Announcement

Collapse
No announcement yet.

Questions about F4-3466C16Q-32GTZR

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

  • Questions about F4-3466C16Q-32GTZR

    Hi,

    First, maybe i need to tell about my rig :

    I9 9900k 5ghz @1.25 (stable on occt)
    2080ti FTW3
    H500m Case
    PSU : corsair rmi 850


    I have this kit of 32Gb Ram and I have few questions :

    BDIE : I havebeen told different things about if it's bdie or not. In typhoon it is but i've been told it can't be sure 100%
    Also,

    Overclock : Once again some told me it was great gain to OC and other told be that above 3200 it's quite useless especially with an Intel CPU. What are your thoughts about it?

    I tried to do an OC with the help of someone but it does not seem that stable here what I got but i had a crash in game so ...

    Temperature : My ram seems to get quite hot : 37 to 38c idle (my ambiants are around 29c those days, hot weather in paris) and up to 48 to 52c in game (red dead redemtpion for instance) is that normal? Thanks for the help !



  • #2
    You can verify the ICs in different ways and overclocking the modules is one of them. There is not really another 8Gbit IC that can hit 3733 at a tRFC this low, so B-Die is already confirmed from that alone. Even very good Micron 8Gbit E-Die or Hynix CJR/DJR would need 500+ for that.

    For my use cases (2D, EBV) memory OC is actually pretty relevant and helps with speeding up decompression of the RAW images of my camera and many of my very frequently used photoshop actions. There is lots of other stuff where the addtional bandwidth and lowered overall latency can help, though. Even some competitive games show significant gains in minimum frame rates as long the settings are not GPU limited.

    Low fifties should still be ok for temperatues, esp if there is no direct airflow from a CPU cooler but only from case/rad fans. If you ever run into stability issues and want to confirm, if they are temperature caused, just give the modules some additional airflow and test if it helped.
    Team HardwareLUXX | Show off your G.SKILL products!

    Comment


    • #3
      Well even before oc I had the same temps kinda and it was fine indeed.

      But I didn't managed to get a stable oc, tried 3733 (1.4 voltage) and 4000(1.45) with different timings but I had crashes in game.

      Also couldn't get less latency than 45 even at 4000

      Comment


      • #4
        I would recommend stability testing your memory overclock before actually using the system. Even a single timing being off can cause crashes and if you change them in bulk it is hard to determine which one needs to be adjusted for stability. In many cases adjusting the DRAM voltage is not the answer, even though voltage can sometimes compensate for wacky timings.

        Latency measurement in Aida64 is very sensitive to any kind of disturbance from other software and background tasks. So if your OS installation is older and pretty loaded with stuff, that can result in higher latency values. Because of that many poeple prefer to run Aida64 memory benchmark in safe mode instead.

        Here is my old and pretty moderate DDR4-4000 setting, also with a 9900K @ 5GHz:

        Click image for larger version

Name:	G.SKILL_F4-400C19-32GTRS_4000c16.png
Views:	391
Size:	115.4 KB
ID:	164551
        Team HardwareLUXX | Show off your G.SKILL products!

        Comment


        • #5
          it's so time consuming, overclocking CPU is kindergarden overclock compared to it

          if i have to test every timings by one for hours i think i'll have to do that H24 for a week... I think i'll call it a day and leave it at 3466 honestly... Compared to 4000 what is the best i could expect in terms of FPS for instance?

          Comment


          • #6
            Yeah, CPU overclocking is prettty simple and boring af in general^^

            I don't play any 3D games, so no ingame numbers from me. However you can compare performance yourself with many games and their integrated benchmarks. Forza Horizon 4, Shadow of the Tomb Rider and Ashes of Singularity Escalation get commonly used for that in addition to one of the Battlefield titles (don't remember which one, though). The impact of memory overclocing can vary a bit. Sometimes the gains are already visible from the average FPS alone, with others you need to look at the minimum FPS or frame times instead.
            Team HardwareLUXX | Show off your G.SKILL products!

            Comment


            • #7
              Hm well i guess i'm not patient enough haha...

              but i suppose i'm fine with those already !

              For my cpu yeah i managed to get it "stable" at least on occt avx2 / small without avx

              9900k, 1.24vcore, turbo LLc
              Io sa 1.15
              no avx offset

              That's pretty ok i guess

              Comment


              • #8
                Is there any point of lowering TRFC alone?

                i found this tab with impact of timings and i saw that my ram XMP is 607 trfc but i can go way lower, if i leave all on auto but set TRFC at 300 for instance, could it still be a boost, without touching frequency or voltage?.
                Last edited by Elric; 06-03-2020, 05:56 AM.

                Comment


                • #9
                  Yes, it does. At DDR4-3466 a tRFC of 300 should be pretty easy for B-Die. Some of the other timings should also have plenty of room to tighten them down.
                  Team HardwareLUXX | Show off your G.SKILL products!

                  Comment


                  • #10
                    Which one should I lower you think without big issues?

                    Comment


                    • #11
                      Almost all of them really. How much each one can be tightened, you would have to try for yourself though. There really is no shortcut to memory OC, so unless you already know very well what you are doing, most of the process is step by step and one by one.

                      This is probably the closest to your XMP if have ever tested (underclocked):

                      Click image for larger version

Name:	GSK_3466c16.png
Views:	339
Size:	109.6 KB
ID:	164569
                      However the same kit and system was also able to this (XMP frequency, tweaked timings):

                      Click image for larger version

Name:	GSK_4000c16.png
Views:	355
Size:	109.3 KB
ID:	164568
                      and even max out the platform at DDR4-4133.
                      Team HardwareLUXX | Show off your G.SKILL products!

                      Comment


                      • #12
                        Should I try those? I have 4x8 too don't know it that has a big impact

                        Comment


                        • #13
                          1:1 copying values from other systems usually does not work, even with very similar configurations.

                          You can however use these (or others) as a rough guideline and step by step work your way towards them.

                          Edit: Just leave the command rate at 2T.
                          Team HardwareLUXX | Show off your G.SKILL products!

                          Comment

                          Working...
                          X