system stability issues. need advice

test_tube_tonytest_tube_tony Dallas TX Member
edited September 2009 in Hardware
heres what ive got....
used evga 780i triple sli mobo, (not ftw version) with bios p08
used e8400 cpu E0 stepping (stock speed)
used evga 9800 GTX
new corsair tx750 psu
new 4x 2 gig g.skill ddr2 1100 clocked at 1066 (8 gig kit)
mem and fsb are linked with a 5:4 devider

the system ran great for the first almost week. trouble started (or at least noticed) after updating the bios. tried reverting back with no luck so i dont think the flash was the real issue. system runs fine for about an hour and crashes with fah running. tried prime95 and it ran an hour before both instances reported failures with in 6 minutes of eachother. have run memtest for 2 passes and it says good. going to run a full 24 hour memtest this weekend tho and see what happens. so im thinkin either mobo or cpu. unfortunately tho, i dont have any hardware that i can test the cpu in. i was wondering if anyone else can suggest anything that i can try? is it wishful thinking that prime failures for sure mean bad cpu? i just dont have a lot of money to sink into this. thanks.

Comments

  • lordbeanlordbean Ontario, Canada
    edited August 2009
    Prime95 failures can also mean data corruption through the RAM or north bridge. Try removing 2 sticks of RAM (aka run on 4GB) and then see if it works any better. A lot of motherboards have trouble running with 8GB RAM (especially when it's high-speed RAM).

    Btw, memtest doesn't stress north bridge stability as much as prime95 blend does. It can appear stable in memtest and still fail in windows, because prime95 blend mode makes the CPU and RAM work hard, and pass lots of data through the NB. Memtest86+ just sends commands to the RAM.

    If removing 4GB of your RAM makes it stable, you have two choices... you can either choose to stay with 4GB, or you can choose to put the other 4GB back in and try removing the divider. I can't provide any empirical evidence (but I'm sure Thrax will back me up here), but using a divider to run your RAM faster offers little to no performance gain on any Core 2-based platform. You'd probably find it more valuable to run 8GB 1:1 than to run 4GB 5:4.
  • test_tube_tonytest_tube_tony Dallas TX Member
    edited August 2009
    ok. i pulled 2 sticks. running a blend test this time. was doing the large fft test. 3 hours and so far so good. noticed another anomaly, tried running the half life 2 lost coast video test for the fun of it. it looked good for the first run, but the 2nd and 3rd time some reflective surfaces where purple-ish.
  • lordbeanlordbean Ontario, Canada
    edited August 2009
    All I can say at this point is keep testing and see what happens... if you get a failure in Prime95, it could be you're still slightly unstable, if not, maybe your graphics card is simply running a little hot (in regards to the purple surfaces).

    Good luck :)
  • test_tube_tonytest_tube_tony Dallas TX Member
    edited September 2009
    just for an update, stable right now at ddr2 800 with the spp overvolted to 1.45. working with the people on the evga forum right now to try and pull off the 1066. there are people that are running the same hardware at 1066 just fine. its just a very fussy board with 8 gigs.
  • lordbeanlordbean Ontario, Canada
    edited September 2009
    That's fairly normal. I've heard of this before, and seen it with my own PC as well... when running 4 DIMMs on your motherboard, maximum achievable FSB is quite a bit lower.
Sign In or Register to comment.