LinX the undisputed champion

Discussion in 'Intel x86 CPUs and chipsets' started by shane41, Dec 10, 2017.

  1. shane41

    shane41 Member

    Joined:
    Nov 10, 2008
    Messages:
    5,441
    Location:
    on the edge
    So many times I see members belting processors hours on end Prime95 for a stable Overclock. :tongue:
    This is an old program too heavy for testing.

    15 minutes anyone can dial in with LinX 0.64. http://www.mediafire.com/file/f1pbo2o3zmer500/LinX.zip
    ___________

    Why are we doing this? To see how low you can run your Vcore. But it's useful for other testing as well ...
    Your scores are in Gflops. If the averages are close your doing well. But if the score drops a BSOD imminent.
    ___________

    My system is example only. 7350K defaults. Win7 no sp1 installed.
    Set problem size 9 to start ... bring up to 18 problem size later but watch your temps. Usually 3x run
    23 problem size & beyond unnecessary testing.

    Pic's are stock. Then manual volts. Rest & Load. Each time lowering Vcore in Msi App.
    Enjoy :)

    [​IMG]

    https://www.mediafire.com/view/c8n7z8ezpa9i7xn/2 rest.jpg
    https://www.mediafire.com/view/owoc3ocvojmw5vp/3 load.jpg

    https://www.mediafire.com/view/sptiqyzasu7mict/4 rest.jpg
    https://www.mediafire.com/view/x1rri3wnsa5bmyk/5 load.jpg

    https://www.mediafire.com/file/qxfustox0l3gkiy/6.jpg
     
  2. aussie-revhead

    aussie-revhead Member

    Joined:
    Feb 21, 2005
    Messages:
    19,216
    Location:
    St Clair 2759
    How many loops do you do?
     
  3. OP
    OP
    shane41

    shane41 Member

    Joined:
    Nov 10, 2008
    Messages:
    5,441
    Location:
    on the edge
    I set it 3 runs Aussie. But I'm used to LinX. 5-10 wont matter ... it just takes longer each run as problem size is increased.

    My final testing usually results on 18 problem size 3x runs. But for eg. I set 1.2V for 5Ghz & it fell over @ 15 problem size.
    Then I increase to 1.25V. This is for low Volt testing as you guys want.

    But I'm cranking these machine the other way for max OC. Set 1.4V & wind up multi's till it falls over. That's a speed thing just for checking quick grunt & belt out a score.
     
  4. aussie-revhead

    aussie-revhead Member

    Joined:
    Feb 21, 2005
    Messages:
    19,216
    Location:
    St Clair 2759
    Yeah that's cool, I like to keep volts down when looking for stable clocks, benching clocks just need to produce a score.

    :)
     
  5. demiurge3141

    demiurge3141 Member

    Joined:
    Aug 19, 2005
    Messages:
    995
    Location:
    Melbourne 3073
    I've had oc fail beyond the 20th run, it's just like prime95, you have to run it for hours.
     
  6. OP
    OP
    shane41

    shane41 Member

    Joined:
    Nov 10, 2008
    Messages:
    5,441
    Location:
    on the edge
    View it as a quick dialing in on cpu multi's. See your BSOD & look it up. Usually low volts.
    Prime95 does not output a score. LinX will test the strength of your settings.

    eg. 7700K 50x 1.2v ... 51x 1.3v ... 52 1.4v

    No x1 stress test can set absolute best for you. Run your computer in that week, if BSOD then adjust.
    _________

    Overclocking I can get on an unknown board / processor. LinX 15 minutes & I have a rough idea on volts for each multi.
    Then I'll throw a motherload of benchmarks @ it.
     
  7. aussie-revhead

    aussie-revhead Member

    Joined:
    Feb 21, 2005
    Messages:
    19,216
    Location:
    St Clair 2759
    I just did three 18k runs and I noticed as I reduce my vcore, my Tflops are also dropping .... it still passes so what is that?

    :)
     
  8. OP
    OP
    shane41

    shane41 Member

    Joined:
    Nov 10, 2008
    Messages:
    5,441
    Location:
    on the edge
    Ah good :thumbup: That means your processor has less strength as you reduce volts.

    So you could search for ideal volts now where the Gflops are strongest. eg say between 1.25 & 1.3v but there was no extra benefit
    when you pushed beyond 1.285v Then you can move on & try sub volts IMC & tune memories as well.

    Combine all these ideas when you do your Cinebench runs to find improved scores.
    __________

    Low scores I see all the time. Result is brute force OR cpu starving of energy. Testing is faster & you'll be the one to benefit.
     
  9. aussie-revhead

    aussie-revhead Member

    Joined:
    Feb 21, 2005
    Messages:
    19,216
    Location:
    St Clair 2759
    I dropped further and LinX found an error in 39 seconds, but no bsod.

    I pumped it back up and output is up, so I will test higher and see what it does.

    :)
     
    warcraft likes this.
  10. Deano_20802

    Deano_20802 Member

    Joined:
    Apr 21, 2009
    Messages:
    288
    Location:
    Tas
    I first noticed this years back when I moved to IBT (LinX based) as my quick dial in program. It's still in my list of test app's today!

    I've also had errors well past 20 runs, but for quick initial dial ins it's useful. I still run all my tests for days when I think I'm 24/7 stable, just to be sure.
     
  11. terrastrife

    terrastrife Member

    Joined:
    Jun 2, 2006
    Messages:
    18,207
    Location:
    ADL/SA The Monopoly State
    Linx stresses memory and avx, prime small fft for temperature stressing :p
     
  12. LinX

    LinX Member

    Joined:
    Jan 17, 2002
    Messages:
    509
    Location:
    Digital Nomad .. I go where my whimsy takes me.
    Program is named after me and I'm not even the author :lol:
     

Share This Page