Quick answer: A DisplayPort to dual HDMI adapter contains an MST hub that splits one DisplayPort stream into two independent HDMI outputs, each with its own timing. One 4K60 output at 8 bit 4:4:4 needs about 12.54 Gbit/s of payload, so two of them need 25 Gbit/s, which is essentially the entire HBR3 budget of 25.92 Gbit/s. That works only if the source supports HBR3 and the adapter uses DSC. Without DSC, the adapter falls back to two 4K30 outputs or to 1080p60 on both.
The adapter is not a splitter in the video sense: it does not duplicate the image, it allocates lanes and time slots so the host sees two monitors. That is why the GPU must support multi-stream transport, why some models need external power, and why the resolution you get depends on what the host reports rather than on what the box promises.
What the adapter has to do
MST, not a passive split
Multi-stream transport packs several video streams into one DisplayPort link, each addressed to a different monitor. The hub enumerates two sinks, the GPU composes two framebuffers, and the hub unpacks them into separate HDMI outputs with separate audio and EDID. A passive Y cable cannot do this, because it has no way to carry two timings on one pair of lanes.
Conversion to HDMI in each branch
Each output runs a DisplayPort to HDMI converter, so the adapter has two of them plus the hub. That is why these units are larger than a single-output dongle, why they warm up, and why some need a USB power lead: the converters and the hub together can draw more than a DisplayPort source guarantees on its 3.3 V rail.
The lane budget, in numbers
Two 4K60 streams
4K60 at 8 bit 4:4:4 with CVT-R2 timing needs 12.54 Gbit/s of data. Two outputs therefore need 25.08 Gbit/s, against 25.92 Gbit/s of HBR3 payload after 8b/10b encoding. The margin is under four percent, which is why a marginal cable or a hub that adds overhead pushes the second display to 4K30 rather than failing outright.
DSC changes the arithmetic
Display stream compression at roughly 3:1 reduces each 4K60 stream to about 4.2 Gbit/s, leaving room for two outputs and even for a third. The GPU and the adapter must both support it, and the monitor accepts the decompressed stream transparently. When all three conditions are met, two 4K60 outputs from one port are routine.
What each configuration supports
| Source link | Without DSC | With DSC | Note |
|---|
| DP 1.2, HBR2 | 2 x 1080p60 | 2 x 1440p60 | 17.28 Gbit/s payload |
| DP 1.4, HBR3 | 2 x 4K30 | 2 x 4K60 | 25.92 Gbit/s payload |
| USB-C Alt Mode, 4 lane | 2 x 4K30 | 2 x 4K60 | Depends on GPU DSC support |
| USB-C with USB 3 active | 2 x 1080p60 | 2 x 4K30 | Only two lanes for video |
Where these adapters fail
Four faults cover most cases. The source runs DP 1.2 while the adapter assumes HBR3, so both displays drop to 1080p. The HDMI ports are 1.4 rather than 2.0, so each output caps at 4K30 regardless of the DisplayPort side. HDCP negotiation fails on one branch, which shows as a black screen on one monitor only. Or the adapter is bus-powered and browns out under load, producing dropouts that look like a cable fault.
Diagnose by connecting one display at a time: if each works alone but not together, the limit is the lane budget, not the converter. Then check the GPU’s reported link rate and lane count. A DisplayPort MST hub explanation covers enumeration, the DisplayPort 1.4 cable page covers the HBR3 cabling requirement, and the HDMI cable rating categories guide explains why each HDMI branch needs a cable rated for 18 Gbit/s.
Audio, EDID and mixed monitors
Each HDMI branch carries its own audio, so a hub with two outputs can feed two different sinks, but the host still sees one audio device per branch only if the hub passes the audio infoframes through. Many adapters forward audio to the first output only. If sound disappears when the second monitor is connected, set the audio output explicitly in the operating system rather than relying on the default device.
EDID handling decides what resolutions are offered. A hub that stores a fixed EDID will report 4K60 even when the attached panel is 1080p, and the host will then render at 4K and let the panel scale, which costs performance for no visible gain. Adapters that pass the real EDID through behave better with mixed monitors.
Practical scenarios
- Trading desk with two 4K60 monitors: use a DSC-capable adapter and confirm the GPU supports it; without DSC, plan for 4K30 on the second screen or use a second GPU output.
- Laptop with one C port and a dock: the dock already contains an MST hub, so adding a second adapter in series adds no bandwidth and may not enumerate at all.
- Meeting room with a 1080p projector and a 4K TV: mixed timings are fine, but set both outputs explicitly rather than relying on the adapter default, which may mirror at the lower resolution.
- Digital signage with two identical panels: if the content is the same, a splitter that duplicates one stream is cheaper and avoids the lane budget entirely.
FAQ
Can I get two 4K60 displays from one DisplayPort?
Yes, if the source supports HBR3 and display stream compression, and the adapter implements DSC. Without compression, two 4K60 streams need 25.08 Gbit/s, which leaves almost no margin on a 25.92 Gbit/s link, so most adapters fall back to 4K30 instead of risking instability.
Does the adapter work with a USB-C laptop?
Only if the port supports DisplayPort Alt Mode with four lanes and the host supports MST. Many laptops expose MST over a C port, but some route video through a converter that does not, in which case only one display appears.
Why does one monitor show a black screen?
Usually HDCP: one branch authenticated and the other did not, often because one HDMI cable is older or longer. Swap the cables between outputs to confirm, then replace the failing one with a Premium High Speed cable rated for 18 Gbit/s.
