Announcement

Collapse
No announcement yet.

14900CL9D - Failing Prime 95 Blend on Gigabyte MOBO

Collapse
X
Collapse
 
  • Page of 1
  • Filter
  • Time
  • Show
Clear All
new posts
Previous template Next
  • #1

    14900CL9D - Failing Prime 95 Blend on Gigabyte MOBO

    Hey there,

    I recently built a new computer, including 4 sticks of F3-14900CL9D-8GBSR memory. Thing is, my computer can not pass Blend in Prime 95 longer then 20 minutes or so being an error occurs. I have it set to XMP Profile 1 in the BIOS, which automatically sets it to the 1866 speeds... And it runs fine it seems, but why does it keep failing Prime?

    Thing is, Prime will run hours on small FFT or large FFT, which rules on the CPU for me. Blend can fail minutes in to about 20-30 minutes. One time I got a BSOD running the 1600 "stock" speeds. Seems like I have a memory issue of some sort. What settings should I play with in the BIOS to stabilize this RAM at 1866? It is also set to "Turbo" out of the "Normal" and "Extreme" settings available... If that helps any (default).

    Specs:

    Windows 7 64bit Professional OS
    Gigabyte GA-Z68XP-UD3P
    Intel 2600K i7
    4x4GB G.SKILL Sniper14900CL9D
    128GB Crucial SSD + 1.5TB Seagate HDD
    1000W Cooler Master Silent Pro Gold PSU


    Appreciate any help...

    -Steven
    Tags: None
  • #2
    Try manual settings, XMP may set some advanced timings too tight for full slots.

    Also, make sure each kit is in the same colored slots.

    Thank you
    GSKILL TECH

    Comment

    • #3
      I did try doing it manually as well, setting the speed to 1866 and switching voltages up through 1.65V with no change in reliability. Even at stock settings (reset bios) the Blend test failed. The other two, however, pass just fine. Which makes me look at the RAM.

      What other settings should I try to make it more stable? Both sets of RAM were also installed in the same color slots so they run together.

      Comment

      • #4
        Might try the sticks at 1.6 and raise the QPI/VTT + .05 (prob about 1.05 currently?)


        Pls offer comments on support I provide, HERE, in order to help me do a better job here:

        Tman

        Comment

        • #5
          Ill try raising that QPI/VTT setting when I get home and see if that helps any. I have tried running the RAM at 1.65V already with no luck.

          I have noticed no issues with day-to-day use with the RAM, only failing stress test after 20mins or so. Will this ever cause any issue in real-life use or just something that Prime does with it? If it will never crash the computer or anything like that I might not worry so much about it, but I just want to make sure its stable before I go and push my CPU any more. It always shows a rounding error or something when it shuts down the worker, rounding was .05 and not .04 or something it says. Don't remember exactly.

          And like I said, Small and Large FFT's run with no failing at all when I run them for an hour or so. But blend only makes it around 15-25 minutes before I get the rounding error hardware failure notice.
          Last edited by Invertalon; 08-22-2011, 09:18 AM.

          Comment

          • #6
            QPI/VVT voltage was at 1.05... Raised it to 1.085. Failed within 15 minutes. Still upping it little by little to see if any improvement.

            What is my *safe* limit as far as this voltage goes? I have done quick google searches and got numbers as high of 1.35V?
            Last edited by Invertalon; 08-22-2011, 11:51 AM.

            Comment

            • #7
              I have my voltage set to 1.18V right now and the test has been running longer then it ever has. Running about 40 minutes now without any issues or Blue Screens... I got one that was a 0x124 so I raised the voltage more... I then got a 0x7F fault so I raised it higher to just under 1.2V... So far, no issues.

              Wish I didn't have to do all these adjustments just to run the RAM as specified (stock) but what can you do. I will let you know how the testing goes. Probably run Prime again tonight for another hour and see if it will pass it for more then 30 mins.
              Last edited by Invertalon; 08-22-2011, 02:52 PM.

              Comment

              • #8
                I ran Memtest last night and got 16 errors.

                It always failed Test 8:

                Test 8 [Moving inversions, 32 bit pat]

                This is a variation of the moving inversions algorithm that shifts the data pattern left one bit for each successive address. The starting bit position is shifted left for each pass. To use all possible data patterns 32 passes are required. This test is quite effective at detecting data sensitive errors but the execution time is long.

                It failed Test #8 on passes 1, 7 and 8. Now I have to isolate the bad RAM.
                Last edited by Invertalon; 08-23-2011, 03:29 AM.

                Comment

                • #9
                  After I got home from work today I decided to isolate the RAM...

                  First, I ran Memtest again and I got the errors again during same test. So I started to find out which was causing me problems.

                  I tested stick 1 (far left) and it passed entire test... Same with stick 2 and 3, all in the same slot... Ran my final RAM module, and it failed.

                  So I put the other three good sticks back in and ran Memtest 3 passes, no faults or errors. Looks like I found the bad cookie.

                  So now to contact Gskill for a replacement/RMA...

                  Comment

                  Previous template Next
                  Working...
                  X