Give that radxa states the possibility to supply power via GPIO Pin 2+4 explicitly on their own wiki to the product, I would say that is normal usage and thus covered by acceptable use.
Apart from that tkaiser is right. (and I wasnt aware that a direct fixed voltage power supply was a possibility, just read up on it)
To clarify up on this.
My suggestion to power the board up using alternative power paths(in this case GPIO) exactly has the point of bypassing the first(and in this case broken) buck converter. It doesnt matter, since as tkaiser rightly summarized thereās a secondary on the board.
My whole reason of asking the OP to do this was my curiosity about whether the whole board was fried or whether we actually got some quality and the board has secondary protections against power surges etc(as in a smoked primary buck converter doesnt kill the entire board but rather has a protection that prevents the rest of the board from getting the spike).
As an addition to satisfying my curiosity this could also help in the rough investigation on the source of the problem that occurred here. The OP states that it went up in smoke when he attached a hdmi to vga converter. This is something that shouldnt ever happen, which leaves me to believe the source of the fault could have either been a short caused by dust et al, a fault in the power supply or maybe even static electricty that got transferred from the op when attaching the adaper, even though I highly doubt that.
If however the board works beyond the known-broken circuitry that would at least hint at an acquittal for the hdmi adapter used.
From a guy who recommends 5V @ 6A which is pure nonsense. 
For the GPIO power, currently I do not have 5V 6A DC supply only 5V 2.5A RPi PSU. Not entirely sure if it is possible for me to test.
I can confirm no dust on the HDMI port of Rock5B nor HDMI-VGA converter, anti-static precautions have been observed. I had used other SBCās before this one. The same HDMI-VGA adapter continues to work on my Laptop and Raspberry Pi 3B+ SBC.
You donāt need 6A. As per (I assume they are one and the same) @tkaiser s page here: https://github.com/ThomasKaiser/Knowledge/blob/master/articles/Quick_Preview_of_ROCK_5B.md
The RPI power supply should be plentiful for a quick test run
I tried to power on the device via 5V on GPIO pins and it does boot loop, cannot get anything on the display except (No signal to detecting input to no signal again, loops). Will get a serial console later and check output. After this, will schedule the return to allnetchina. I need to look for cheap ways to return this as the fee for return is expensive on where I am at.
Hm. Taking in account that there is no such thing as āpassive hdmi vga adapterā - I believe that adapter tried to take voltage it needs to run from HDMI pin 18, which lead to frying the HDMI slot. As for protection circuit that usually prevents it - i donāt think it was either in adapter or rock5 (even x86 iGPU not always have them)
Thatās why you donāt use āpassiveā adapters.
Update: The busted IC has been replaced by allnet.
All is good now.