1. Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Low RAM DDR3 2400 MHz TridentX performance

Discussion in 'Overclocking & Cooling' started by mandragora665, Jul 16, 2017.

  1. mandragora665 New Member

    Joined:
    Jul 16, 2017
    Messages:
    5 (0.05/day)
    Thanks Received:
    0
    Hi, I have problem with low RAM write speed on AIDA64 benchmark.

    My platform is: i5-4690k@4.7 GHz, cache 4.2 GHz Asrock Z87 Extreme3 2x8 GB G. Skill TridentX 2400 CL10 (double sided) Windows 10 Pro

    Ram is in double channel configuration with XMP profile active. Before TridentX I had Kingston 1333 CL9 sticks. After upgrade I see massive boost in WinRAR benchmark (~4900 kb/s to almost 7200 kb/s), boost in fps in games, faster rendering and video compression. RAM read tests give me almost 36000 MB/s, but RAM write tests give me only ~21500 MB/s. Shouldn't it be much higher?

    I've tried CPU stock clock setting, manual timings configuration, changing RAM 1T/2T and nothing helped.

    Motherboard have newest BIOS, Ram are in A2/B2 slots, with slots A1/A2 no difference.

    Lowering RAM frequency to 2133/1866 impacts on read speed but write speed doesn't change.
     

    Attached Files:

    • 2.jpg
      2.jpg
      File size:
      88.4 KB
      Views:
      37
  2. sneekypeet

    sneekypeet Unpaid Babysitter Staff Member

    Joined:
    Apr 12, 2006
    Messages:
    24,436 (5.81/day)
    Thanks Received:
    9,317
    The write and copy results are about what is expected from that kit. The 36000 read is way above what any of the reviews show. Of course most of the reviews have the CPU at stock levels too.
     
    EarthDog says thanks.
    10 Year Member at TPU
  3. mandragora665 New Member

    Joined:
    Jul 16, 2017
    Messages:
    5 (0.05/day)
    Thanks Received:
    0
    Reviews with Haswell CPU shows write speed similar to read speed, I have much lower score.

    For example: https://www.purepc.pl/pamieci_ram/j...re_i5_haswell_test_ddr3_13332400_mhz?page=0,4 (MemCopy)

    http://forum.pclab.pl/topic/1014134-GSkill-TridentX-DDR3-2x4GB-2400MHz-CL10/ (screenshots attached)
     
  4. sneekypeet

    sneekypeet Unpaid Babysitter Staff Member

    Joined:
    Apr 12, 2006
    Messages:
    24,436 (5.81/day)
    Thanks Received:
    9,317
    The first link does not have your kit in the chart. The second link, I see the guy in the first few posts is using a Z97 motherboard. You need to find charts that use similar components to base your opinion off of.

    Also, have you tried to update the bios? Have you cleared the CMOS after removing the original RAM. There are a few things that could be playing up and causing weird issues with the results. Personally, I would use the default CPU speed and settings, set XMP, reboot and bench. This way you can look at reviews and see where you stand against the masses.
     
    10 Year Member at TPU
  5. vega22

    vega22

    Joined:
    Jul 18, 2007
    Messages:
    2,425 (0.65/day)
    Thanks Received:
    653
    turned virtualisation off?

    i would guess the timings you have are not efficient and causing the issue but it could be a bios issue too. sneeky has covered what to do there :)

    this is my older set @2400mhz to give you an idea of where you should be.

    https://www.techpowerup.com/forums/attachments/aida-jpg.52379/
     
    10 Year Member at TPU
  6. mandragora665 New Member

    Joined:
    Jul 16, 2017
    Messages:
    5 (0.05/day)
    Thanks Received:
    0
    I'll try turn off virtualisation. I've cleared CMOS after upgrade of course.

    I have the same score with Virtualization off. I'm afraid I have to contact with Asrock, maybe they can help with another BIOS or sth.
     
    Last edited by a moderator: Jul 16, 2017
  7. vega22

    vega22

    Joined:
    Jul 18, 2007
    Messages:
    2,425 (0.65/day)
    Thanks Received:
    653
    my cache might be clocked upto x45 too in that, but the impact of that is not that great.

    i would look into your timings and see if running them cl11 with tighter subtimings helps. i think it did for me :D
     
    10 Year Member at TPU
  8. mandragora665 New Member

    Joined:
    Jul 16, 2017
    Messages:
    5 (0.05/day)
    Thanks Received:
    0
    Higher timings makes my read score lower but write stays the same :/
     
  9. eidairaman1

    eidairaman1 The Exiled Airman

    Joined:
    Jul 2, 2007
    Messages:
    18,503 (4.92/day)
    Thanks Received:
    4,502
    If you noticed gains, the job of the ram you bought accomplished that. Synth benchmarks are not realworld.
     
    10 Year Member at TPU
  10. mandragora665 New Member

    Joined:
    Jul 16, 2017
    Messages:
    5 (0.05/day)
    Thanks Received:
    0
    I agree, real world performance is the key, but I'm curious what happens to AIDA benchmark.
     
  11. eidairaman1

    eidairaman1 The Exiled Airman

    Joined:
    Jul 2, 2007
    Messages:
    18,503 (4.92/day)
    Thanks Received:
    4,502
    Aida only gauges theorhetical performance.

    @cdawall @OneMoar
     
    10 Year Member at TPU

Currently Active Users Viewing This Thread: 1 (0 members and 1 guest)