Hello, I never thought I'd actually get to the point of asking the interwebz for help on this, but I have a computer problem that I am having a lot of trouble solving... so any help, even if it's not a perfect solution, will be appreciated.
Here's the basic problem: I recently started a major overhaul of my gaming PC, and since then I have been experiencing random crashes that always have identical symptoms: whatever image was being output to the screen becomes heavily distorted, like it is being shattered into small lines of bright colors. When it freezes on whiter things (like the Win7 login screen) it's usually only the colored areas that undergo the distortion. Additionally, an incredibly loud, horrible grating/shrieking/popping/crackling sound starts emitting from any speakers. The computer in this state is entirely unresponsive, and I can only stop this by holding down the power button to kill the power.
Now, I am a scientist, so I've tried to eliminate as many variables as I could. I honestly thought I had solved the problem by replacing my motherboard, but it just happened again. THat's why I've broken down. I'll provide as much info as I can about what my components are, which setups the problem has happened with, and which it hasn't.
As a note, I have studied my Windows Event Logs exhaustively, and there is never a single event from around the time when the crash happened. For example, the one today, there is a critical event logged when I restarted the computer at 9:02, telling me a crash occurred at 9:00. However, the first event before the 9:02 power on is more than 15 minutes prior, before the crash occurred (and it's just a normal service log). Additionally, I use Lavalys Everest Ultimate logging, and I log every system parameter possible; the logs always go right up until the crash, but there is never anything extreme or abnormal in the seconds or minutes leading up to it.
Setup when I first encountered the problem:
New AMD Phenom II x6 1055 CPU
New MSI cheapo board
New Diamond Radeon HD 5850 GPU
Old Patriot DDR3-1333 RAM (2 x 2GB DIMMs)
Old WD Barracuda 750GB HDD
Old Creative X-Fi Fatality XtremeGamer sound card
Old Tagan 850W modular psu
Windows 7 Professional
The first time it happened, with the above setup, I had my TV connected to the GPU via HDMI. I believe I had not overclocked anything yet, but I don't remember for sure. Under this setup, I have had the problem happen under every combination of the following variations:
-Graphics card disabled/enabled/removed
-Monitor connected via VGA/DVI/HDMI to onboard GPU/Radeon
-Speakers connected to X-Fi/motherboard/frontpanel/HDMI
-CPU default everything/TurboCore off/OC
The other day, I replaced the motherboard with a less bargain-bin one and got some new RAM.
-New ASUS M4A88TD-V EVO/USB3
-New 2 x 2GB OCZ DDR3-1600 Gold
At first, everything seemed to be working great... I went more than 24 hours with no problem (previously it had been happening at least every 4 hours) I then tried putting the two old DIMMs in as well, thinking it looked like the old mobo was the problem. Everything seemed fine for a day... started to overclock a bit (bumped ref clock up to 250MHz from 200, which also brought the RAM up to 1333 MHz - stock rating for the Patriot set, and below the 1600 rating for the new OCZ.) After a day or so, however, I had the problem happen again.
I've since then removed the 2 old sticks (which I was almost sure were the problem) of memory. When I tried getting the OCZ RAM to actually run at its specified rate and timings (1600 MHz, 8-8-8-24, 1.65v) the computer froze and glitched out shortly after loading Windows. I still eventually encounter the issue even if all BIOS settings are left at their default.
I'm not sure what other information I should add... this is already getting really long, so I'm going to leave it there and hope someone even reads it all... I'd really appreciate any help or advice, and I'll provide any additional info that might help!
hm, you know I hadn't really even considered the psu. Do you know of any ways I could test it before i actually go out and buy a new one? Unfortunately I'm just a poor grad student and I've pretty much spent all of my nerd-funds for a while
i always assume the power supply is either working or dead and there being no in between. and going from 200 to 250 on the ratio over clock is new to me i cant budge my phenom2 720 over 210 or stability goes right out the window and speaker glitching happens when the cpu crashes after tweaking the ratios too high.
switch the ratio back to 200 and over clock using the offered multipliers for over clocking. ? maybe ? vague suggestion.
Well I have the lower tier 6-core that doesn't have the unlocked multiplier, so it pretty much stays at 14x (making it 3.5GHz). I've read reviews where people have clocked it up over 4GHz successfully with increased voltage, and up to 3.8 with stock settings and cooler. Again that's with the default multiplier, and with "AMD Turbo Core" turned off, because that bumps the ratio on 3 cores up to 16.5x and also overvolts them temporarily.
The times when I have turned on "Turbo Core" from the OS with AMD's Overdrive utility, sometimes it has seemed fine, sometimes not. Eventually it always seems to end up freezing, but it also does that when the CPU clock is at the stock 200 Mhz. That's what's driving me crazy about this, even with no OC at all it's still happening eventually.
On the other hand, I've been trying to find any reports on the web of people having the same power supply I have die on them... from everything I've seen it's generally very reliable. In general it seems like people experience similar symptoms to mine when their PSU can't handle the load, but there's no way an 800W psu can't handle my setup, unless it is starting to die. I haven't even had the thing for 2.5 years though.
I'm going to try to get a hold of a multimeter soon, but just observing the values in the BIOS and from Everest logs, the voltages SEEM ok (?) I'll try to post a log from the run-up to one of the crashes, because I don't really know enough about power supply stuff.