Announcement

Collapse
No announcement yet.

Looks like i did the 8x500 = 4GHz trick

Collapse
X
Collapse
 
  • Page of 1
  • Filter
  • Time
  • Show
Clear All
new posts
Previous template Next
  • #1

    Looks like i did the 8x500 = 4GHz trick

    I am currently almost 10 hours! (and counting) Orthos stable @ 8x500FSB = 4GHz .
    I really figured out the minimum voltages since setting one of them lower causes Orthos to crash.

    One step lower Vcore causes crash after 2 hours, anything else within an hour for sure.

    I have set 1.5v PLL, 1.34vTT with auto references, 1.3875vcore, 1.9v RAM,
    1.3v MCH.

    Setting a higher Voltage on my Dimm's (which are rated 2.1v) gives no extra stability and doesnt make me able to run a lower VTT and/or MCHvoltage.

    CPUz currently reads 1.312v under load all the time, and 1.344v IDLE.
    Highest read is in the BIOS which is 1.35x.

    With EIST and C1 state enabled for now i figured out it DOES lower the multiplier IDLE to 6 (making it clock back to 3GHz). But the voltage is no longer lowered when IDLE...

    Who agree's with me these voltages/clocks temps are 100% safe for 24/7 use?

    Thanks to OCZ Vendetta 2 the cpu hits max 55 degrees on a room temperature of 20 degrees (seems very ok to me).
    Might run 3.6Ghz in the summer when this room's temperature rises up to 30 degrees.

    I can try if i can get my memory timings a little tigher, by then my overclock is done. I will also show the bandwidth difference between 400fsb - 3.6ghz (tight timings) and 500 fsb - 4ghz (looser timings) in a benchmark later here!

    I can tell sPI takes me 11,669 seconds to calculate a million digits.
    It's not shown on the screenshot below since im still stresstesting (around 15 hours orthos stable will do for me i think).
    Attached Files
    Last edited by MadDoGz; 11-15-2008, 05:17 AM.
    Primairy system: GA-EP45-DS4 / E8400 C0 / 2x2GB OCZ ReaperX PC8000 / Sapphire HD5870 / Creative Xfi Xtrememusic PCI / Intel Postville 160GB for OS & Programs / Antec Neopower 650w / Antec 900 / Windows7x64pro RTM (build7600)
    Tags: None
  • #2
    Re: Looks like i did the 8x500 = 4GHz trick

    Trd 11? Why, 9 and 10 did not work? That is odd looking at your timings, you should be able to use a lower one. Maybe your lower ram voltage is what is not allowing that?

    Looks good though man! Nice, I know it can be pretty hard and time consuming to overclock and get stable! So good job man!!!

    Comment

    • #3
      Re: Looks like i did the 8x500 = 4GHz trick

      yes i will try to get my tRD to 9. Guess 8 won't be quitte achieveable?
      I will also try to get my tRAS lowered to 15.
      And maybe try to get the maximum out of the other advanced timings (this will take much time doing memtest!).

      EDIT: currently running Orthos blend again with tRD 9 and tRAS 15, nice add to the bandwidth again (mostly caused by tRD for clarity).
      Figured out it needed more volts than 1.9 to post tRD 9, currently made a big step of 0.1 since 2.0v wont harm these dimm's
      Last edited by MadDoGz; 11-15-2008, 06:05 AM.
      Primairy system: GA-EP45-DS4 / E8400 C0 / 2x2GB OCZ ReaperX PC8000 / Sapphire HD5870 / Creative Xfi Xtrememusic PCI / Intel Postville 160GB for OS & Programs / Antec Neopower 650w / Antec 900 / Windows7x64pro RTM (build7600)

      Comment

      • #4
        Re: Looks like i did the 8x500 = 4GHz trick

        Im still not stable. I ran small FTT's last night on Orthos for over 10 hours, without an error. Just started Blend test and after 2 hours if gave up an error again. Rounding error @ 640K FTT (thats cpu related, not RAM).
        Guess i need slightly more Vcore?..... im not willing to put much more in it.
        Strange it didnt pop out @ pure small FTT test.
        Primairy system: GA-EP45-DS4 / E8400 C0 / 2x2GB OCZ ReaperX PC8000 / Sapphire HD5870 / Creative Xfi Xtrememusic PCI / Intel Postville 160GB for OS & Programs / Antec Neopower 650w / Antec 900 / Windows7x64pro RTM (build7600)

        Comment

        • #5
          Re: Looks like i did the 8x500 = 4GHz trick

          So... How's your tests coming along?

          Comment

          • #6
            Re: Looks like i did the 8x500 = 4GHz trick

            Originally posted by Lsdmeasap View Post
            So... How's your tests coming along?
            Looks like the ram doesn't like tRAS 15 / tRP 9 at this speed.
            Switched tRAS back to 18 straight away (doesnt give any noticeable performance-decrease) and have tRP set to 10 for now... (tRP does affect performance tho).

            Didnt run memtest for long with the tRAS at 15 and tRP at 9... just like 3 hours. To bad it didnt even pick out a single error in that time (1 error is a no-go for me).

            Ill keep u informed with my latest findings.
            Primairy system: GA-EP45-DS4 / E8400 C0 / 2x2GB OCZ ReaperX PC8000 / Sapphire HD5870 / Creative Xfi Xtrememusic PCI / Intel Postville 160GB for OS & Programs / Antec Neopower 650w / Antec 900 / Windows7x64pro RTM (build7600)

            Comment

            • #7
              Re: Looks like i did the 8x500 = 4GHz trick

              tRP 9? Why so high?

              Comment

              • #8
                Re: Looks like i did the 8x500 = 4GHz trick

                Originally posted by Lsdmeasap View Post
                tRP 9? Why so high?
                Typo , that was meant tRD

                I'm nearly stable, my MCH just really needed 1.32v @ 400 FSB!
                I finished 8 hours blendtest @ Orthos (the MCHvoltage of 1.3v caused to faill this after 2 hours, Memtest didnt give error's, Small FTT's didnt crash after 10 hours... so had to be the MCH).

                Look what i found about P45 chipsets:
                http://www.reghardware.co.uk/2008/09/04/review_intel_p45_chipset/page5.html

                According to that info 1.32v requirement for 500fsb stable seems perfectly normal.
                I think that will really be ok here for 24/7 use since i have the DS4 (with heatpipes).
                Besides P45 is a 65nm chip(set) so it can take a bit more volts than a 45nm (CPU's) imo.

                After finishing the latest 8 hours Orthos blend (without an error!) i had these voltages:

                1.3750vcore in bios (~1.31v load), before bumping MCH voltage i've been testing @ 1.4v which was 1.36v idle @ cpuZ
                1.34vTT
                just 1.5v PLL ofc.
                1.32v MCH
                2.0v RAM (can probably go back to 1.9v again since it was really stable, will figure this out soon)

                So for the VTT and MCH that will be the minimum voltages, also think for the Vcore that is, which i am currently testing.
                After im assured this is stable with the lowest voltages i will see if i can get a tRD of 9 working on 1000mhz ram.
                Also test if all is stable without skews, i currently use both 50ps for cpu and mch (think i dont need them)

                question:
                Is it possible a MCH clock skew will make the difference in stability in tightening tRD from 10 to 9?
                Last edited by MadDoGz; 11-17-2008, 11:05 AM.
                Primairy system: GA-EP45-DS4 / E8400 C0 / 2x2GB OCZ ReaperX PC8000 / Sapphire HD5870 / Creative Xfi Xtrememusic PCI / Intel Postville 160GB for OS & Programs / Antec Neopower 650w / Antec 900 / Windows7x64pro RTM (build7600)

                Comment

                • #9
                  Re: Looks like i did the 8x500 = 4GHz trick

                  Ahh! So funny us overclockers are

                  Comment

                  Previous template Next
                  Working...
                  X