I report my experience with RX 590 and RX 5700.
Gigabyte X299 - RX 590 (Mojave):
- On Clover/Graphics I was already using the Orinoco framebuffer, so I left it like this.
- AGPM Controller was already loaded by default, so I didn't use dAGPM.
- agdpmod=pikera was not needed in my configuration.
- On System Info/PCI there is no news, there is only RX 590, but currently almost all SSDTs that I use is not displayed on the PCI information.
Power consumption
- IDLE: 90-100 Watt (with and without your SSDT)
- On rendering: up to 350 Watts and sometimes over (not tested with your SSDT)
But I think they are incorrect readings from iStat, because Sapphire declares a consumption of <235 watts, and 350 seems to me to be really too much for this card.
I was already using an SSDT for the RX 590, that of the KGP guide.
After replacing the new SSDT RX580, adapted to my ACPI paths, I didn't notice any difference in geekbenck, sometimes the score is higher, sometimes lower, exactly as before (around 51000 OpenCL - 56000 Metal).
The only real difference is that this new SSDT sometimes on waking from sleep, the GPU fans spin at maximum for a few seconds, while at other times it breaks sleep and the result is a reboot.
Gigabyte Z390 - RX 5700 XT (Catalina):
- On Clover/Graphics no framebuffer set, I don't know which one is for the RX 5700.
- AGPM Controller was already loaded by default, so I didn't use dAGPM.
- agdpmod=pikera already present.
- On System Info/PCI it is empty without SSDT, while with the SSDT the RX 5700 and the HDAU output are displayed.
After adding the SSDT it seems that the performances shown by geekbench are better even if we are far from what the performance of this card should really be.
The problem is that after a while the screen freeze and the computer stuck. This happened shortly after a sleep, but I don't know if sleep was the cause. Also during the benchmarks the computer made strange noises, like buzz. For now I have removed the SSDT.