1. OCAU Merchandise now available! Check out our 20th Anniversary Mugs, Classic Logo Shirts and much more! Discussion here.
    Dismiss Notice

Intel's Haswell-E...

Discussion in 'Intel x86 CPUs and chipsets' started by stmok, Jun 5, 2013.

  1. glnn_23

    glnn_23 Member

    Joined:
    Jan 19, 2012
    Messages:
    473
    Looking to try some new ram soon and have ordered some Ripjaws V to try.

    Have been very happy though with my Ripjaws 4 2400Mhz C15 16Gb that I bought when they first came out and pretty much settled on a stable 3200Mhz 16 17 17 43 1T @ 1.35v that passed any test I could throw at them.

    Thought I'd loosen timings right off today and went for a highest speed for a bit of fun using the 167 strap

    Ended up with 3540Mhz 17 19 19 46 2T @1.35v


    Click to view full size!


    http://valid.x86.fr/j1pz1i
     
    Last edited: Aug 17, 2015
  2. m0n4g3

    m0n4g3 Member

    Joined:
    Aug 5, 2009
    Messages:
    3,652
    Location:
    Perth, WA
    So i decided to try and get a lower cpu strap @ 100mhz, as i've seen some people say that the lower cpu strap is more efficient.

    Unfortunately.......... no bueno! Can't seem to get it to boot at 3000mhz ddr4 at all.... :(

    What are people doing to get bclk at 100 and ddr4 speeds < 2400?
     
  3. Supersize

    Supersize Member

    Joined:
    Jul 30, 2008
    Messages:
    1,832
    Location:
    Gold Coast QLD
    I know it's not 3000Mhz ram, but I'm running 2666Mhz Ripjaw's, this doesn't push the strap over 100mhz, though a lot of people saw all 2666mhz ram does.

    Have you tried resetting the BIOS and starting from scratch, first set the XMP profile on the ram and then lowering the CPU ratio to match.
    I'd go with lowering the ratio's so the CPU/cache clocks are default speeds and then try to boot.
     
  4. mshagg

    mshagg Politburo

    Joined:
    May 8, 2007
    Messages:
    23,089
    Location:
    Adelaide
    By more efficient, I suspect they mean you can use adaptive voltage, rather than being forced to use offset mode?

    Im only using a povo 2133 kit @ 2666. Raja suggested that intel didn't tune DDR4-3000 very well on the 100 strap and it typically requires a fair bit of tuning (system agent and uncore voltage are key).

    http://rog.asus.com/forum/showthread.php?53306-DDR4-3000-Setup-Guide-For-Rampage-V-Extreme
     
  5. m0n4g3

    m0n4g3 Member

    Joined:
    Aug 5, 2009
    Messages:
    3,652
    Location:
    Perth, WA
    Thanks guys, i'll give it a go when i get home.
     
  6. glnn_23

    glnn_23 Member

    Joined:
    Jan 19, 2012
    Messages:
    473
    Have you tried skipping 3000mhz and going up to 3200mhz. I have always found this is much easier. Also what are your main timings and voltage?
     
  7. m0n4g3

    m0n4g3 Member

    Joined:
    Aug 5, 2009
    Messages:
    3,652
    Location:
    Perth, WA
    timings are default for my kit.

    3000mhz 15-17-17-35-2T @ 1.35v
     
  8. OJR

    OJR Member

    Joined:
    Jan 19, 2013
    Messages:
    4,593
    Location:
    Melbourne
  9. m0n4g3

    m0n4g3 Member

    Joined:
    Aug 5, 2009
    Messages:
    3,652
    Location:
    Perth, WA
    What's a safe "max" voltage for cache voltage?

    Custom water loop with 2x 360 Rads, 5930k, and 2x 980ti's in the loop, and i think noctua NF-F12's running at full speed?
     
    Last edited: Aug 20, 2015
  10. mshagg

    mshagg Politburo

    Joined:
    May 8, 2007
    Messages:
    23,089
    Location:
    Adelaide
    The ROG guide suggests up to 1.45 but i think most people settle for 1.3v. If you have to push it harder than that i think your efforts are better directed at tuning the cores.
     
  11. m0n4g3

    m0n4g3 Member

    Joined:
    Aug 5, 2009
    Messages:
    3,652
    Location:
    Perth, WA
    ATM just testing stability. Settings are

    Bclk 100
    CPU Multi 45
    Core Multi 40
    DDR4-3200 16-17-17-43-2t
    Vcore 1.3v
    Vcache 1.45v
    DDR 1.4v
    VCCSA 1.216v (auto).

    So far 15mins aida64 stress test stable, max temp so far i've seen is 71C.

    Might bump up the cache multi up abit, but that's netting me much more balanced read/write/copys in AIDA

    60/61/65 respectively. (GB/s)
     
  12. m0n4g3

    m0n4g3 Member

    Joined:
    Aug 5, 2009
    Messages:
    3,652
    Location:
    Perth, WA
    Managed to push ram to 3200 15-17-17-35-2t on 1.375v, and dropped the cache voltage down to 1.25v @ 4000.

    Think i'll settle on this for my 24/7 o/c as it's only hitting 65C after 30 mins.

    Managed to get 4600 core bootable in windows @ 1.3v with cache at 4200 @ 1.45, but temps are getting abit higher 79C. Although i think i have a weak cache.... can't go higher than 4200 even @ 1.45v... :(

    This will probably be my benching speeds.

    Thanks for the help guys! :)
     
  13. glnn_23

    glnn_23 Member

    Joined:
    Jan 19, 2012
    Messages:
    473
    Started playing with the new Ripjaws V 3466Mhz 4 x 4gb on my R5E.

    XMP for 3466 uses the 167 strap so not much use really but managed to hit 3626mhz. Not stable at all.

    At 3400mhz however on the 125 strap appears to be HCI memtest stable 16 18 18 44 2T @ 1.385v


    Click to view full size!


    Also ran a couple of the memory presets on R5E for Samsung 1 sided 4 x 4 and will run C14 3200 and C15 3300 in Aida 64 Benchmark


    Click to view full size!



    Click to view full size!
     
  14. m0n4g3

    m0n4g3 Member

    Joined:
    Aug 5, 2009
    Messages:
    3,652
    Location:
    Perth, WA
    Nice numbers on that ddr man.

    I might try pushing this corsair abit more, seems like it's handling 3200 at stock timings really well.
     
  15. Supersize

    Supersize Member

    Joined:
    Jul 30, 2008
    Messages:
    1,832
    Location:
    Gold Coast QLD
    I have a strange issue, I'm running a 4.2Ghz overclock with 3.3Ghz on the cache.
    Now AIDA64 has this sudden drop in clocks, CPU usage stays the same and the machine doesn't crash.
    When this happens it doesn't show on Hardware Monitor, that continues to show 4.2Ghz.

    When I ran OCCT I noticed the clocks dropped again, this was picked up by hardware monitor.
    Again, test continues, doesn't fail, machine doesn't crash at all.

    Could this actually be happening or could it be a senor issue, the machine will pass all tests, 2 hours of AIDA64 (CPU,FPU,Cache), 2 hours of OCCT Large data set, 2 hours of Realbench, 2 hours of Intel Extreme Tuning Utility stress test, and Cinebench.
    It only happens in the first 2 - 10 minutes of testing and then never happens again.
    Here's some screen shots so you can see what I'm talking about:

    [​IMG]

    [​IMG]
    [​IMG]
    [​IMG]
    [​IMG]
     
  16. glnn_23

    glnn_23 Member

    Joined:
    Jan 19, 2012
    Messages:
    473
    Playing with timings a bit more now with Ripjaws V 3466 on X99 and can run a few benches with ram 3400mhz @ C15 17 19 42 1T 1.385v.


    Click to view full size!
     
  17. Moochin_Bastard

    Moochin_Bastard Member

    Joined:
    Dec 3, 2003
    Messages:
    468
    Location:
    Stafford, Brisbane (4053)
    I've got a kit of 4x8gb g.skill 2400mhz running at 2666 at stock settings so far, im just wondering how much voltage is safe in these ram chips? 1.2v is standard for these chips.
    Running my 5820k @ 4ghz on 1.025v atm and wanna see how high i can get the ram.
     
    Last edited: Aug 23, 2015
  18. Supersize

    Supersize Member

    Joined:
    Jul 30, 2008
    Messages:
    1,832
    Location:
    Gold Coast QLD
    Wow man, is it stable, usually the stock 5820k voltage is between 1.055-1.065v,
    I've had to use 1.110v for a stable 4Ghz but that's with adaptive + offset, with a VCCIN of 1.900v.
    (see post above with the sudden bus speed drop, seems to be due to lack of voltage, yet tests pass..lol).
     
  19. glnn_23

    glnn_23 Member

    Joined:
    Jan 19, 2012
    Messages:
    473
    Not exactly sure but I tend to stay under 1.4v. I have used g.skill 2400 4x4gb ram since it came out and settled on a stable 3200mhz 16.17.17.43.1T at 1.35v. It will be a little more difficult with 8gb sticks though I imagine
     
  20. Moochin_Bastard

    Moochin_Bastard Member

    Joined:
    Dec 3, 2003
    Messages:
    468
    Location:
    Stafford, Brisbane (4053)
    Turns out I had to increase the volts to 1.035 on the cpu, the quick test I ran for stability wasn't long enough this time it's stable while priming. Will have a little go with the DDR now.

    Played with the cpu a bit more got 4.6ghz @ 1.275v.

    Would that sorta volts be fine for a 24/7 use?
     
    Last edited: Aug 24, 2015

Share This Page

Advertisement: