Possible CPU Overheating

Discussion in 'Silicon (v)Alley' started by tranced22, Feb 6, 2011.

  1. Hey, so I have a couple questions, I have tried looking around on google but didn't find much useful, so I figured i'd just ask my fellow GC members.

    About a 2 weeks ago(approx.) my computer started shutting off when I would start playing games, I first suspected it was due to my GPU overheating(as i had just upgraded that several weeks before then. After I had played with the fan settings and stuff i was able to drop the temperature to no where near what would cause it to overheat, however this didn't solve the problem, next I thought it might be the CPU overheating, wasn't really able to determine if this was in fact the problem, until last weekend my computer shut off and wouldn't power back on. I then found out my PSU had died and I needed a new one. So i purchased a new one, it got here last wednesday and i installed it that night powered on and had no issues. I played a couple video games (for multiple hours at a time), mainly Minecraft and Counter-Strike: Source. However a couple hours ago I decided to play some Bad Company 2. A couple hours into playing my computer shut off on me again(for the first time since the new PSU was installed) So after this happened, i began to suspect the CPU was overheating again. After letting it sit for a while, i booted up and went directly into the BIOS. I checked the CPU temp and my bios was reading it at somewhere in the mid to high 30's (in degrees C). However this temperature was rapidly climbing and eventually got to the mid to hight 50's! This is when i started becoming concerned(as the auto shut off temp for my processor is at 62*C) So i powered my computer off and started looking around on my laptop for ways I could verify that this was in fact too hot as well as attempting to find some software that would help me find out what my CPU temp was while inside the windows OS(as i read somewhere that the BIOS will put a strain on on your CPU to read the max temps). Some of the software i found were: CoreTemp - This was supposed to simply tell me what temp my cpu was, and at first it read 0*C(?), So i rebooted and launched it again. After that it started out reading 27*C, and so i thought that wasn't too bad, however it didn't stay there, it just kept dropping until eventually it reached 0*C. So I figured this wasn't an accurate read. After looking around a little bit more online I found suggestions for this program Called OCCT. Which basically runs a stress test to see if your computer is stable, i ran it for the CPU option at the very top(i think it was just called CPU - OCCT). The website suggested the 30 minute test for a pretty reliable read through(however did suggest that some people liked to do 2 hours or even 24 hour tests.). And(im guessing) at the end of the test is when it would give me the results about the CPU temps as well as any issues, however about 5-10 minutes in my computer shut itself off.

    Does this confirm my suspicions of the CPU overheating?
    Are there any other programs people may suggest to read my CPU temp?
    I also have a formatted bootable flash drive with Hiren's Boot CD v13.0, however I was unable to navigate through that and comfortably select any of the options other than running the Memtest(which checked out fine after i believe 3 passes) And if anyone can help me through running the diagnostics on that/would suggest or not suggest it. I would appreciate any help. thanks.

    Computer Specs(nothing is overclocked):
    CPU: AMD Phenom II x4 955 BE @ 3.2ghz
    GPU: ATI Radeon HD 6870
    PSU: OCZ Modxstream-PRO 700w
    Case: Antec Nine Hundreed Black Steel ATX Midtower
    Mobo: Biostar TA890FX
    HDD: WD Caviar Green 1TB


    If theres any relevant information I left out I'd be glad to add it, just ask.
    Thanks again :)
     
  2. #2 Hydroriffic, Feb 6, 2011
    Last edited by a moderator: Feb 6, 2011
    First check out everest ultimate edition,it cost but check out the bay;)... It reads just about every temp, it also has its own stress test option. Do you get BSOD? or does it slow down before it crashs?

    My first guess was the PSU, but i see you have a quality PSU.
    The only way to tell if its overheating is to watch the temps while stress-testing. Get everest then watch the temps while u stress it.
    Id get some replacement thermal paste and check that out, usually thats the problem with overheating. Did you use the stock thermal paste? Or did you put your own on, maybe incorrectly??:confused:not saying you did but its often the case.

    Your case has GOOD airflow, so theyre really isnt a reason your cpu should be overheating. As long as the heatsink fan is on, that case should supply it with enough cool air. What heatsink do you have, the stock, or maybe one from a previous build?
     
  3. #3 tranced22, Feb 6, 2011
    Last edited by a moderator: Feb 6, 2011
    Unfortunately i have been completely cut off from the bay :( as i am living on campus and they have strict no P2P software rule and i get a warning simply from opening on the software on my computer while im on the network. And no I don't get the BSOD, and it doesn't even slow down before hand it just suddenly powers down, and then when i reboot(i have to do so by manually hitting the power button, its not a restart), is that windows tells me that it didn't shut down properly. As for the software, is there a free version of everest? or possibly a .zip or .rar download somewhere(like megaupload or one of those sites).

    As for the thermal paste/heat sink, they are both the stock ones that came with my processor, and since this was my first tower that I built I had a friend who has done this a couple times before help me and hes the one who actually applied it. And as far as I could tell the heat sink fan is running fine, it was reading at about 3300-3400 rpms in the BIOS(which i compared to my brothers on his Phenom II x4 965 BE @ 3.4ghz which was at 3500 rpms and was in the mid to high 40's in Celcius in the BIOS-and he has had no issues when running games since we built his tower a couple months ago).

    The only other thing i can think of is that when my PSU failed last week and my computer wouldn't boot back up, i was still concerned it was the CPU overheating i took the heat sink off to look at it and make sure that it didn't fry the CPU and it looked fine, could taking the heat sink off have messed up the thermal paste that was on there?(i was pretty careful when taking stuff off/putting it back on)

    Thanks again for the help. :)

    Just found this link, is this the same version or edition as the one you had in mind? it says ultimate edition, but you never know http://download.cnet.com/Everest-Ultimate-Edition/3000-2086_4-10499291.html ----> lol just noticed it was a 30 day trial for it, but that should work right?
     
  4. There's your problem... 62 C is "high-warm" for a decent CPU. They're rated up to 95 C (and I usually put the shutdown at 90 C), and many desktop CPUs get up to 65 C under load.

    That's still pretty warm, but not cause for auto-shutdown... I'd say bump up your shutdown temp to 75, and clean/replace the CPU grease if you haven't done so recently. Applying CPU grease is a bit of an art but once you grasp the fundamental concepts of how the grease works and how much to apply, you'll be fine. Be sure to COMPLETELY CLEAN both the heat sink and the CPU of any remaining crud... use rubbing alcohol and a cloth, or if you're feeling spendy, get an Arctic Silver "Arcticlean" kit (I swear by that kit, it's amazing). Get it completely clean, nothing but metal left. Then, apply just enough to cover the thermal surface without spilling over, and DON'T spread it out with a card or anything (old PC repair myth/mis-advice). If it's a large surface (like a modern desktop CPU, derp), just make an "X" in the center (about 2/3 the height/width of the chip) of a thin bead of grease, then let the CPU cooler flatten it out.

    Then you should have no more cooling problems. :D

    edit: Also, screw that Everest crap, just get HWMonitor (and grab CPU-Z too, while you're at it, just as a cool tool to have). It tells you everything you need to know, it's free, and it's no-BS.
     
  5. #5 tranced22, Feb 6, 2011
    Last edited by a moderator: Feb 6, 2011
    Hmmm, alright ill check out the HWMonitor as well as the CPU-Z, and the 62*C auto shut down temp is the temp that i got from the AMD site, and its labeled as Max Temps. ( AMD Processors for Desktops: AMD Phenom )

    If i do end up modifying the Auto Shut down temp where would i do that? in the BIOS?

    As for the thermal compound/grease my whole tower(aside from the cd/dvd drive) are less than a year old, i bought all the parts last summer. And from what i understood that is supposed to last for several years(even though many people suggest to change it on a yearly basis)

    I just opened up the Hardware Health Configuration in my BIOS the Shutdown Temperature is marked as disabled?

    Also just ran HWMonitor and these are the temp readings:
    THRM 48*C(current) 52*C(max)
    CPU 48*C(current) 54*C(max)
    Mainboard 63*C(current) 64*C(max)
    TMPIN2 21*C(current) 21*C(max)

    and then under AMD Phenom II X4 955
    each of the cores is listed 0*C(32*F) for all 4 of em
     
  6. Strange that AMD would list a 125w CPU (HOLY SHIT that's a hot fucking CPU! My laptop is 35w and my Core 2 Quad desktop is like 65w!) as having a max temp of 62 degrees. o_O

    Yeah, the grease is supposed to last quite a long time, but the thing is, it's rarely properly applied. Most "I built my own computer, herp!" computers I've seen are absolutely gut-wrechingly poor thermal installations... that's why I recommend re-applying it, just to be sure. But a good way to find out is to feel the heat sink. If the heat sink FEELS like 60 C (quite hot to the touch), then the grease is fine, you just have a problem with getting the heat off the heat sink. If it's COOL to the touch and saying 60 C, then you have a grease/thermal conductivity problem (and grease would likely fix it).

    Of course, there is another oft-misadvised habit a lot of folks have: running their computer in the "high performance" power profile. All that does is prevent the CPU from cooling itself properly... it does NOT actually provide any additional performance gain (the "top-end" settings are exactly the same as the "balanced" profile). So if you changed that, might want to switch it back to "balanced" :)
     
  7. #7 tranced22, Feb 6, 2011
    Last edited by a moderator: Feb 6, 2011
    Hmm, would changing the profile setting make that much of a difference, cause i've just left it at high performance(kind of scratch that question, cause i did just see a HUGE drop in temp from just changing that setting... lol its now reading at 36*C and possibly still dropping) . And also i've felt the heatsink a couple times before and it had felt pretty hot (not so hot that i couldn't touch it though) and also the only other high temp i see is the mainboard, which is still around 62*C is that anything i need to worry about? and what is that exactly? lol

    Thanks again

    PS
    The power for my processor is at 19.5W and the max is at 113.4W

    And at this point the CPU temp is down to 29*C lol... thats pretty ridiculous...
     

  8. And there you have it, folks. :p

    Yes, the power profile setting makes a huge difference. I actually usually customize "Balanced" on notebook PCs to max the CPU out at 50% on battery, to save battery life (and minimal performance impact unless you're encoding movies or something), that's how I run mine.

    MB temp is a "mystery" sensor somewhere on the board... no MB manufacturer puts it in the same place, and sometimes (like it seems in your case) it's monitoring the northbridge or southbridge temperature. Check those heatsinks and see if it feels like 60 C to you ;)
     
  9. Haha yeah, so I restarted my computer, just to see if everything would remain normal after a reboot, I booted into the BIOS first and the temp starts in the high 30's then it would still go up into the 50's(i went ahead and just booted up once it got to like 51*C). So im guessing the BIOS does in fact stress the CPU to get more of a max temp reading? and will those be the temps i will likely see whlie gaming?

    Also now that im booted up it seems that the CPU temp is stabilizing around 28*C-29*C. However since I am still yet to play any games I don't know if the random shut down problem is actually resolved, I'll try and figure that out by some point tomorrow for sure, and repost the results then. (I will also probably try running the OCCT 3.1 CPU Stress test).

    Anyways thanks for suggesting that, I honestly never would have thought to adjust the power settings... lol I'm still semi shocked at both how fast it dropped the temp as well as how much it dropped the idle temp.(i honestly think it dropped around 6*C instantly when i changed the poewr setting).

    And as for the Northbridge and southbridge things, where would i find the heatsinks etc for those? And is that a different possible alternative for the auto shutdown thing? (it seems its very slowly dropping and is at 56*C atm) What is a common overheating temperature setting for the auto shutdown on those?

    Thanks for the help... I love GC! :)
     
  10. (I do sort of work for +likes and +reps, just sayin' :) )

    The reason the CPU cools off in Windows (under the Balanced profile) is because it actually shuts down idle parts of the CPU, lowers the voltage and lowers the clock speed when it's not needed. It does this about 200 times a second, adjusting its speed and power to match what you're using it for... as it loads a web page, for example, it might speed up to load it quicker, then cool back down. In high performance mode, it's basically just free-wheeling, running at a billion miles an hour, producing a shitload of heat, waiting for you to do something. Windows doesn't let the CPU "idle down" when it's in High Performance mode. That's what makes it so hot.

    The BIOS isn't advanced enough to handle cooling the CPU down, so the CPU gets quite hot. Depending on the type of BIOS, it might be running "idle loops" or simple instructions that don't make the CPU as hot as it would be in a game, though. Different activities produce different amounts of heat depending on how many parts of the CPU need to be activated to process it (I could get into a lot more detail here, but it's basically that ;) ). So it gives you a good idea, but not necessarily a spot-on idea. If it's getting to 50 C in the BIOS, expect it to get around 60 C in a game or other intensive app.

    The northbridge is pretty much the only other big heatsink on the board, usually right next to the CPU. They often don't have fans, but the heatsink is usually used to show some kind of company logo (like Asus)... or, hell, I'll just look up your board ;)
    [​IMG]
    Ah, yeah, there it is. It's that big metal block under the CPU cooler, that says "BIOSTAR" on it. Check how warm that's getting. Your other chip, the southbridge, is the much smaller heat sink located near the battery, under the memory slots. It's just a metal block with fins. Under each of those is a tiny chip, like the CPU, that produces a good amount of heat. Those metal blocks are the only thing keeping them cool, and they rely on air flow from the CPU cooler and other fans in the case. :)
     
  11. Haha, yeah man for sure, i actually didn't even realize there was +rep on this site(i've seen people mention it but it seemed jokingly and not too serious). but i see a button for liking your post so ill go ahead and click that and see what happens. Also will the north/south bridge likely heat up much during games or are those generally going to stay around the same temp?
     
  12. Rep is that little button next to the post number in the dark green bar, looks like a scale :) But don't mind rep if you don't want to, just saying.

    From what I've seen, NB and SB temps generally stay about the same all the time. So if it's not burning up after 10-15 minutes, it'll probably stay OK while gaming. The only thing I can think would be different is the heat coming from the CPU cooler (during gaming) affecting the NB temps... but they don't heat and cool like the CPU does. Just make sure they're cool and it'll all be cool.
     
  13. Ahh found it, yeah i went ahead and gave you some rep for the post about the power profiles... cause seriously thats pretty ridiculous about the High Performance thing, i looked around in the advanced settings for it and sure enough, the minimum performance setting for the CPU was set to 100% for the High Performance one, vs the 5% setting on the balanced(i checked on my laptop, but i imagine they are the same as on my desktop). Anyways ill post back here tomorrow if that fixed the problem.

    Thanks again for the help :)
    Goodnight!
     
  14. #14 tranced22, Feb 6, 2011
    Last edited by a moderator: Feb 6, 2011
    Hey, so i just ran the OCCT v3.1 CPU - OCCT test. And I was watching the results on HWMonitor while it was going to watch my temps during the test.
    The first minute or two of the test its just idle, and recording the temps. then after another maybe 3 or 4 minutes of the test, I watched the CPU temp climb all the way to 87*C before i stopped the test to let it cool back down.

    The only thing im unsure of is i get a message saying "OCCT detected that CPU-Z or HWMonitor is running. As OCCT uses the CPU-Z and HWMonitor SDK, this can lead to weird results in System Information or Monitoring Information. Do you wish to continue?" then i can either click "OK" or "Annuler". I went ahead and just closed both programs before I ran the test, and then I ran OCCT first, then opened up HWMonitor, I didn't get an error or warning message of any kind doing it this way, but I doubt that would change much. Could these be just innacurate results? And if not, does this likely imply that I do infact need to reapply some thermal grease?

    And also, I noticed that the Mainboards temp reached around 66*C during the Stress test, is that likely just from residual heat from the CPU getting to such high temps?

    Also just found out that im still using the stock thermal grease that came pre-applied on my heatsink.
     
  15. Chyeah. So um... how did you "find out" that you're still using the stock grease? And please don't tell me you took off the heat sink to check ;)

    And yes, that's a really high temp of 87 C. That's much more likely to cause a shutdown. It's unlikely that having hwmonitor running would cause "inaccurate" readings, either it'll work or it won't work. So that's probably right...

    Definitely time to re-apply that grease if the heat sink doesn't feel like 87 C...
     
  16. Yeah, I actually found out that i was using the stock thermal grease by asking the friend who originally helped me build the computer, and while talking to him it came up that i did in fact do that last weekend(take the heatsink off that is) when my PSU died and i was worried the CPU may have been fried(which now looking back on it was fairly silly lol) but yeah I figured out that i basically just need to buy some new thermal grease and re apply that. I'm looking to either get Arctic Silver 5 or the Antec Formula 5(i think thats what it was called, ima check out what they have at some of the local computer stores around campus before i go to order it online). I'm just glad im finally getting to the bottom of this, lol i wanna get back to gaming! haha.

    So having either air pockets or some grime or something like that can really up the heat on the CPU while its under load? (it still idles around 27-28*C)
     
  17. #17 FalconFour, Feb 7, 2011
    Last edited by a moderator: Feb 7, 2011
    Even taking the CPU heat sink off will destroy the grease used on it. Unless you JUST installed it with fresh grease like 2 minutes ago, any time you take the heat sink off, the grease is too screwed up to work. (And even if you just installed it, removing/reinstalling it - even to "check it" - significantly reduces its performance) Reason being is the same reason you never want to flatten it out with a card... when you press the cooler into place, it fills in all the gaps perfectly, and makes a perfect conductive surface. When you take it off, the grease can't just magically "re-blob", so it stays in the same shape it was before: completely flattened. When you put them back together, it doesn't fill properly and traps a lot of air in it. That's why it's important to never remove the heat sink unless you're ready with a rag, alcohol, and a tube of Arctic Silver ;)

    Especially if the grease is the original stuff... that crappy, dry, flaky junk pre-applied to the heat sink? Dries up within hours of first use... so when you take it off you're basically turning it into useless powder.

    I'd recommend this kit if you're looking for "doing it right":
    Amazon.com: ArctiClean 60ml Kit (includes 30ml ArctiClean 1 and 30ml ArctiClean 2) and 3.5grams Arctic Silver 5 Thermal Compound: Electronics
     
  18. Hmmmm, if i end up buying it online, ill probably get that, the main issue i have right now is that money is a little tight atm, mainly cause of the PSU i had to buy last week. so i don't wanna spend to much, but that actually doesn't look much more expensive than just the grease i was looking at. But i am still probably going to check out the local stores first and see what i can find there. i also plan on checking out a couple guides or videos and stuff online to make sure i do it right. Would you say that using the remover and purifier are a lot better than just cleaning it myself?(i'm not even sure what/if anything i'd have to do that atm)
     
  19. Well, here are your choices:
    1) use the kit...
    a) couple drops of material remover ("1"), rub it all over the CPU until the grease turns to liquid
    b) couple drops of remover to the heat sink, rub it around...
    c) wipe it off with a rag, repeat if anything's still stuck to either
    d) couple drops of purifier ("2"), rub it around a bit to coat the surfaces
    e) wipe it off with a rag until it's shiny clean
    [ame]http://www.youtube.com/watch?v=2MBWJ4K5UD4[/ame]

    or
    2) use rubbing alcohol
    a) douse the end of a rag in rubbing alcohol
    b) wipe wipe wipe wipe
    c) scrub scrub scrub scrub
    d) douse the other end of the rag in alcohol
    e) wipe wipe wipe wipe
    f) scrub scrub scrub
    g) repeat until it looks "clean enough"
    [ame]http://www.youtube.com/watch?v=r5QUggzDTw4[/ame]

    Maybe it's the shininess that the ArtiClean kit brings to a cooler and CPU, but I tend to like the (1) method better ;)
     
  20. Solid advice in this thread.

    Don't let that chip get as hot as you are though. Definately stay below 60 under load if you want it to last.
     
Loading...

Share This Page