With limiting RAM bandwidth increases, and other interconnects reaching close to current RAM bandwidth speeds, APU's don't benefit from using system RAM as GPU buffers as much.
So i was thinking aloud of what the PCIe 4.0 standard really brings to the table in performance.
For a x16 lane configuration that is upto 32GB bandwidth per second by directional, equalling to dual channel DDR4 memory at 2133MHz on it's own.
So in theory if you use dual channel DDR4 memory at 2133MHz, and some other usecase can handle a different memory pool using the PCIe 4.0 interface, that would be very nice to offload system RAM, right?
Then i came to the realization a possibilty of having a PCIe 4.0 enabled APU and mainboard combo, with the GPU on the APU using the PCIe 4.0 interface to connect to it's own memory pool using a modified High Bandwidth Cache Controller (HBCC for short), offloading system RAM alot in memory sensitive tasks like games.
This essentially can mean a add-in PCIe 4.0 memory card can be easily put into a small form factor PC, giving a upgradable video buffer of max 32GB (bi-directional) per second in addition to say DDR4 running at 2133MHz (32GB/s). this would offload the memory controller alot, and double the bandwidth for the whole system (32GB/s for GPU, and 32GB/s (or more) for CPU), especially in RAM heavy tasks (1440p gaming as example).
I know it is unlikely to happen soon, and i even do not know if there will be too much latency and interface overhead doing so.
But it could be another good way marketing to push for the PCIe 4.0 standard, making it easier to justify the mainboard cost price increase, and give a more realistic use of PCIe 4.0 for APU gamers.
So in summary imagine a scenario like this:
A 65W TDP AMD APU with:
6 to 8 Zen2 cores (1 chiplet) at 3.2GHZ (consuming around 35watts at those clocks)
A graphics chiplet consuming around 35watts max.
With another 12watts for a slightly new io chip to allow use of the main PCIe 4.0 16 lanes being used as video memory if present in the PC.
Customizable add-in card for VRAM buffer size.
Which should be capable of 1080p gaming in 60fps with most titles at decent quality levels.
imagine PCIe 5.0, 64GB per second throughput bi-directional?
That would easily enable entry 1440p gaming with more than 60fps on a APU in the future, having a 65W-95W TDP.