Announcement

Collapse
No announcement yet.

F3-12800CL9D-8GBXL + Intel DZ68BC + Intel i5-2500

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

  • F3-12800CL9D-8GBXL + Intel DZ68BC + Intel i5-2500

    Hi,

    That RAM is listed on your site as "9-9-9-24-2N," but I can find nothing beyond that.

    For a few years, I've had one set (8GB total) running fine as seen in the "Active" column here. Until recently, I hadn't noticed that it was just on the default and that XMP wasn't in effect, so I was running at 11-11-11-28-1T (the 1T confirmed with CPU-Z).

    Having tried XMP today ("Proposed" column), 9-9-9-24-1T seems fine (1T again confirmed by CPU-Z), and as you'll see, tRASmin, tRFC, tRC, and especially tFAW are also different.

    I now have another set of the above RAM ready to install (bringing me to 16GB), but I understand from other posts that XMP will no longer be available or appropriate. The default way that I had always used with one set may work (11-11-11-28-1T), but I don't think that's ideal (not looking to OC but just get as close as possible to what the RAM is spec'd to do).

    How do the XMP numbers shown in the Proposed column mesh with the full specs of the RAM beyond just 9-9-9-24-2N? tFAW, for one, seems weirdly high at 50, though maybe there's a good reason for it.

    Side-question: Should the age difference of the two sets really matter?

    Thanks
    Attached Files
    Last edited by rseiler; 05-07-2015, 11:35 AM.

  • #2
    In this case, age does matter. Manually input the proposed specifications to see how it works.

    Comment


    • #3
      I removed the old RAM and put the new RAM in its place, just to run like that for a day to see what's what.

      Despite the new set being the identical model to the old set, the numbers are different.

      Given that, when all four are in place which values should I input manually to try to emulate what XMP would be? Even if it was on Automatic, how would it know which to use?

      tCL, tRCD, tRP, tRASmin, tRFC, tRRD, tWR, tWTR, tRTP, tRC, tFAW

      Old (Automatic): 11,11,11,28,128,5,12,6,6,39,24
      New (Automatic): 11,11,11,28,208,5,12,6,6,39,24

      Old (XMP): 9,9,9,24,130,5,12,6,6,40,50
      New (XMP): 9,9,9,24,15,5,12,6,6,33,24

      CR continues to be 1, at least for Automatic.

      tRFC is the one that seems the most strange. For Automatic, it almost doubled; for XMP, it dropped radically (you would think that it would also be relatively higher).

      Update: I think this thread might explain the general differences if not the particulars. I can't tell right now if the new set is single-sided, but the old one is double sided. So, I guess from the comment there, it should be OK despite the chip difference, though I wonder if it makes more sense to have the old or new set in Channel A for when the system is on Automatic.

      For Manual, it sounds like I should input the higher values (so, old XMP). Correct? Based on the phrasing in the screenshot, my board looks like it supports multiple XMP profiles when all 4 DIMMs are installed, so maybe it's OK just to choose the higher one that way.
      Attached Files
      Last edited by rseiler; 05-07-2015, 09:15 PM.

      Comment


      • #4
        Of the timing differences, manually input the higher values. tRFC can't be 15, you should probably set it to the maximum value with full slots.

        Comment


        • #5
          With all 4 slots populated, Auto worked fine but configured 2T for the first time, which I think is normal in this config (one pair was always 1T).

          So I then ran with this manually, and it's worked over a couple days:
          9,9,9,24,200,5,12,6,6,39,24 and 2T (leaving CR on Auto does not work now, it must be set).

          200 is the max tRFC settable, or I would have gone 208.

          If that sound fine to you, I'll stick with it.

          As you can see from the earlier screenshots, the board does support per-channel config, but I'm not sure if that's advised. About the only change that I would make, I think, would be the tRFC=130 for the older memory, which is in fact "2 ranks, 8 banks" vs the new memory being "1 rank, 8 banks." If it's worth bothering with that, please let me know. If it is, I suppose it would make sense (given that the first 8GB is used a lot more than the second) to put the old set back in Channel A, where for some reason the new set ended up.

          Comment


          • #6
            Sounds like it's good where it is. Enjoy!

            Comment


            • #7
              Alright, thanks.

              For future reference, what is your general guidance about per-channel settings? I didn't find much discussion of it here, maybe because it's not that common to have different RAM in each channel.

              Comment


              • #8
                You got it, in general it is difficult to get different RAM working together.

                Comment

                Working...
                X