I feel so dumb right now

Discussion in 'Desktop Hardware' started by Ethrem, May 18, 2016.

Thread Status:
Not open for further replies.
  1. Ethrem

    Ethrem Notebook Prophet

    Reputations:
    1,404
    Messages:
    6,706
    Likes Received:
    4,735
    Trophy Points:
    431
    So I built myself a great desktop back in 2014

    ASUS Maximus VI Formula
    i7-4790K @ 4.6GHz (auto overclock, the chip can handle 4.8GHz stable but the voltage is too high)
    32GB G.Skill Ripjaws X 1600MHz kit (F3-1600C9-8G)
    EVGA 780 Ti SC w/ ACX

    Well... Today it failed to POST saying the overclock failed. Now the system has had an issue since I got it where it turns itself off and back on when it boots and then works fine. Well I went in and looked at the settings and the voltage was set at 1.5v but I had manually set the memory to 1600MHz. Looking online I realized this memory is meant for 1.55v to be stable at 1600MHz but I realized something else... Its insanely overclockable. With a voltage bump to 1.65v, I just pushed these to 2133MHz 11-12-12-33 @ 1T. I never rally bothered with overclocking RAM before but I know Haswell loves memory bandwidth so I gave it a little nudge. Wow did that ever pay off! All this time ive been running these excellent Micron 8GB modules at a lowly 1600MHz without even the proper voltage (you would think the XMP profile would take care of that and it would if when I set it up I used XMP instead of manually setting it to 1600MHz). The 1T command rate surprised me as well, especially on chips made for 9-9-9-24 2T. So yeah I feel stupid now. All this time I thought it was a motherboard issue... And even dumber, my motherboard has a display that shows codes. I could have figured this out ages ago lol.

    I don't know why it decided to fail today but I'm glad it did. Even with the higher CAS, 2133MHz is much faster than 1600MHz, you can feel it in the whole system.

    Unfortunately it won't drop to 10. I haven't tried dropping the other values but I'm running memtest86 and so far it has no errors with these timings.
     
  2. bloodhawk

    bloodhawk Derailer of threads.

    Reputations:
    2,954
    Messages:
    5,851
    Likes Received:
    8,561
    Trophy Points:
    681
    Jesus CL9 @ 2133 !!!???? Those modules are ****ing amazing.
     
    TomJGX, Jarhead and Ethrem like this.
  3. Ethrem

    Ethrem Notebook Prophet

    Reputations:
    1,404
    Messages:
    6,706
    Likes Received:
    4,735
    Trophy Points:
    431
    No, I wish! Its CL11 @ 2133. I could probably pull off 9 @ 1866 but 2133 needs 11. I've thought about pumping the voltage on the IMC a bit and see if that makes 10 stable but honestly its not really all that important. There's definitely a performance gain even with the higher timings but the 1T command rate surprised me. Its a minimal gain (9/1600 = .0056 and 11/2133 = .0052) but it didn't cost me anything extra to get that gain, G.Skill just used really good Micron chips when they didn't have to - although maybe they did... 8GB modules back in 2013 when these were released were mostly CAS 10 or even 11 @ 1600.

    There's probably room to tweak some more, I just don't feel like fussing with it right now. Waiting for X/X2 HD Remaster to finish so I can see how stable these new 359.16 drivers are compared to the rest of the r364 branch which has all been trash.
     
    bloodhawk and hmscott like this.
Thread Status:
Not open for further replies.

Share This Page