I have searched the archives of this forum, and googled, but noone seems to be having quite same same issue as me. Also, most are having it with the IP35 Pro, rather than the Pro XE. BTW, if it matters i'm running an Intel Core 2 Quad Q9550 in there, at stock speeds. My OS is Vista 64 bit, fully patched and up to date.
Anyway, I am having issues with my Corsair Dominator RAM. Ever since I built this PC i've had occasional BSOD issues that always complain about atikmdag.sys. Sometimes the BSODs were random, and would happen in the middle of playing WoW or just browsing the web, but the system would *always* BSOD if I let the monitors go to sleep/powered them off, then tried to bring them back online again.
I looked online, only to find a lot of people saying that this was often an indication of a RAM error rather than a graphics driver or card fault. I checked my BIOS, only to find i'd been running my memory at 800Mhz instead of the 1066 it was supposed to do, and also I was running the DDR voltage at 1.8V instead of the 2.1V the modules required.
I upped the voltage to 2.1V, and changed the CPU:RAM multiplier to 1:1.50, which gave me a RAM speed of 1020Mhz. This was the closest I could get to 1066Mhz. I also manually set the timings to 5-5-5-15-2T. The other post I found on the forums of someone having similar (but worse) issues with the IP35 Pro he eventually fixed by changing his multiplier from 1:2.00b to 1:2.00a. As the IP35 Pro XE only has one 1:2.00 setting, and as my CPU is 1333Mhz FSB so this would severely O/C my RAM, I didn't want to go that high.
These changes (well, the voltage increase, the rest of it was just a speed boost for the system) got rid of my 'reliable' BSODs when sleeping/waking up or power cycling the monitors, but didn't get rid of the random BSODs (about 2 or 3 a week) I am having.
I then downloaded and ran memtest x86, and it crashed and rebooted my PC within 20 secs of starting. I tried removing one memory module, and it passed fine. I then swapped to the other module (both in both DIMM slots I had been using previously) and it passed fine as well. It appears the fault only shows up when running the RAM dual channel. I tried swapping it briefly for 2 generic DDR2-800 sticks out of my PC at work. My home machine completed memtest x86 fine with these DDR2-800 modules in running dual channel, while my work PC crashed memtest after 20 secs with the Corsair modules (it is worth noting, however, that my work PC *can't* run these modules at anything other than 800Mhz and 1.8V. It doesn't have any O/C ability in the BIOS).
I thought it was definitely a RAM fault, so I ordered a new matched pair of the same memory. Unfortunately, the new RAM is doing exactly the same thing. I am now wondering if I need to tweak up another voltage somewhere, or if I have a motherboard fault (I am doubting the latter, as my work box did the same thing). Unfortuantely having never overclocked before all the different voltages in the BIOS scare me, and I have no idea what most of them are for. I have options to alter CPU VTT, MCH, ICH, ICHIO, DDR 2 Reference Voltage (-2% to +2%, rather than a set value), and CPU GTLREF 0&2 and 1&3 (also percentage values, but larger. Currently at 67%). I really don't want to fry my system, as I can't really afford to replace it at the mo, so I was hoping someone on here could tell me what I could try upping to attempt to get this system to run stable, while hopefully not toasting anything.
I did manage to find out that MCH stood for 'Memory Controller Hub', and found a ticket on here where someone else was having issues with the IP35 Pro and Corsair Dominator RAM, so I tried following the advice in that thread and upping my MCH to 1.33V from stock (1.25V), but I still had the same issue. I went up to 1.37V, same issue, and was scared to go any further in case I toasted something It's now back at 1.25V.
Can anyone offer me any advice on how to get my system running stably so it'll actually pass a dual channel memtest x86 test?
Thanks