What is the technical reason for modern graphics cards only supporting four monitors?

Question

In January 2022, consumer-grade graphics cards support only four monitors each in the higher tiers. What is the technical limitation that dictates only four?

DisplayPort 1.4a MST can support 32.4 Gbit/s and in MST mode this will support up to four monitors 1080p@60 Hz.

The 11G-P4-6696-KR graphics card has three DisplayPort ports. What technical limitation prevents the vendor from allowing all three ports to support x4 1080p@60 Hz?

With 11 GiB of graphics memory I would expect that many frame buffers could be supported for simple things like web browsing and other office type applications.

1080p ARGB 8 bit depth (32 bits), color buffer only, is slightly under 8 MiB.

11 GiB / 8 MiB = 1408 frame buffers for basic desktop tasks. For the uninitiated, this is a gross simplification if the desktop composition environment which will have many rectangular surfaces that have textures applied to them, but this is a lightweight computation for this class of graphics device.

So with that in mind, the limiting factor must be something else in the graphics hardware, possibly the bandwidth of the ROPs. Another possible bandwidth limitation would be the encoder that has to transform the in-memory frame buffer representation into a format understood by the target monitor.

Answer 1

Generally even a protocol such as DisplayPort requires some kind of transceiver electronics that are completely distinct from the other buses. The way a graphics card talks to a monitor over HDMI or DisplayPort is going to be significantly different from the way it talks to memory or the internal units. Either the voltage will be different, it will have different current usage, or will have error correction to handle going down a cable of questionable quality.

What you are probably seeing is some limitation in tagging hardware streams out to various display transmitters. There may be some complicated multiplexer/switching arrangement to merge MST streams together and the actual number of outputs from the main GPU to the multiplexer could only have 4 ports. There are apparently graphics cards that do support 6 streams so they likely have more ports going out from the GPU to the display transmitter electronics.

It is not a factor of memory bandwidth, nor likely anything to do with ROPs or anything internal to the GPU or memory bus, and it likely boils down to a marketting and technical choice to say that "most users probably only need X number of displays" along with "it's easy enough for the user to buy another card if they need more".