Question : Problem: Vista 64+ & 8GB OCZ 5-5-5-12.2

Team I have built a high spec PC with the bellow configurations.
At the moment I have only installed the single graphics card.
2GB Memory to install the system and it worked fine.
System Drivers & Windows Updates done and dusted.

As soon as I add the extra RAM the computer starts to blue screen.

However when I add any extra memory it gives me the BSOD.
I have attempted to change the Memory Timings in the Bios to 5-5-5-12, this did not work.
Then I changed the North Bridge Voltage to 1.30V under the instructions of the company I got the System from.
I was advised that the North Bridge does not like the full 4 sticks of RAM.

Any ideas?

1 - Enermax Galaxy 1000W EGX1000EWL ATX2.2 Modular PSU
1 - Intel Core 2 Quad Extreme Edition QX9650 "LGA775 Yorkfield" 3.00GHz (1333FSB)
2 - Western Digital Raptor 150GB 10000RPM SATA 16MB Cache
1 - Asus P5N-T Deluxe nForce 780i (Socket 775) PCI-Express DDR2 Motherboard
1 - Microsoft Windows Vista Ultimate 64-Bit Edition DVD - OEM - 1Pk (66R-00838)
2 - OCZ 4GB (2x2GB) PC2-8500C5 1066MHz Reaper HPC Edition Dual Channel DDR2 (OCZ2RPR10664GK)
2 - BFG GeForce 8800 GTX OC2 768MB GDDR3 HDTV/Dual DVI (PCI-Express) - Retail

Answer : Problem: Vista 64+ & 8GB OCZ 5-5-5-12.2

If you had an oscilliscope and could actually "see" the waveform on the address and data buses, you'd easily understand why the system is not reliable with 4 modules installed.

With unbuffered memory, every memory chip (16 per module with double-sided modules) represents one load on the address and data bus drivers ... with 4 modules that's 64 loads !!   The square wave looks more like a very sick sine wave at that point ... with very poorly defined transitions.   For a bit more detail, play Item #10 in this presentation:
http://www.corsairmemory.com/memory_basics/index.html

As sparkmaker noted earlier, you can sometimes help with this by increasing the voltage somewhat => this makes the initial waveform have more displacement; so even after substantial degradation it may work better.   Another helpful technique is to slow down the memory timings --> try add 1 to each of the timing elements.   This gives the signal more time to stabilize.

You may also want to try less aggressive RAM modules ... perhaps DDR2-800 modules with a basic voltage of 1.8v.   These present somewhat lighter loads ... so that the same # of loads will have slightly lower signal degradation (not a lot ... but a very small difference can often mean the difference between reliable operation and not).

The BEST approach on systems with unbuffered modules is to simply never load the buses with more than 32 chips (36 if ECC modules are used) => i.e. 2 modules if they're double-sided; 4 modules if single-sided.    That will always give you the most reliable operation.

... There's a reason server motherboards use registered modules => I simply don't understand why more desktop boards don't ... I suspect that may change as systems with a lot of memory become more commonplace.