Clevo Overclocker's Lounge

Discussion in 'Sager/Clevo Reviews & Owners' Lounges' started by Spartan, Mar 4, 2016.

  1. hissy

    hissy Notebook Geek

    Reputations:
    45
    Messages:
    83
    Likes Received:
    16
    Trophy Points:
    16
    Update:
    3400 is unreliable no matter what. If it boots, it works flawlessly at least within several hours of gaming and random activities, with not just normal, but even undervolted Uncore, but main issue here is IF it boots.
    On restart, it tends to stuck in reboot cycle for long time, ignoring Fn+D, until eventually it boots again. No voltage or timing tricks can change this.
    Still refuses to boot with 133 base clock and multiplier above 10.
    So highest reliable clock is 3100 (100x15 + QCLK Odd Ratio=Enabled) @1.3v, default Uncore, CL15: this at least boots without issues every time and still didn't die on me in windows. Tweaking timings left for tomorrow, it's night here.
    They are Hynix MFR btw.

    On temps, apparently under load everything inside the P775 chassis gets uniformly heated to the same temp as CPU and GPU. Poor ram is doomed to be scorched all the time.
     
    joluke likes this.
  2. jc_denton

    jc_denton V̖̟en̰g̻̼̰̩͙ea̲n̪c̭e̼ ͍̘̤͓̟̤Is̙͔ ̤Mi̻̭̣͎͍ͅn̰ḙ

    Reputations:
    6,544
    Messages:
    2,385
    Likes Received:
    4,061
    Trophy Points:
    281
    Last trick that worked for my p775tm1-g, was having the ram sticks in slots 1 and 4. Instead of having both slots in the two dimms under the bottom cover. They were more stable and achieved higher OC, you can also try if the same applies for 2 and 3.
     
    hmscott likes this.
  3. Meaker@Sager

    Meaker@Sager Company Representative

    Reputations:
    8,731
    Messages:
    54,446
    Likes Received:
    15,872
    Trophy Points:
    931
    The same here, there does seem to be less interference, likely socket to socket interference as sodimms are noiser than desktop slots.
     
    jc_denton likes this.
  4. hissy

    hissy Notebook Geek

    Reputations:
    45
    Messages:
    83
    Likes Received:
    16
    Trophy Points:
    16
    Oh forgot to mention. I've seen someone's post about exactly that, and this is why i've touched the ram again. They already in slots 1 and 4. Other combinations seem to work not better.
    In bottom slots they never boot at anything higher than 3000.
     
  5. Meaker@Sager

    Meaker@Sager Company Representative

    Reputations:
    8,731
    Messages:
    54,446
    Likes Received:
    15,872
    Trophy Points:
    931
    The machine is mostly trace limited for speed.
     
  6. hissy

    hissy Notebook Geek

    Reputations:
    45
    Messages:
    83
    Likes Received:
    16
    Trophy Points:
    16
    Inevitable cost of packing a high end computer into this small box.
    Well, finale here: just reverted everything. While it can run at 3100. higher timings make it slower than my last settings.
    I think it's indeed the board's traces.
    Initial goal with picking this ram was to have enough headroom for overclocking (actually it was one of the main goals for whole system, sadly the 2070 was a mistake), and i'm happy with results.
    As i know, exceeding 3000 mhz is uncommon for these chips, so i can say I squeezed everything from them, and apparently, from the platform as well. And that with very good timings.
    (more numbers: 1.30v in memory voltage, -117mv in core (and ring as they tied in bios), all other voltages unchanged)
    And it's faster than any high-speed sodimm ram available in local stores x)

    3036.png
     
    hmscott and jc_denton like this.
  7. Meaker@Sager

    Meaker@Sager Company Representative

    Reputations:
    8,731
    Messages:
    54,446
    Likes Received:
    15,872
    Trophy Points:
    931
    Also sodimms do not help as I said, especially when stacked.
     
  8. jc_denton

    jc_denton V̖̟en̰g̻̼̰̩͙ea̲n̪c̭e̼ ͍̘̤͓̟̤Is̙͔ ̤Mi̻̭̣͎͍ͅn̰ḙ

    Reputations:
    6,544
    Messages:
    2,385
    Likes Received:
    4,061
    Trophy Points:
    281
    I got inspired and tried to squeeze a little bit more out of my sticks.

    Before 3333Mhz at CL15-17-17-36 (276)
    Read 49128
    Write 51994
    Copy 49457
    Latency 42.2 ns
    Timings and subtimings:
    [​IMG]

    Tightened 3333Mhz at CL14-15-15-28 (276)
    Read: 50762
    Write: 52408
    Copy: 50335
    Latency: 39.1 ns
    Timings and subtimings:
    [​IMG]
    At 1.35v, stock VCCIO/VCCSA and +0 uncore. Passed MemTest 200% 0 errors.

    TimeSpy saw some gains at 5Ghz:
    CL15 12620
    CL14 12794

    14-14-14-28 might be possible without errors, if given more voltage.
     
    Papusan, joluke, jaybee83 and 2 others like this.
  9. Meaker@Sager

    Meaker@Sager Company Representative

    Reputations:
    8,731
    Messages:
    54,446
    Likes Received:
    15,872
    Trophy Points:
    931
    You might find errors start to creep in when pushing it though. A 5% latency improvement is nice.
     
  10. hissy

    hissy Notebook Geek

    Reputations:
    45
    Messages:
    83
    Likes Received:
    16
    Trophy Points:
    16
    Wow, that's how a good ram looks like.Pretty numbers.
    though i'm a little bit scared to push so much voltage. It's already hot enough in there under load
     
Loading...

Share This Page