I can, but memory will not get trouble too much.
If error was cause of memory, I'm going to do it.
I will focus on HANA and GPU.
GPU will be main problem. GPU got wider area to cause hogging and sacking.
Also more balls to arrange then other.
HANA will be enough reflow with SMT are flow rework station.
Big problem is BGA ball seat on chip was damaged, we can't bring back.
I'm not sure any encrypted information was store on GPU or not.
One GPU was take me about 45 minutes to place balls, re-arrange and sit on GPU by heat.
