Announcement

Collapse
No announcement yet.

Downclock 4 x Ripjaws V F4-4000C18-8GVK

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

  • Downclock 4 x Ripjaws V F4-4000C18-8GVK

    Hi everyone,
    I bought 2 kits of 16Gb DDR4 (F4-4000C18D-16GVK) at a nice price on AZ Warehouse to - temporarily - replace a perfectly working AEGIS kit (F4-3000C16D-16GISB) on my ASUS MAXIMUS HERO VIII (Z170 chipset, which claims to support up to 3800MHz XMP) equiped with i7-7700.
    Of course they won't boot at 4000 XMP so, after some test, I've succesfully passed 2x MemTest (v9.4 b1000) cycles @ 3200 / 16-18-18-38 (same AEGIS XMP timings).

    I would like to squeeze the maximum speed (note that I don't care much about speed but about latency, due do my DAW/NLE applications) out of these RAMs until I'll renew the whole platform (next will be i5 12500 on z690), but I don't want to overvolt the (CPU) memory controller.

    Any suggestion ?
    Last edited by forart.eu; 06-22-2022, 09:28 PM.
    »Forward Agency NPO
    In progress we (always) trust.

  • #2
    Try DDR4-3600 or a lower value until the system can be stable, then you can leave it at that.

    Comment


    • #3
      Originally posted by GSKILL TECH View Post
      Try DDR4-3600 or a lower value until the system can be stable, then you can leave it at that.
      As said, I don't care mutch about frequency, but I need lower timings.

      Are there any recommended combinations, or I have to choose between JEDEC/XMP ones only ?
      Last edited by forart.eu; 06-22-2022, 09:28 PM.
      »Forward Agency NPO
      In progress we (always) trust.

      Comment


      • #4
        Sure, you can lower and optimize timings once you find the highest stable DRAM Frequency. You don't want frequency to be lower than what the system is natively capable of.

        Comment


        • #5
          I've found this post that gives some interesting infos about a similar kit underclocking on a Gigabyte z690 board:
          I've got this same motherboard on bios F5a with a 12600k and my G.Skill Ripjaws 4000mhz CL18 ( F4-4000C18D-32GVK ) worked out of the box with XMP, no issues at all. Also recently underclocked the same kit to 3600 16-20-20-39 to get it to run gear 1 and this setup also seems stable for now. No guarantees on that yet though since I haven't done proper testing.
          Hope that helps.
          »Forward Agency NPO
          In progress we (always) trust.

          Comment


          • #6
            Wow, another very interesting find:

            Originally posted by amit_talkin
            Here is my XMP for F4-4000C18D-16GVK kit. Which originally had only 4000MHz XMP profile.With 4000MHz XMP my fabric clock was running at 1000 mhz. so I tuned and created new 3600MHz profile.

            Click image for larger version

Name:	xmp-jpg.2540304.jpg
Views:	367
Size:	513.3 KB
ID:	168358
            »Forward Agency NPO
            In progress we (always) trust.

            Comment


            • #7
              OK, 1st feedback: the 2 x F4-4000C18D-16GVK kits works without any issues @ XMP 4000 on the Gigabyte Z690 UD DDR4 v1.0 (upgraded to latest F20a BIOS) + Intel® Core™ i5-12600 CPU.

              Geekbench v5 results

              (live usb) Linux Mint: Gigabyte Technology Co., Ltd. Z690 UD DDR4 - Geekbench Browser

              (NvME) Windows 11: Gigabyte Technology Co., Ltd. Z690 UD DDR4 - Geekbench Browser


              Then we "played" "played" a bit RAM settings in order to optimize latency and here's results:
              MHz
              2000​
              1600​
              1700​
              1800​
              1800​
              1800​
              Gear
              2​
              1​
              1​
              1​
              1​
              1​
              CR
              1​
              2​
              2​
              2​
              2​
              1
              CL-RCD-RP-RAS
              18-22-22-42​
              18-22-22-42​
              18-22-22-42​
              18-22-22-42​
              16-20-20-39
              16-20-20-39
              Read (MB/s) ¹
              54876​
              47950​
              49034​
              51625​
              52063​
              52016​
              Write (MB/s) ¹
              56813​
              47403​
              50597​
              53658​
              53584​
              53730​
              Copy (MB/s) ¹
              54785​
              47197​
              49909​
              52497​
              52962​
              53324​
              Latency (ns) ¹
              75.2​
              66.8​
              63.6​
              61.1​
              58.6​
              58.0​
              Read ²
              1111.29​
              955.25​
              1012.95​
              1071.78​
              1081.04​
              1077.83​
              Write ²
              1092.18​
              914.08​
              966.84​
              1027.45​
              1031.21​
              1026.48​
              Combined ²
              1145.19​
              978.53​
              1040.67​
              1101.66​
              1117.48​
              1104.75​
              Memory Mark ³
              3658​
              3727​
              3860​
              3934​
              4025​
              4018​
              Read Uncached ³
              25297​
              25180​
              26509​
              27442​
              27515​
              27927​
              Read Cached ³
              36286​
              36222​
              36243​
              35829​
              36181​
              34388​
              Write ³
              18681​
              17134​
              17945​
              19101​
              19478​
              19307​
              Latency ³
              41​
              37​
              35​
              35​
              33​
              33​
              Threaded ³
              53093​
              44600​
              47159​
              49587​
              50104​
              50141​
              Database Operations ³
              6666​
              6466​
              6854​
              7015​
              6945​
              6941​
              ¹ = AIDA64 v6.75
              ² = OCCT v11.0.11
              ³ = PerformanceTest 10.2 x86 Build 1008

              Of course now Geekbench has lowered, so now questions are:
              1. what's the best setting for A/V editing usage in your opinion?
              2. does CR1 really improves ALL performances?
              3. can you suggest better settings?
              Thanks in advance to all that can/would help.
              »Forward Agency NPO
              In progress we (always) trust.

              Comment


              • #8
                1. Highest frequency with lowest timings while staying within system temperatures ranges and stability.

                2. It can but at some point it needs to be increased to achieve stability, similar to any other timing.

                3. That is near system capability so anything more you will need to fine tune to see what you can squeeze out of it.
                Last edited by GSKILL TECH; 08-10-2022, 09:05 AM.

                Comment


                • #9
                  Thanks for your replies.

                  Last question: can other settings (RC-RFC1-RFC2-RFC4-RRDL-RRDS-CCDL-FAW) help to achieve better performances/latencies ?
                  »Forward Agency NPO
                  In progress we (always) trust.

                  Comment


                  • #10
                    Yes, however requires trial and error testing to confirm stability

                    Comment

                    Working...
                    X