I do agree on principle that a bios update wont work, but may as well try. The 4in1 IS the MB chipset drivers, which is what I'm pushing at. VIA is known to have similar problems and they have update the 4in1 drivers for this... I wasn't very clear on the memory registers (memory). I'm sure you know this, but for his benefit I'm explaining the bootstrap process. During initial bootup the bios initializes its own program (basically, turning the hardware to a known/found state) then when thats done it passes control to the o/s bootstrap. This bootstrap either loads the o/s or contains more specific initialization strings. This info is on the HDD 0,0,1 (T,H,S) and it in turn loads the correct os operating environment. I suppose it POSSIBLE for the bios to pass the memory off as a known state and a bios flash is needed for correct recognition, but I'm not to keen on assembly. Anyhow, the HDD then transfers the bootstrap info into memory (512k byte???) and the bios checks the last few bytes for a memory register to tell it the bootstrap is valid. I'm guessing that its very possible to have invalid bootstrap but still show up as valid by the last few characters... but again, this is programming that I've NEVER seen, just know in theory. After all this verification, the bios will hand complete control to the os bootstrap. The memory register would be the location (example 0000:7C00 is the location in memory where the HDD loads the bs.)
Also, the MBD will default to the slowest RAM. You want the largest DIMM in slot 0.
As far as PSU, that CAN be an issue, too. Depends on if the system was already taxed out or not as well as if the PSU is subpar and is not a constant but fluctuates as they tend to get worse not only over time but also over load ratios. Adding the mem can do that. I would still go for the 4in1 drivers, and yes I'd agree that its probably the slott if the drivers dont work, but figured if he got a volt calculator to figure things up it would be the cheapest route.
One other thing to try. I could be the voltage regulator in the bios. Update the Vcore from 1.5 to 1.65 see if that helps. Then try to disable shadowing and cache in the bios (if it boots, it'll be slow but itll give an idea)... If not, then yeah... probably the mem controller.