1. Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Sensor Test Crash

Discussion in 'RealTemp' started by w1ldm4n, Nov 11, 2009.

  1. w1ldm4n New Member

    Joined:
    Nov 11, 2009
    Messages:
    10 (0.01/day)
    Thanks Received:
    0
    I am running Windows 7 Professional (64 bit) and my computer crashes when I try to run the sensor test.
    Details:
    E8400 (not overclocked)
    My Idle temps are supposedly around 45-50c

    When I click sensor test it opens the box, I select Prime95 and let the test run. Everything seems to be going fine until the idle test when it says 'stopping Prime95'. At this point, my entire computer freezes, nothing on the display changes at all. Numbers stop changing, even the second hand on the windows clock gadget on the desktop stop, no a single pixel on my display will change. I have to use the hardware reset button on my computer and then everything boots normally.

    Why is this happening, and what can I do to fix it.
    I tried the 32 and 64 bit versions of Prime95, is there a 64 bit realtemp?

    P.S. I've heard around the internet that Tjmax for an E8400 is 100c, is this correct?
     
  2. mlee49

    mlee49

    Joined:
    Dec 27, 2007
    Messages:
    8,488 (3.45/day)
    Thanks Received:
    2,105
    The TJMax is 100 for almost all Intel chips, so your temps should be fine.

    Something is overloading and halting, does it BSOD? What voltages are you running on the chip, even though its not overclocked?

    Realtemp doesn't have a 64bit version btw.
     
  3. unclewebb

    unclewebb RealTemp Author

    Joined:
    Jun 1, 2008
    Messages:
    948 (0.41/day)
    Thanks Received:
    420
    I just ran the Cool Down Test on my E8400 using Windows 7 x64 and there weren't any problems.

    [​IMG]

    That leads me to believe that the test itself is OK.

    What this test does to vary the load level of your CPU is it adjusts the Clock Modulation feature of your processor at each step. This feature has been built into all Intel CPUs since the Pentium 4 era and provides a user with a way to manually lower the performance of their CPU to control heat. That can be handy when you have a laptop burning through your lap.

    This feature can be accessed in the RealTemp settings window. A stable CPU should be able to run at any Clock Modulation setting from 12.5% to 87.5% and should have no trouble going from one Modulation level to another on the fly. Those are Intel's theoretical numbers and aren't 100% accurate but as these percent settings decrease, the performance and heat output of your CPU should also decrease.

    Is your computer 100% stable?

    I recently bought 2 x 2GB memory modules and installed them in an older motherboard. I tried for a day or two and tried every possible bios setting, etc., and I found absolutely no way to get my computer 100% stable. I tested at different CPU speeds from 1600 MHz to 4000 MHz, I tried loose memory timings like CL 6-6-6 and tight memory timings like CL 4-4-4, I tried fast memory speeds, slow memory speeds, lots of northbridge voltage and hardly any. I played infinitely but could not get the Blend test in Prime 95 stable for more than about 30 seconds. This exact same computer and same memory in Windows Vista x86 is 100% stable for hours when testing with Prime95.

    When I try Windows 7 x64, the x86 version of Prime 95 is 100% stable but my computer is not stable when running the x64 version of Prime 95. It has nothing to do with Prime 95. Lots of users can pass the x64 test. As soon as I switch back to 2 x 1GB memory modules, my computer can pass the x64 test too, even when overclocked to 4 GHz.

    Try adjusting some settings on your computer. There's no reason why RealTemp should be locking up. Rather than run the Cool Down Test, just run Prime 95 separately and play around with the Clock Modulation settings when trying to recreate the problem.

    TJMax for the average E8400 is 100C but Intel admits that this can vary a little bit from one CPU to the next, even with the same part number.
     
  4. w1ldm4n New Member

    Joined:
    Nov 11, 2009
    Messages:
    10 (0.01/day)
    Thanks Received:
    0
    Here's a screenshot of my computer after running Prime95 small FFTs for about 10 minutes

    http://img691.imageshack.us/img691/4459/tempsz.png
    (that's SpeedFan at the bottom btw)
    Edit: OK, I just noticed that the picture says 2.9994, but it goes around 2.9994-2.9998

    What do you mean by 100% stable?
    I have a Gigabyte EP45-UD3P mobo (P45 chipset) with 2x2GB DDR2 1066
    For whatever reason, the core speed is like 2.9997, and my Core Voltage (Vcore) is 1.814.

    Yesterday, in accordance with http://www.tomshardware.com/forum/221745-29-sticky-core-core-temperature-guide I lowered my clock speed and Vcore way down, and accidentally crashed my computer - I had to open it up and pop out the CMOS battery to reset everything in the BIOS and get it to boot. Then I uninstalled all EasyTune 6 (Gigabyte's overclocking program) and the computer seems to run fine, I just did a Prime95 x32 blend test while typing most of this post and everything ran perfectly, except that occasionally in CPU-Z the Core Voltage would jump to 1.200v and then go back to 1.814.

    But I'm worried about 2.9997gHz and 1.814V (isn't 1.25 normal?). Could I still be experiencing residual side affects of lowering Vcore too much, (I took it down to like 1 or 1.1 volts)? And if so is there a way to get rid of them?
     
  5. unclewebb

    unclewebb RealTemp Author

    Joined:
    Jun 1, 2008
    Messages:
    948 (0.41/day)
    Thanks Received:
    420
    100% stable means your computer can run a variety of stress testing programs like Prime95, 32 and 64 bit with the Small FFT and Blend options, LinX, SuperPI 32M as well as a 3D based benchmark program like 3D Mark06.

    You can buy a brand new computer from a store and even if you don't overclock it, there's no guarantee that it is 100% stable and can pass a variety of the above tests.

    It's very easy to end up with parts that are not compatible or need to be adjusted with changes in the bios to voltage, memory timings, etc. Defective memory is probably the number one cause of system instability and even brand new memory needs to be tested with programs like MemTest 86+ to make sure your memory is working correctly. Whether your computer is brand new or a few years old, you can never take its stability for granted without running a battery of tests on it.

    A 3 GHz CPU may not run at exactly 3 GHz. RealTemp is pretty accurate and it is showing 2999.69 MHz so that's pretty damn close.

    Core voltage of 1.184 while your computer is fully loaded is also normal. It is the Intel design spec for voltage to drop slightly when fully loaded. Some motherboards have options to prevent this but if your computer can run stable at 1.184 volts then there's no need to worry about this.

    SpeedFan is using the wrong TJMax for your CPU. You can adjust this to the correct 100C in SpeedFan if you want to. That's why RealTemp and SpeedFan report 5C different.

    There shouldn't be any residual side effects. Every time you go into the bios and change something, you should have a fresh start.

    The UD3P is a great board but there are also hundreds of things that can be adjusted in the bios and need to be set correctly. It might be a good idea to learn more about your motherboard and what settings are important for stability.

    Edit: Here's a more recent version of RealTemp with a few more features for Nvidia graphic cards.
    http://www.techpowerup.com/downloads/1691/Real_Temp_3.40.html
     
    Last edited: Nov 12, 2009
  6. w1ldm4n New Member

    Joined:
    Nov 11, 2009
    Messages:
    10 (0.01/day)
    Thanks Received:
    0
    Well, even with the new version of RealTemp my computer still freezes about 5 seconds int test 9 (right after it kills Prime95), but I wrote down and graphed the numbers and they make a perfect line all the way down so I don't think the sensors are stuck. Although Core 1 reports between 1 and 2 degrees C higher than core 0, but that's not much. So if I know that me sensors aren't stuck is there any other reason that I would run the test?

    My computer is stable when I manually run a Prime95 test (32 or 64), and I passed Memtest86 so I think I'm OK.
     
  7. w1ldm4n New Member

    Joined:
    Nov 11, 2009
    Messages:
    10 (0.01/day)
    Thanks Received:
    0
    In other news, I decided to use EasyTune6 to overclock my E8400 to 3.6 gHz (400x9). It seems to run fine, but my cores (Tjunction) idle at around 50 and Tcase runs at just under 40.
    When I started a Prime95 small FFT, I decided to stop it when Tcase hit 70C after less than a minute (Tj was around 75). Is this safe, or should I go back to normal speed?

    I have a Dynatron Genius Cooler with a hefty amount of Arctic Silver 5 (which may still be breaking in) in an Antec nine hundred 2 case that has plenty of airflow (200mm exhaust right above cooler, 120mm exhaust in the back next to the cpu, and an intake fan between my graphics card and my cpu). Do these temps make sense? because I built a system for a friend and his i7 hits 40C under a heavy load and idles at room temp with a very similar cooler
     
  8. sneekypeet

    sneekypeet Unpaid Babysitter Staff Member

    Joined:
    Apr 12, 2006
    Messages:
    21,569 (6.99/day)
    Thanks Received:
    6,100
    the "hefty amount" can be counterproductive....try reseating the cooler with as little AS-5 as needed to spread an thin, almost "see-through" layer on the CPU IHS. Reseat the cooler and retest. Too much TIM on a cooler actually screws with the transfer of heat just as much as seating it improperly.

    Googling your cooler, I would expect way better temperatures at stock clocks with your CPU as well. AS-5 is by no means the best paste, but for your purposes it will sufice. I do suggest pulling the cooler off and cleaning it all really well, then reapply and reseat.
     
  9. unclewebb

    unclewebb RealTemp Author

    Joined:
    Jun 1, 2008
    Messages:
    948 (0.41/day)
    Thanks Received:
    420
    What does CPU-Z show for core voltage when you are overclocking? A core temperature of 75C as reported by RealTemp won't hurt anything but with that cooler at 3.6GHz, that sounds a little high. As long as your computer is running reliably, that temperature is OK. I'm not a big fan of Easytune or other Windows based overclocking programs. I prefer to adjust the settings in the bios so I have better control and so the memory isn't running out of spec.

    I have no idea why the sensor test freezes your computer. It works correctly and completes on all the computers I've run it on. You might have some other protection software on your computer that thinks RealTemp is doing something bad when it kills the Prime95 task.

    Your sensors sound OK and it is normal for core 0 to report a couple of degrees higher than core 1. These sensors are far from perfect and your sensors sound like they are better than most.
     
  10. w1ldm4n New Member

    Joined:
    Nov 11, 2009
    Messages:
    10 (0.01/day)
    Thanks Received:
    0
    @sneekypeet:
    First of all, the temperatures that I listed were when I OCed to 3.6gHz, they were 5-10C cooler at factory clocks.
    And I feel kind of dumb because the first time I put the heat sink on I just used a little line in accordance with the instructions on the AS5 site, but when I thought my cores were idling at 50C, I cleaned everything off and put that line on, plus filled in the cracks between the heatpipes and the base of the cooler. (Although in retrospect that was a bad reading because SeppdFan was using 105Tjmax instead of 100, so I guess it was 45) According to the AS5 instructions, the chip that actually makes the heat is like a rectangle and thats where to put the AS5, and I also have my cooler oriented so that the 4 heatpipes are perpendicular to that chip (so that the heat goes to all of them and not just one or two), is this right?

    If I were to go back and redo the AS5, do I have to clean everything off and reapply, or can I just take off some of the excess and leave a little on the chip and not have to use any more?

    Also, from almost everything I've heard, Arctic Silver 5 is the best thermal paste out there, what do you know of that's better?
     
    Last edited: Nov 14, 2009
  11. burebista

    burebista

    Joined:
    Sep 8, 2005
    Messages:
    612 (0.19/day)
    Thanks Received:
    181
    Location:
    Romania
    It was at his time and it's still OK. Now choose one for one-two-three degrees less. :)

    [​IMG]
     
  12. w1ldm4n New Member

    Joined:
    Nov 11, 2009
    Messages:
    10 (0.01/day)
    Thanks Received:
    0
    CPU-Z shows a core voltage of 1.280V while running a large FFT test in prime95 (and my temps after 10 minutes are Tcase:68-69 and cores are both 72)

    I'll get rid of ET6 and just change stuff in the BIOS and see if that works better.
     
  13. unclewebb

    unclewebb RealTemp Author

    Joined:
    Jun 1, 2008
    Messages:
    948 (0.41/day)
    Thanks Received:
    420
    At a core voltage of only 1.28 volts, a core temp of 72C is higher than normal. Your room temperature and case air flow play a big part in this. When you're bored you can try re-mounting your heatsink but it might not make a big difference. AS5 is more than adequate. There will always be something better but a couple of degrees isn't worth driving to the store and spending money for.
     
  14. unclewebb

    unclewebb RealTemp Author

    Joined:
    Jun 1, 2008
    Messages:
    948 (0.41/day)
    Thanks Received:
    420
    w1ldm4n: I'm curious about why your computer crashes during the sensor test and would like to try and get this to happen on my computer. Do you use UAC and does your account have administrator privileges? Also, what antivirus program or anti spyware programs do you have running in the background or anything else you can think of.

    In Windows 7 x64, even a bad program shouldn't be able to lock up the entire operating system. Do you let RealTemp shutdown Prime95 (that's what you should do) or do you shut down Prime95 manually? If I can duplicate the problem you're having and if it's anything I'm causing then maybe I can find a fix for this issue.
     
  15. w1ldm4n New Member

    Joined:
    Nov 11, 2009
    Messages:
    10 (0.01/day)
    Thanks Received:
    0
    I have the final version of Windows 7 Professional x64
    My account has admin privileges and I click 'allow' to a UAC prompt when starting RealTemp
    I have a BFG Geforce GTS 250 OC out of the box, and then manually OCed all the way I can without artifacts (I did this a while ago and everything runs fine, and when I reset the clocks RealTemp still crashed)
    I have BitDefender internet security 2010
    I run a copy of Apache in the background, but that shouldn't do anything.

    If there's anything that could be screwing stuff up it could be this:
    About a month ago when I was overclocking my GPU, I tried to install NVIDIA nTune, but for whatever reason it failed to install right, but I have some nTune services running so it looks like it halfway installed or something...

    And I let RealTemp kill Prime95 doing small FFTs
     
  16. unclewebb

    unclewebb RealTemp Author

    Joined:
    Jun 1, 2008
    Messages:
    948 (0.41/day)
    Thanks Received:
    420
    w1ldm4n: I tried the CPU Cool Down Test with UAC enabled to the highest setting, "Always notify", but I still couldn't find any problems.

    [​IMG]

    Maybe BitDefender is making your computer extra secure. RealTemp starts the Prime95 task so it should be allowed to end this task without any security software complaining. Even if something was trying to protect your computer, no software should lock up your whole computer.

    I think I'll go download the trial version of BitDefender and install that to see if I can isolate this problem further.

    Edit: BitDefender 2010 with UAC at Always notify doesn't seem to be causing a problem either.
    You've got me stumped. I'm not sure why the RealTemp Cool Down Test locks up your computer.

    [​IMG]
     
    Last edited: Nov 16, 2009
  17. w1ldm4n New Member

    Joined:
    Nov 11, 2009
    Messages:
    10 (0.01/day)
    Thanks Received:
    0
    Well for whatever reason, without changing anything except overclocking my processor to 3.6gHz (it crashed at that speed before) I started some music and ran the test to see if it was my entire computer that was locking up or just the display, and low and behold, the test completed perfectly all the way through.
    I don't think I did anything differently than before besides the music (windows media player), but RealTemp 3.39.5 wasn't the problem.

    In other news my sensors aren't broken and my cores idle between 45 and 50 and maxed out at 71-72 at 100%. Is this safe to run? (My computer was stable through PCMark Vantage and 3DMark Vantage)

    In other other news, in the next version of RealTemp, could you add a reading for Tcase (and a tray option), I keep getting annoyed at having to open SpeedFan to get a Tcase temperature.
     
  18. unclewebb

    unclewebb RealTemp Author

    Joined:
    Jun 1, 2008
    Messages:
    948 (0.41/day)
    Thanks Received:
    420
    Tcase temperatures are often times not very accurate. Intel CPUs are controlled by what core temperature they are running at. That's the important number to watch.
     
  19. w1ldm4n New Member

    Joined:
    Nov 11, 2009
    Messages:
    10 (0.01/day)
    Thanks Received:
    0
    This is straight from Intel (http://processorfinder.intel.com/details.aspx?sSpec=SLAPL)

    Correct me if I'm wrong, but that sound like Tcase to me. I read some more into the Core 2 Duo datasheets it seems that for the E7 and E8 series CPUs, it is Tcase that is important. (see http://download.intel.com/design/processor/datashts/318732.pdf section 5)
     
  20. unclewebb

    unclewebb RealTemp Author

    Joined:
    Jun 1, 2008
    Messages:
    948 (0.41/day)
    Thanks Received:
    420
    TCase is important if you are a system builder and choosing heatsinks and fan speeds and stuff like that but for the average user, it's completely unimportant.

    The correct way to measure the TCase temperature that Intel is talking about is to cut a groove into the heatspreader on top of your CPU so that you can run a calibrated thermocouple to the geometric center of your heatspreader. Then you have to solder that into place. How many end users do this to correctly measure TCase. I'll give you a hint. Nobody.

    The TCase temperature as reported by most motherboards is an approximation at best.

    The whole point of the TCase spec is that if manufacturers keep the measured TCase temperature within this limit then a CPU should very rarely if ever reach the thermal throttling point even on a hot day in a small case full of dust bunnies while running a high stress program like Prime 95 on all cores. That's the ultimate point of all this. To keep the CPU from reaching the thermal throttling point so the end user is able to get the full performance from his Intel CPU that he paid for.

    The one and only thing that controls thermal throttling is the core temperature. For the end user, it's the only important number to keep an eye on. RealTemp goes one better than that and also keeps track of the thermal throttling bit within the processor. That information is reported in the Thermal Status area of RealTemp. If RealTemp shows OK at the bottom then that means that there have been no thermal throttling episodes since you powered up.

    That is ultimately the only thing that Intel cares about. That your CPU does not thermal throttle during normal usage. On most CPUs if you keep your core temperature at a maximum of 97C as reported by RealTemp, you won't have to worry about reaching the throttling point and your CPU will be running at full speed and within its design spec so Intel will be happy too.
     
    Last edited: Nov 18, 2009
  21. w1ldm4n New Member

    Joined:
    Nov 11, 2009
    Messages:
    10 (0.01/day)
    Thanks Received:
    0
    It seems like you got my post that I made yesterday but for whatever reason it isn't showing up here.

    In any case, that makes enough sense and I guess I won't worry about it.

    Thanks for all of your help on this thread.
     
  22. OceanBreeze New Member

    Joined:
    Nov 24, 2009
    Messages:
    5 (0.00/day)
    Thanks Received:
    0
    I'm having the same problem as w1ldm4n's original post (Crash when trying to run a sensor test).

    I am running Windows 7 Home Premium 64 bit and my the computer crashes (hard) when the sensor test gets to the point where is says "Stopping Prime95". The last "CPU Load" reported was 32.6%.

    Symptoms of the crash is that the Prime95 window closes then everything is frozen - mouse cursor won't move, keyboard keystrokes not recognized, clock gadget quits ticking. The only thing that works is the reset button.

    System details:
    Q9550 (not overclocked)
    Gigabyte GA-EP45-UD3P Motherboard
    4 Gb Corsair PC2-6400 memory
    RealTemp Ver 3.40
    Prime95 Ver 25.8 Build 4
    No virus software installed
    Nvidia 9500GT graphics card - no overclocking

    FWIW, I can run the Prime95 torture test for hours without incident.
     
    Last edited: Nov 24, 2009
  23. burebista

    burebista

    Joined:
    Sep 8, 2005
    Messages:
    612 (0.19/day)
    Thanks Received:
    181
    Location:
    Romania
    Seven x64 and everything is fine here.
     
  24. unclewebb

    unclewebb RealTemp Author

    Joined:
    Jun 1, 2008
    Messages:
    948 (0.41/day)
    Thanks Received:
    420
    Here's the latest version of RealTemp:
    http://www.fileden.com/files/2008/3/3/1794507/RealTempBeta.zip

    OceanBreeze: I wish I had an answer and a solution for you but I don't. I've never once had a problem running the Cool Down Test on XP x86, Vista x86 or Windows 7 x64. There is nothing this test does that is wild or crazy. Built into Core CPUs is the ability for a user to throttle them to control power output and temperatures. This feature is called Clock Modulation and if you do a Google search and include that and Intel you can probably learn more about it.

    For some reason, this test is causing some CPUs to lock up solid. w1ldm4n had this same problem but then it magically fixed itself so I have no idea what changed on his system to fix it.

    I wish I could fix this for you but I can't fix a problem that I can't recreate. All I can suggest is to avoid running the Cool Down Test.

    Maybe instead of running Prime95, try running Orthos instead and only load one core. You can also go in the Settings window and play around with the Clock Modulation feature. That should not cause your CPU to lock up and killing a process that RealTemp started (Prime95) shouldn't cause your computer to lock up either.
     
  25. OceanBreeze New Member

    Joined:
    Nov 24, 2009
    Messages:
    5 (0.00/day)
    Thanks Received:
    0
    I tried the beta version and even updated my MB with the latest BIOS. Still the same result. Maybe I'll try what w1ldm4n did with the music and overclocking and see if that changes anything. I'll let you know.
     

Currently Active Users Viewing This Thread: 1 (0 members and 1 guest)

Share This Page