1. This forum section is a read-only archive which contains old newsgroup posts. If you wish to post a query, please do so in one of our main forum sections (here). This way you will get a faster, better response from the members on Motherboard Point.

65c+ temps at idle on celeron d 3.06GHz

Discussion in 'Overclocking' started by Luvrsmel, Jan 26, 2010.

  1. Luvrsmel

    Luvrsmel Guest

    not overclocked but the system was given to me as a junker because of
    "overheating." The bios was reading 90+C and I could only read it for a
    second before the system re-booted itself. So I replaced the fan and HS with
    a copper core intel HS and the temps are now in the mid 60's as according to
    CPUID HARDWARE MONITOR. Is this normal for this cpu or has it been damaged
    from too much heat? It still runs as evidenced by this posting from this
    machine :)
     
    Luvrsmel, Jan 26, 2010
    #1
    1. Advertisements

  2. Running in the mid 60s with no load is high, my suggestion would be to
    underclock that system. Go into the BIOS and set the clock speed down to
    2.5G, that should lower your CPU temperature by 10 degrees or so.

    When you put on the new heatsink how much thermal paste did you use? You
    don't want to overdo it, too much paste is as bad as too little.

    One more thing, if you want to test the stability of the system run
    sys_basher on it.

    http://www.polybus.com/sys_basher_web/
     
    General Schvantzkoph, Jan 26, 2010
    #2
    1. Advertisements

  3. Luvrsmel

    Luvrsmel Guest

    That would be idle. It fluctuates anywhere from about 60 -72 depending on
    the HD activity..its just up and down constantly. I thought it might be the
    sensors on the board malfunctioning but the air temps read higher than usual
    too 30 - 35C.
     
    Luvrsmel, Jan 27, 2010
    #3
  4. Luvrsmel

    Luvrsmel Guest

    Would love to but its an Intel D865GVHZ board...can't overclock, can't
    underclock :(
    Very little paste. Just a film really. Not an Oreo cookie.
    Thank you ..will try.
     
    Luvrsmel, Jan 27, 2010
    #4
  5. Luvrsmel

    Paul Guest

    You might say the temperature measurement system had a fault with it, but when
    the computer reboots or switches off (THERMTRIP), that tells you it probably
    is pretty hot. So the hardware monitor might not be telling lies.

    It could be that the processor is a Prescott family (90nm) Celeron. That
    family of products had leakage currents consuming up to 25% of the power
    the thing uses. Still no excuse though, for it running hotter than the
    cooling system can handle.

    What to do with it, depends on what your objective is for the system.
    Are you using it yourself as your primary computer, or wishing to
    resell the box to someone else ?

    If a BIOS has no options for clock rate control, you can do a BSEL
    socket mod. I did that to my current Core2 processor. By grounding or
    leaving open circuit, a certain pin on the bottom of my CPU socket,
    I can change the FSB speed indication from FSB800 to FSB1066. That allows
    me to overclock by 33%. Since my overclock was never 100% stable (errored
    out on games), I had to disable my mod. Since I made the mod switchable,
    it is always ready at a moments notice, even if it is useless.

    Your computer might support a similar option. What you have to do though,
    is examine the BSEL hardware encodings, to see what two or three bit
    pattern controls the FSB speed. Some mods are hard to do, or in my case,
    the mod was pretty easy (it still required soldering, to make it secure
    and not fall off). Just say for the sake of argument, your
    processor was a FSB533 one and you could drop it to FSB400. Your CPU core
    would then run at 75% of its normal speed. If the processor is an FSB400,
    then there is no lower setting than that, so a BSEL mod would then be
    useless.

    *******

    You can also try underclocking while in Windows. Motherboards have a
    clock generator chip. If the clock generator chip happens to be listed
    in one of the popular "clockgen" programs, then you could modify the
    speed while in Windows. The computer must remain stable enough, to boot
    into Windows without overheating. Then, in Windows, you could crank the
    FSB down to FSB400 that way. Since I don't know what Clockgen chip your
    board uses, I cannot do the necessary research for you.

    http://web.archive.org/web/20070929075711/http://www.cpuid.com/clockgen.php

    On that web page, you can see ICS technology clock generator part numbers
    in the left hand column. Like ICS950403, would be a part number printed
    on a 48 or 56 pin chip on the motherboard. If there is a match (which is
    highly unlikely but it could happen), then that program could be your
    "friend". Otherwise, the program would be useless to you, as it would
    not know how to adjust the motherboard speed in Windows.

    In this picture of the D865GVHZ, look just above the main aluminum heatsink.
    There are four caps with blue marks next above that. The rectangular
    chip just above the leftmost of those four caps, is your clock generator
    chip. Look for a part number on the top.

    http://www.tnh.com.vn/tnh/data/tnh_prod/d865gvhz.jpg

    Another program you could try, is SETFSB. I've used this one on one
    of my systems, and it was fun to play with. It has a list of supported
    hardware as well. Only supported clockgen chips will work.

    http://www13.plala.or.jp/setfsb/

    *******

    Another alternative, is to buy a replacement processor, assuming this
    is not a problem with the motherboard itself (like too-high Vcore).
    But whether that makes sense, depends on your objectives for the
    system. Maybe pouring any money into it just doesn't make sense.
    I'd probably replace the 90nm Celeron, with a 130nm Northwood family
    P4. As long as the motherboard supports that, it might run a bit
    cooler than your current one.

    There are various examples here.
    http://www.pricewatch.com/gallery/cpu/pentium_4_2.8ghz_478

    Take the SL number and look them up here, to learn more about your
    potential purchase.

    http://processorfinder.intel.com

    For example, if I look up the first one listed on that pricewatch page
    right now, the SL6WJ, the info is here. It is a 0.13u processor, which
    is the Northwood family. The FSB runs at FSB800. If it said 90nm,
    you wouldn't want it (as it could run hot too).

    http://processorfinder.intel.com/Details.aspx?sSpec=SL6WJ

    The manual for your motherboard is here.

    http://downloadmirror.intel.com/15203/eng/D865GVHZ_ProductGuide01_English.pdf

    "Support for:
    * Intel Pentium 4 processor (2.2 GHz or higher) in the mPGA478-pin
    package with 800/533/400 MHz system bus
    * Intel Celeron processor (2.0 GHz or higher) in the mPGA478-pin package
    with 400 MHz system bus"

    So it looks like a P4 2.8GHz Northwood would be a replacement. That particular
    SL6WJ processor has a TDP of 70 watts, so relatively speaking is a cooler running
    processor. If you could get it for a decent price, that might be a solution
    for you.

    *******

    Putting a gigantic cooler on the Celeron would be another option,
    but I have seen reported cases where no matter what cooler is used,
    the thing still overheats. And I don't have good explanations for
    why they should run that hot (without dying). A gigantic cooler needs
    plenty of space within the computer case, and may be tricky to install.
    Some have screws that are hard to access with a screwdriver. If
    any mechanical tolerances are astray, any clips or levers can be
    extra hard to put in place. While you could try a solution like that,
    I'd probably only do it, if I happened to already own a spare
    heatsink and fan. If you have plenty of parts from other busted
    computers, maybe that would be a partial solution.

    (Example of huge old clunker - Tuniq Tower)

    http://ncix.com/products/index.php?sku=17867&vpn=T-120&manufacture=Sunbeam

    (It is 6" high and weighs a couple pounds.)

    http://images17.newegg.com/is/image/newegg/35-154-001-S01?$S640W$

    Have fun,
    Paul
     
    Paul, Jan 27, 2010
    #5
  6. Luvrsmel

    Luvrsmel Guest

    Paul, thank you very much for that extensive research in trying to help me
    out.
    As an update, CLOCKGEN doesn't see my PLL and neither does SETFSB.
    Both support ICS952607EF and unfortunately mine is ICS952601EE which is a
    close as it gets.
    Interestingly enough, this system, as hot as it is, has not crashed since
    replacing the heatsink.
    I plan on using this unit for wireless internet from another room, but I
    certainly don't want to risk a fire hazard.
    Rather than soldering, you certainly found some great CDN pricing on used
    cpu pulls.
    It sounds that cpu replacement is the way to go.
    I found an Intel pdf tech sheet on this board and it does support a 3.4GHz
    800fsb cpu.
    Presently the memory in it is 1 GB of DDR400 so it's all good.
    Thank you again! :)
     
    Luvrsmel, Jan 27, 2010
    #6
  7. Luvrsmel

    Luvrsmel Guest

    a little video present..and quite entertaining actually :)
    http://www.youtube.com/watch_popup?v=jEjUAnPc2VA#t=20

    converted to H.264 mp4 with this machine and the cpu temp climbed up to 83C
    (181 F) as per CPUID HM and still completed the job.
    I'm just amazed at how these temps are possible?
     
    Luvrsmel, Jan 28, 2010
    #7
  8. Luvrsmel

    Paul Guest

    The THERMTRIP on some of the older processors was set
    pretty high, so I guess that's why it is still running.
    If you want to see it shut down, just loosen the clip
    on the heatsink :)

    Paul
     
    Paul, Jan 28, 2010
    #8
  9. Luvrsmel

    him Guest

    Well after picking up another cpu (Cel D 2.66Ghz) and trying in this intel
    board, I'm getting the same overheating problems.
    I've determined that it is not the cpu but the board as the culprit.
    What could be causing the board to overheat the cpu(s)?
     
    him, Feb 11, 2010
    #9
  10. Luvrsmel

    Paul Guest

    The temperature measurement itself could be defective.

    The software reading out the temperature, could be using the wrong equation.

    The processor power is F*C*V**2. Which means power is proportional to the
    square of Vcore. If Vcore goes from 1.35V to 1.45V, the power increase would
    be proportional to

    1.45 * 1.45
    ----------- = 1.15x or a 15% increase in power for a 7.4% increase in voltage
    1.35 * 1.35

    Power dissipation is directly proportional to frequency, but it is
    relatively easy to get frequency information, and determine whether
    your CPU is overclocked.

    If the CPU silicon die is not making good contact with the cooling
    system, that too can lead to excessive temperature. But since
    you've changed the processor, and probably applied fresh thermal
    paste before reinstalling the CPU, I doubt a second processor would
    have a cooling system defect.

    That leaves too much Vcore, or too high a frequency, as contributing
    factors. Or just as easily, the temperature measurement itself is
    garbage. If the claimed CPU temperature is high, and the CPU
    heatsink is making good contact, then sticking a finger on the
    heatsink, should give you a reassuring toasty feeling. If the heatsink
    is cold to the touch, and the temp readout is still high, then make sure
    the heatsink is actually touching the processor. In at least one
    case here, I had a heatsink fit crooked, because some metal prevented
    the heatsink from dropping into place. It is one reason, I always install
    the heatsink, outside the computer case, verify the heatsink is
    fitted properly, and then drop the motherboard into the computer
    case.

    *******

    If you need to look up details on your processor, you can get them at
    processorfinder.intel.com . The box your processor came in, should have
    codes printed on the label. One of the codes is an "SLxxx" code. You
    can enter that code, and get the processor data. For example, this
    Celeron claims to use 1.25V to 1.30V, which means some boxes contained
    1.25V processors, while other boxes had 1.30V processors etc. There is
    no way to know, unless you have a way to get the VID code value, to know
    what shipped in the box. At one time, just a single voltage part shipped
    in the box, but now, a range of products qualify for the same SLxxx code.

    http://processorfinder.intel.com/details.aspx?sSpec=SL9KN

    6 bit VID code
    Processor ----------------> Vcore_regulator ----+---> Hardware_monitor
    ^ (wires on mobo) | (can read Vcore)
    Vcore | |
    +-------------------------------------------+

    HTH,
    Paul
     
    Paul, Feb 11, 2010
    #10
  11. Have you checked the temperature in the BIOS? It's possible that the
    software is using the wrong parameters, if the BIOS has a wildly
    different value then the problem is the software. One more thing, are you
    looking at the thermistor reading or the on die diode reading? The on die
    diode will give you a more accurate reading.
     
    General Schvantzkoph, Feb 11, 2010
    #11
  12. Luvrsmel

    him Guest

    Interesting approach to properly seating the cpu.
    I can tell that the heatsink is seated properly as I gently wiggle the
    aluminium HS while I'm securing it down and any attempt to pull it off at
    that point is difficult as the compound is holding it.
    In addition, when I touch the HS during any HD activity the HS is not
    burning to the touch but warm, certainly not at the temps the cpuid program
    or bios claims that it is.
    Then is it a faulty sensor throwing everything off or could the PS be
    supplying too much voltage to the cpu?
    Even cpuid shows the voltages for my P4 2.6Ghz/512/800 as between 1.46 -->
    1.52 max if that helps
     
    him, Feb 15, 2010
    #12
  13. Luvrsmel

    Paul Guest

    That is a Northwood (0.13 micron) part. I can tell by the 512KB cache.
    There are two processors matching that description on the Intel site.
    This is one of them. The part is listed as having "Multiple VID", so
    short of checking out the processor itself, it is hard to say from here,
    what the correct nominal Vcore value would be.

    http://processorfinder.intel.com/details.aspx?sSpec=SL6WH

    I have a P4 2.8GHz/512/800 and Vcore is listed as "1.525V max"
    on the sticky label fastened to the box it came in. That
    seems consistent with the value you're seeing in software.
    If I was using an Asus motherboard, at idle, when the processor
    isn't pulling very much current, I would expect to see as high as
    1.525 + 0.060 = 1.585 measured value. The Asus motherboards tend
    to overvolt a bit at idle, but then, at idle the processor
    won't be getting as hot anyway.

    A Northwood shouldn't get quite as hot as a Prescott, which is
    the next generation of processor at 90nm internal geometry.

    The issue might be how the temperature is being read out, whether
    the hardware monitor chip channel making the measurement should be
    in thermistor or diode mode.

    Are you reading out the temperature with Speedfan from almico.com ,
    or are you using some other utility ?

    I checked the Intel site, but don't see a specific tool for
    reading out the temperature.

    In the D865GVHZ user manual, I see an option for "Fan Control Configuration",
    and if you disable fan control, it should run full speed. If the motherboard
    is allowed to control the fan speed, it may turn the fan speed down.
    Automated fan speed control, has a "temperature target" it is aiming
    for, which might be hotter than you're comfortable with. I run all
    my CPU fans here at full speed, and don't seek to compromise on temperature.
    I'm not really worried about the processor, and just like the extra airflow
    over the Vcore circuit and the like. It helps keep my (passively cooled)
    Northbridge cool, as my CPU cooler blows air on the Northbridge. Not
    all fan schemes blow air in that direction, and for the side mounted fans,
    their "spill air" might not be aiding the cooling of other components.

    In some cases, it is hard to tell the difference between the various
    temperature readout channels. You can get some idea as to whether you're
    looking at the CPU channel, via the dynamics. If you run a utility like
    "CPUBurn", or "Prime95 Torture Test", you should see a rapid change in the
    CPU temperature. Other channels should respond more slowly to the change
    in computing activity.

    I've always had relatively small temperature jumps on my CPUs, when a load
    is applied. I don't think I've ever had a system that was hard to cool.
    Just lucky I guess. Maybe it's because I can't afford the expensive,
    higher speed processors :)

    Good luck,
    Paul
     
    Paul, Feb 15, 2010
    #13
  14. Luvrsmel

    ~misfit~ Guest

    "Warm" doesn't do it for me. If both the BIOS and the utility that you're
    using within the OS are reporting temps in excess of 65C the HS should be
    more than just 'warm'. 50C is uncomfortable to hold your finger on for any
    length of time and I;d expect that at least. I'd be looking toward your CPU
    / HS interface or even the efficiency of your HS itself as being the
    problem.
     
    ~misfit~, Mar 7, 2010
    #14
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.