1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

CPU QE9650 prime 95 questions and overclocking

Discussion in 'Hardware' started by cookie! nom nom, 6 Jun 2020.

  1. cookie! nom nom

    cookie! nom nom Minimodder

    Joined:
    27 Apr 2012
    Posts:
    970
    Likes Received:
    86
    Hi everyone

    My work was getting rid of a load of old CAD computers and one of them was a p35 chipset.

    Now from what I understand the p35 wasn't an amazing chip for overclocking and same for the QE9650?

    I currently have a overclocking of 3.69ghz and seem stable although each time I adjust the voltage I always seem to have one core/worker out of balance in prime95. Is this normal? Currently I have 3 workers on 640k test 2 and one core/worker 672k test 1. If I adjust the voltage then it's one core runs slow the rest are all closely matched.

    Very confused

    Edit running win 10 64bit because that's the nearest thing to hand.
     
  2. Big Elf

    Big Elf Oh no! Not another f----ing elf!

    Joined:
    23 Apr 2009
    Posts:
    3,940
    Likes Received:
    595
    I've had CPU cores completing stages of Prime95 at different speeds although usually I think they were within a few seconds of each other.

    I only had one P35 board and the only thing I remember was massive vdroop. The P45 boards were generally a lot better.

    I had the non-extreme Q9650 which was a sweet little CPU that did 4GHz at around 1.27V

    What settings are you using for the FSB, Multiplier and vCore?

    My memory is a bit hazy of overclocking the early Core CPUs but I do have some settings I used for the Q9650.
     
  3. cookie! nom nom

    cookie! nom nom Minimodder

    Joined:
    27 Apr 2012
    Posts:
    970
    Likes Received:
    86
    Feb is 410 and CPU 9

    Fab strap 400
    Pcie 101

    Vcore 1.38125 (working down)

    CPU pll 1.6
    Fsb 1.26
    NB 1.37
    Sb1.2

    Loadline cal enabled

    CPU gtl ref 0.63

    NB gtl 0.60

    SB 1.5v 1.5v

    CPU spread disable

    Pcie spread disable
     
    Last edited: 6 Jun 2020
  4. cookie! nom nom

    cookie! nom nom Minimodder

    Joined:
    27 Apr 2012
    Posts:
    970
    Likes Received:
    86
  5. Big Elf

    Big Elf Oh no! Not another f----ing elf!

    Joined:
    23 Apr 2009
    Posts:
    3,940
    Likes Received:
    595
    vCore is way too high. I'd only go to around 1.55v under water and then only for benching. At 1.8v it won't last long.
     
  6. cookie! nom nom

    cookie! nom nom Minimodder

    Joined:
    27 Apr 2012
    Posts:
    970
    Likes Received:
    86
    Blast ! Sorry trying from my phone and it seems selective what keys it will output, I ment 1.38125v.

    Sorry for confusion :(
     
  7. Big Elf

    Big Elf Oh no! Not another f----ing elf!

    Joined:
    23 Apr 2009
    Posts:
    3,940
    Likes Received:
    595
    You might find some of these settings useful for quick calculation of FSB and Multiplier to give optimum settings for RAM speed if you're using PC-8500.

    [​IMG]
     

    Attached Files:

    Arboreal likes this.
  8. thewelshbrummie

    thewelshbrummie Minimodder

    Joined:
    29 Aug 2009
    Posts:
    414
    Likes Received:
    47
    Still have my old Q9650 build. Admittedly mine is on a 2008 era X38 motherboard for which I paid £250 back when £250 boards were top of the range. It's probably the same age as your P35.

    I started out with a cheap £50 Pentium for that board, from which I got a 45% overclock (1.8 to 2.6) just from boosting the FSB. From memory I put the Q9650 in a few years later and never got close in % terms (I never messed around with Vcore) but running it at an FSB of 400MHz at stock voltage was pretty easy (my Asus board, despite X38 only being rated for 333MHz, was advertised as supporting 400MHz FSB which it does without issue). It's a DDR3 board - which from memory most LGA775 boards weren't but with 8GB of 1,600MHz DDR3 it's still pretty stable (not 100% though - even in 2010 it would occasionally crash and age hasn't been kind to it). I was really happy though with hitting a 2:1 FSB/memory ratio though with those settings...

    I can't help any further as the rig is in the UK... and I'm not. But from memory 4GHz should be attainable for it with the right board - I just wasn't too fussed with pushing it too hard as I just wanted mine to last as a 32bit XP build for Win95 era 16bit games, and for that purpose it's holding up well.
     
  9. cookie! nom nom

    cookie! nom nom Minimodder

    Joined:
    27 Apr 2012
    Posts:
    970
    Likes Received:
    86
    Is it normal for prime 95 to have different speed workers?

    I can't seem to find a clear answer and uping voltage isn't helping.
     
  10. LennyRhys

    LennyRhys Fan Fan

    Joined:
    16 May 2011
    Posts:
    6,411
    Likes Received:
    920
    As Big Elf says, cores would complete stages at different speeds but generally within a few seconds of each other. I did >8 hours of stress testing on my Q9650 and the threads were out of sync by no more than 2-3 minutes.

    Your core voltage is pretty high for 3.7GHz, and if what Big Elf says about vdroop is correct, then the CPU is probably not getting enough voltage under load. Some users report that lagging P95 threads are caused by insufficient vcore, so it's possible that's what is happening here.

    For what it's worth, here are the settings I used for 4.3GHz, but bear in mind I had a good P45 motherboard (Asus Maximus II Formula). Notice that these mobos have "Load Line Calibration" which is intended specifically to counter the effects of vdroop, stabilising the CPU voltage when under load. If your board doesn't have LLC, you will have to compensate by using a higher idle voltage, which isn't desirable!

    Code:
    PCIE Frequency : 102
    
    CPU Voltage : 1.35625
    CPU PLL Voltage : 1.59275
    FSBT : 1.37825
    DRAM Voltage : 1.8
    North Bridge Voltage : 1.45775
    South Bridge Voltage 1.5 : 1.55300
    South Bridge Voltage 1.1 : 1.20600
    
    CPU GTL Reference 0 : +10mV
    CPU GTL Reference 1 : -35mv
    CPU GTL Reference 2 : +10mV
    CPU GTL Reference 3 : -35mv
    North Bridge GTL Reference : Auto
    DDR2 Channel A REF Voltage : Auto
    DDR2 Channel B REF Voltage : Auto
    North Bridge DDR Reference : Auto
    
    Load Line Calabration : Enabled <-----
    CPU Sread Spectrum : Disabled
    PCIE Spread Spectrum : Disabled
    
     

Share This Page