I have this problem with my just built computer. Windows keeps constantly freezing or rebooting during random
operations (ie. copying files, surfing net, opening some application, browsing the hard drive...). I haven't had it
running more than 30 minutes at time, though I can't say if it would chrash when just idling. Regardles of that,
Windows always installs fine without any problems (done it four times now).
AMD Phenom II X4 955 BE
Gigabyte GA-MA770T-UD3P (rev 1.0 with latest BIOS F5)
2x 2GB PNY 1333 MHz DDR3
Seagate Barracuda 7200.12 500 GB
Antec Earthawatts 500W
ATI Radeon HD5770 1 GB PCI-E
Windows 7 Ultimate (64 bit)
IDE DVD drive
I know that it's not a problem with DVD, HD or power (tested) and pretty sure it's not problem with OS. And I'm also
using all the latest drivers I've been able to find. All the settings in BIOS are set to defaults.
And here's what I've done so far:
My first thought was it was a bad memory. So I run bootalble MemTest and found errors using dualchannel mode. So I
ripped of the other module and ran the test again with errors found. Did the same with the other module and again
found errors. Also tried modules in different slots, but still getting errors. Now I've been running MemTest with
the other module for quite many times and errors have occured maybe in 10 different addresses. They don't occur in
every pass ( maybe every second or third) and only one address have occured twice. That made me wonder if it was a
problem with CPU or mobo that was causing the problem. I have also found that using both modules (dualchannel) the
errors occur faster and it sometimes freezes MemTest.
So my next guess was to test the CPU with prime95. If I run the test that uses lots of CPU and little RAM, it runs
fine for at least 30 minutes (haven't tried longer). Then if I run Torture test (stresses CPU and RAM) it takes from
2 to 10 minutes for the test to stop saying hardware error. If I run benchmark test, the computer always freezes in
few seconds. CPU temp has been at 54 C at max so that should be OK. Also other temps are OK.
One thing that allways freezes or reboots computer apart from prime95 benchmark test is extracting large file (5GB)
from rar-archives. So that makes me wonder if it is an issue with CPU's memory controller, RAM or maby mobo? Or
My BIOS settings are all either auto or default. RAM voltage was set to 1.6V (also tried that manually) and PNY recommenends 1.5 +-0.2V. I lowered speed from 1333 to 1066 and was able to run prime95 benchmark with one and two modules installed. Now I'm running the torture test to see how it goes. So if lowering memory frequency helps with this one, does it mean that the memory just ain't as good as it is promised to be? Or can this still be an issue with mobo or cpu? And some timing settings that I had...
Set Memory Clock: Auto
Memory Clock: x6.66
CAS# Latency: 9
RAS to CAS R/W Delay: 9
Row Precharge Time: 9
Minimum RAS Active Time: 24
Thanks for the help so far!
EDIT: Torture test run about 15 minutes with two DIMMs installed and then system froze.
it's going to be pretty hard to say if it's a bad memory controller or bad RAM without trying another stick... But you might try Starting a ticket with PNY and also checking with the Gigabyte website. Look at the memory compatibility list if they have it... not that it really means a lot on an AMB board with the controller on the PSU, but there can be communication problems between the board and RAM.
So, PNY may be able to give you tweak settings to correct.
Gigabyte may have a BIOS update or settings suggestions.
Or your CPU might be at fault.
Yeah I've contacted manufacturers now but I'll guess I'll have to wait through weekend for their replys.
I've been running memtest now with my RAMs one at a time and trying them in different slots. I found out now that the errors occur indeed in pretty much same addresses (many addresses but they do keep repeating same or almost same). So as the addresses stay the same regardles the slot I'm using, I think it could be that both my RAM's are defected. I guess I'll start my RMA round there .
There is only a fraction, a tiny fraction difference in speed running your memory at 1333 over 1066. You can only see it in benchmarks, and it is a small amount. If it runs at 1066 fine, I would play with the timings and see if you can maybe get it to run stable with the timings a little tighter with it at 1066.
On the other hand though, PNY memory, bah, not my first choice by far..........