Announcement

Collapse
No announcement yet.

F4-3000c14q-64gtz, x299

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • F4-3000c14q-64gtz, x299

    Hi everbody!

    I've sent this message to the European support mail yesterday while waiting to get activated. As on the forum more people are able to give me some pointers, I'm posting it here too.

    I bought the ram kit F4-3000C14Q-64GTZ for my new 7820X build (ASUS X299 TUF Mark I Mainboard, BIOS 0802).

    Sadly I am unable to get the ram to the rated speed, doesn't matter if I'm using XMP or not. So I am not sure whether one of the sticks is defective or sensitive to high/low voltages or temperature.

    The kit runs fine up to 2800MHz @ 14-14-14-34, everything higher and I get errors in HCI Memtest & GSAT. GSAT shows an error on the first read, the reread contains the expected data though to what kind of problem that points I don't know.

    I'd like to get the kit to run at the rated speed as it was quite expensive, so could you please help me?

    Thanks!

  • #2
    Which slots are you using?

    Test each module to see if they can individually pass 3000 C14.

    Comment


    • #3
      Thanks for you quick reply.

      I'm using A1, B1, C1 and D1.

      Not tested each module alone yet cause I'd have to get new thermal paste (NH-D15) but will try do it over the weekend.

      Comment


      • #4
        Didn't need new paste after all, with some fiddling I got the modules out w/o removing the cooler.

        All four modules separately passed a 90min GSAT without an error @3000MHz when before with the whole 64GB I couldn't pass this once.

        I then inserted all four again, this time in ascending order, so that two consecutively numbered modules where left/right.

        Then I started a three hour GSAT and got errors:
        Code:
        2017/09/30-14:09:07(DST) Report Error: miscompare : DIMM Unknown : 1 : 5974s
        2017/09/30-14:09:07(DST) Hardware Error: miscompare on CPU 9(0x200) at 0x7ef2f8d3efe0(0x0:DIMM Unknown): read:0x8000000000000000, reread:0x0000000000000000 expected:0x0000000000000000
        2017/09/30-14:10:24(DST) Report Error: miscompare : DIMM Unknown : 1 : 6051s
        2017/09/30-14:10:24(DST) Hardware Error: miscompare on CPU 13(0x2000) at 0x7efa7f432828(0x0:DIMM Unknown): read:0x1000000000000000, reread:0x1000000000000000 expected:0x0000000000000000
        2017/09/30-14:21:11(DST) Report Error: miscompare : DIMM Unknown : 1 : 6698s
        2017/09/30-14:21:11(DST) Hardware Error: miscompare on CPU 6(0x40) at 0x7ef754790828(0x0:DIMM Unknown): read:0x1000000000000000, reread:0x0000000000000000 expected:0x0000000000000000
        2017/09/30-15:00:03(DST) Report Error: miscompare : DIMM Unknown : 1 : 9030s
        2017/09/30-15:00:03(DST) Hardware Error: miscompare on CPU 6(0x40) at 0x7ef177eadee0(0x0:DIMM Unknown): read:0xfffffff7fffff7f7, reread:0xfffffff7fffffff7 expected:0xfffffff7fffffff7
        2017/09/30-15:30:08(DST) Stats: Completed: 588487104.00M in 10800.30s 54488.05MB/s, with 4 hardware incidents, 0 errors
        I continued tinkering (increasing RAS, changing DIMM voltage, etc) but couldn't pass the 3 hour test once - though quite often a single error occurs in the last 40 minutes or so (temps start to stabilize quite early, so don't think that matters).

        When I checked the SPD XMP setting I found that e.g. tRRD_L should be set to 8, but the BIOS only lets me enter up to 7 -- even in XMP mode that isn't set correctly right alongside t_RRDS & t_RRDD (SPD says 6, AUTO & XMP set this to 4) .

        Could that be a case of BIOS still unstable or do I have a hardware problem?

        Comment


        • #5
          Test 2 or 4 at a time to see if you can find a poorly working set.

          The results can give us an idea of whether a module is causing problems or settings.

          Comment


          • #6
            Tested 2 sticks in C1 D1 and afterwards the other 2 sticks in A1 and B1.

            Got errors on both tests, but more than 10 during the first one so that I didn't even finish the 3h run.

            Yesterday I also got a replacement kit from my shop, tested all 4 sticks of this @3000 and got errors, no fault again @2800.

            So I guess the settings have to be changed?

            Comment


            • #7
              What test and version are you using?

              Have you tried boosting DRAM Voltage or SA Voltage to see if that may help?
              Last edited by GSKILL TECH; 10-04-2017, 09:16 AM.

              Comment


              • #8
                Yes, I've tried boosting the DRAM voltage, above 1.39 the faults come really quick.

                I'm using Googles' stressapptest at the moment, version 1.0.7
                Before I used HCI Memtest Pro 6.0 but with 64GB & 16 threads that doesn't work out well.

                Could you clarify what you mean by SoC? SA & IO or something else? And if possible give me starting values cause I don't think 0.765 for SA & 1.0 for IO are (that's what auto gives me) correct. I tried different values already before posting but didn't get to a situation where I passed the test.

                This are the voltages with auto & DRAM set to 1.350. The VTT voltage is reported wrong (in SIV & TurboV too) and set to DRAM voltage / 2.


                Timings (Auto) for 2800:


                Timings (Auto) for 3000 (screen grab comes from testing with 2 sticks only), faults at 2T too:

                Comment


                • #9
                  Try increasing SA Voltage to around 1.0V or 1.10V to see if it works more stable. Then you can fine tune from there.

                  Comment


                  • #10
                    Thanks - tried to get it stable with these voltages, but still getting one error in a 3 hour run.

                    But much better than upping the IO and DIMM voltages, that just led to an extreme amount of errors.

                    I probably have to accept 2800 as max or do you have any other ideas?

                    Comment


                    • #11
                      Yes, if it is perfectly stable, you can leave it at that.

                      Comment

                      Working...
                      X