGDCDiagnose on to get the OUI and FW version for the chip? Or perhaps a DPCD dump from Linux will be sufficient.

I have a Acer XV273K display. It has DisplayPort 1.4 and HDMI 2.0b inputs.

I’m using macOS Big Sur 11.5.1 on a Mac mini 2018 which has an HDMI 2.0b port connected to Intel iGPU (Intel UHD Graphics 630) and Thunderbolt ports also connected to iGPU. I also have a W5700 eGPU (6 DisplayPort 1.4 outputs supporting DSC, one of them is a USB-C port with DisplayPort Alt Mode).

The CAC-1080 doesn’t give me an HDR option but the Cable Matters 201388, Apple USB-C Digital AV Multiport Adapter, and the Mac mini 2018’s built-in HDMI port can all give me HDR from any of the ports (Intel or AMD). The Cable Matters and Apple adapter require 30Hz from Thunderbolt. 60Hz works from the other ports including the Mac mini’s HDMI port…

For the Apple adapter 30Hz is expected from Thunderbolt since it’s two lanes DisplayPort 1.2 and DisplayPort 1.2 doesn’t support 4:2:0.

The Mac mini’s HDMI 2.0b port is using four lanes HBR2 with 4:4:4 10bpc input and 4:2:0 10bpc output 4K 60Hz (according to DPCD registers obtained by using AGDCDiagnose command – the HDMI adapter MegaChips MCDP2920A4 does 4:4:4 to 4:2:0 conversion).

For the Cable Matters 201388 adapter, from Thunderbolt:
4K 30Hz HDR (no format conversion info listed in AGDCDIagnose)
For Intel GPUs, AGDCDiagnose doesn’t give input or output pixel format information.
Since DisplayPort 1.2 doesn’t support 4:2:0, and there’s no 4:4:4 to 4:2:0 conversion, then 30Hz max for 10bpc is expected. It seems to me that the Synaptics VMM6100 should be able to do format conversion, (DPCD register 5 indicates so) but it’s not happening. The MegaChips has register 80 specifying HDMI for DWN_STRM_PORTX_CAP, but the VMM6100 says DisplayPort. I guess I should find the Format conversion info in the DisplayPort 1.2 spec to see how that’s supposed to work (DisplayPort 1.4 spec cannot be found online). Maybe Apple has special software for the Megachips (which it uses in many products including the Apple TV and M1 Mac mini).

For the Cable Matters 201388 adapter, from AMD W5700:
4K 60Hz 10bpc HDR, HBR2 x4, 4:2:0
In this case, DWN_STRM_PORTX_CAP is HDMI so I don’t know what happened in the Thunderbolt case. No Format conversion is required with this connection since the AMD supports DisplayPort 1.4 which supports 4:2:0 which is required for 10 bpc which is required for HDR.

For CAC-1085 from AMD W5700:
4K 60Hz 4:4:4 8bpc (no HDR option).
Maybe macOS is dumb or too smart – it won’t do HDR unless some specific conditions are met. The Cable Matters adapter meets those conditions. So what’s the difference?
One thing strange about the Cable Matters is that it’s showing DSC Support: 0, which seems wrong for a HDMI 2.1 adapter. Maybe it’s reporting the DSC capability of the display. My display is only HDMI 2.0 so DSC is not required. DSC would be required if the adapter was only getting two lanes but that’s an unlikely situation.

There is a single byte change in the EDID between the 201388 and CAC-1085. That shouldn’t happen if I’m using the same connection on the same display for each. It appears the single bit change is in the second byte of the AMD vendor specific data block – maybe it’s just a flag that signals HDR/not HDR? I can’t seem to find info on the AMD vendor specific data block contents.

[joevt]

Optical cables are single direction. I assume you had it connected in the correct direction. What was your HDMI 2.1 source and display? The cable is backward compatible. Did it work for HDMI 2.0 (in case you don’t have any other HDMI 2.1 sources or displays to test)?

[joevt]

club-3d makes several HDMi 2.1 cables (Ultra High Speed HDMI 4K120Hz 8K60Hz 48Gbps). Did you try those?

They have different lengths:
Copper: 1m, 1.5m, 2m, 3m, 4m, 5m
Optical: 10m, 15m, 20m

[joevt]

Is your firmware version the same as mine?

 

[joevt]

You still need to get the firmware update from club-3d but at least you can check the current version.

[joevt]

There is a tool at https://kb.cablematters.com/index.php?View=entry&EntryID=147 (downloaded the 201388FirmwareUpdateTool.zip ) called VmmDPTool.exe that might work. It can extract and update firmware from many Synaptics devices.

The tool says the CAC-1510 contains a VMM-2320. This matches the numbers shown in the AGDCDiagnose output on a Mac […-…-…-…-035-032] = 0x23, 0x20. The firmwares I have for HDCP-ON and HDCP-OFF are version 2.33.102.

 

[joevt]

Any MST hub should be able to do the same. DisplayPort 1.4 MST hubs like the CSV-1550 are the most capable because they support the highest DisplayPort bandwidth (HBR3 x4 with DSC). All the DisplayPort MST hubs I know including the CSV-1550 use the Synaptics chips (click all the “Read More” buttons at https://www.synaptics.com/products/video-interface-ics )

Also, if you have a GPU and OS and graphics driver that supports DSC, then it effectively increases the input/output bandwidth of the MST hub even more. For displays that don’t support DSC, the MST hub can take a DSC input and decompress it for output. This is how it is able to support three 4K 60Hz displays.

GPUs that support DSC are Navi (RDNA), RTX, Ice Lake, Tiger Lake, M1 Mac. macOS Big Sur and later might not support DSC. macOS Catalina supports DSC.

As for your problem with a PC laptop, PC laptop specs often skip details about maximum number of connected displays, or what GPU each port is connected to, or if a USB-C port supports DisplayPort Alt Mode or not.

Only PCs with Thunderbolt 4 are guaranteed to support DisplayPort Alt Mode (and also two displays via Thunderbolt) because Intel won’t let them use the Thunderbolt 4 name without those features. Other PC laptops without Thunderbolt 4 might only support one display from a Thunderbolt 3 port.

All Macs have Thunderbolt 3 ports that support two displays (except the M1 Mac where the second DisplayPort connection of the Thunderbolt port can only be used by dual tile displays like the LG UltraFine 5K or the Dell uP2715K display).

The billboard device is a feature of USB-C devices. If you connect a USB-C device that requires a specific USB-C Alt Mode (such as DisplayPort Alt Mode or Thunderbolt Alt Mode) but your USB-C port doesn’t support that feature then the OS can use the info of the billboard device to display a message.

The manual for the Asus ROG GL502VS-DB71 seems to indicate that the USB-C port only support USB 3.0. That’s 5 Gbps. DisplayPort 1.2 (HBR2) is 5.4 Gbps (per lane). DisplayPort 1.4 (HBR3) is 8.1 Gbps (per lane). The ROG GL502VS is from 2016. The Intel GPU can support 3 displays including the internal display. The Mini DisplayPort and HDMI port add up to 3, so it’s unlikely the USB-C port supports a display. But your laptop has an Nvidia GPU as well? But the specs don’t say what ports it’s connected to if any. Connect a display to all the ports and see what GPU the Device Manager says they are connected to. (view by Connection).

[joevt]

Apple supports mirroring on MST but not multiple displays. I don’t think they changed that with the M1 Macs. Apple also supports MST for old 4K displays that used two streams to get 4K60.

An MST hub can be used to convert narrow/fast DisplayPort to slow/wide DisplayPort or vice versa. For example, the Club 3D DisplayPort to Dual Link DVI adapters use HBRx4 but USB-C docks might only support HBR2x2 or HBR3x2 so you need an MST hub to convert HBR2x2 to HBRx4 to get 2560×1600 from the adapter in that case. This DisplayPort conversion feature of MST hubs also works on Macs.

[joevt]

The CAC-1023 appears to be a dumb cable – it should not modify the capabilities of the source or sink.

Maybe it adds too much noise in the signal because of the extra length and the transition between two cables. Maybe the shorter CAC-1022 would work better. A single cable would have the best signal. Or maybe a repeater like the CAC-1007 should be used in this situation.

[joevt]

Cable Matters says the 201388 uses the VMM6100.

[joevt]

Synaptics has some chips that can output HDMI 2.1 VMM8210, VMM8100, VMM6210, VMM6100 but I don’t know how well they work compared to the RTD2173 chip.

[joevt]

[quote author="joevt"]
Currently, we have DisplayPort 1.4 to HDMI 2.1 adapters (CAC-1085 and Cable Matters 201388 which probably use the RTD2173 chip)
[/quote]

Synaptics has some chips that can output HDMI 2.1: VMM8210, VMM8100, VMM6210, VMM6100

 

[joevt]

Test with a DisplayPort display using a USB-C to DisplayPort cable?

[Author]

Posts