Temperature spikes up and down

Since I installed a new power supply earlier this year (the old one suddenly blew out apparantly as the computer just went off immediately and would not reactivate) I've been getting weird temperature readings. When playing games that use the graphics card the sensors named auxtin and systin in HWMonitor spike up and down even to well below idle temperatures. The temperature also spikes up during other graphically intense activities such as watching HD videos or viewing webpages with a lot of full size screenshots though obviously not to the same extent as gaming.

About 2 weeks ago I finally got around to playing such a game in window mode so I could watch the temperatures in HWMonitor at the same time. This screenshot was taken after shutting the game down having played it for about an hour.

http://i20.photobucket.com/albums/b227/DeathZebraEX/ScreenShot009.png

The temperature usually returns to idle almost immediately (though I have seen it take 2 or 3 seconds once) after shutting the game down or if it's in full screen mode, just switching out of it to a different task. The min/max temps got lower/higher the longer I played the game. The 105C max is the highest I've seen.

This next shot was taken from another test a few days ago which I performed with Speedfan also on so I could get some graphs. As you can see, Speedfan agrees with HWMonitor. I just loaded a save file and left the game idle for 15 minutes though this will have done little to decrease the graphical burden as some of the things on screen are animated. The flat lines at the end show when I turned the game off.

http://i20.photobucket.com/albums/b227/DeathZebraEX/ScreenShot014.png

I also took a graph while watching youtube videos. The first half / two thirds are while watching a video in 720p during which systin momentarily spiked up once or twice. The last third was taken while watching a 1080p video of a new(ish) game. I put a line on there for GPU temperature this time and the systin temperature was going up and down even though the GPU temperature barely moved. Presumably, the flat lines at the end are from me closing the browser window.

http://i20.photobucket.com/albums/b227/DeathZebraEX/ScreenShot017.png


Other things possibly worth noting:

• Systin idles at 41-46C. Oddly, the higher the room temperature, the lower this idle temperature. Auxtin idles at 30C +/- 2C
• I do not remember systin and auxtin really moving at all with the previous power supply. I monitor temperature obsessively and would have almost certainly noticed anything like this. I probably would have been worried had systin gone above 50C.
• None of the other temperature readings behave any differently. Only the auxtin and systin temperatures are affected. I installed an intake fan at the front of the case in June and that affected absolutely nothing.
• The old power supply was a no-name 450W with a 80mm(?) exhaust fan. The new one is an OCZ ZS550W with a 135mm fan on the bottom and just a vent at the back. I don't know whether it sucks or blows air. If it sucked it would be sucking hot air from the case and relying on the vent in the back to expel it. If it blew then surely the vent would be taking some of the hot air just expelled by the case's exhaust fan.
• I haven't noticed any change in the computers performance caused by these temperatures. I have occasionally felt the side of the case and that feels no hotter than it ever was.
• I have run various malware scans several times for unrelated reasons in the past few months and they have detected nothing.
• HWMonitor showed a max systin temp of 50C during one of the malware scans. This, along with an incident wherein some hot air was coming into the room from the kitchen, is the only time I've seen systin go up when the computer wasn't doing anything graphically intensive.
• The removal of the old power supply's 24 pin cable proved troublesome and required a lot of force. The motherboard was bent several times in the process.
• After the old power supply blew out I opened the case and noticed the graphics card was bent. Thinking it wasn't plugged in properly I attempted to push it in harder but to no avail. Presumably it had been bent for some time so this probably isn't the problem.

Any ideas as to what's going on here? I can't tell if there's a real problem or not. On the one hand, the sheer speed with which these temperatures go up, down, go lower than idle (I've seen as low as 2C!), and return to idle makes them seem unreal as does the lack of any other indication that there's something wrong. On the other hand, the fact that the spikes go higher and lower over time and as the PC gets hotter and the incident with the actual hot air seems to show that there is a problem even if the readings aren't entirely accurate.

Frankly, I wouldn't have bothered anyone with this but as support for Windows XP is ending soon I probably need to decide whether to buy a new PC or Windows 7 with an SSD and more RAM for this one.

 

Post a picture of the heatsink and fan unit you have currently on your motherboard, along with a picture showing the case fan. Also is the case inside a cabinet, or out in the open, with nothing stacked on top of it, or around it. How warm is the room that you are operating it in? Is there air conditioning in the room that you are using it in?

Even if you went with a SSD, it still will not lower the temp of the CPU. Here are some various Liquid cooling systems http://www.newegg.com/Water-Liquid-Cooling/SubCategory/ID-575?Tpk=liquid cooler

 

You don't need liquid cooling - you need to determine why your system is getting so hot in the first place. OEM coolers are perfectly suited to cool the CPU they come with - even with mild to moderate overclocking. It is the case's responsibility to ensure the CPU cooler is getting a sufficient supply of cool air.

Is your case interior and all vents free of heat trapping dust? Is your CPU heatsink clean? Are all fans spinning freely? How many case fans? Can you add more, or larger case fans?

Actually, most of your temps are quite good. CPU temps over 60° is not good, but even there (according to your first screen shot) with 65°C that is not extreme. Still, I want my CPU temps to stay below 60°C. GPU temps, on the other hand, tend to be considerably warmer so the 70°C does not bother me (assuming the GPU cooler is functioning properly).

105°C is extreme. That is above the boiling point of water!!! Note "system" normally indicates the chipset's north bridge. Some motherboard have a small (typically 40mm) fan on a heatsink, some use a large passive (no fan) cooler and depend on air turbulence created by the "expected" OEM cooler (another reason not to use alternative cooling).

Sadly, many HW monitors have problems properly identifying and labeling sensors and as a result, often give erroneous readings. For that reason, I recommend using the hardware monitor that came with your motherboard (or downloaded from the maker's site).

You might also try Speccy (from the makers of CCleaner) and see what it says.

PSU fans always exhaust heat out the back of the PSU - thus out the back of the computer case (regardless if top or bottom mounted). And PSU fans should not be counted as case cooling (even though they do help draw cool air in).

As for a new computer - get a good case. I like Antecs as they typically have excellent cooling capabilities and many come with removable, washable air filters - a must-have feature for all my cases.

 

I have borrowed a camera and was finally going to get around to doing this. Unfortunately I noticed a strange process with a long nonsensical name in the task manager while updating AVG on wednesday and got an e-mail from a website I'm registered with saying some e-mail address I've never heard of had requested my password be reset. Naturally, some malware scans are in order before I plug someone elses equipment into my PC.

The case is out in the open, the back 9 inches away from the wall. The monitor's close but not near any vents. I occasionally have some paper stacked on top of the case but since I installed the new power supply, only a few sheets at most; hardly anything that can have any effect. I've never noticed the top of the case get warm anyway.

The room is warm with it being summer but there are large double doors directly to the outside, one of which is nearly always open because there are no other windows and air conditioning.

I gave too much information here; the SSD is just for faster OS booting and some much needed extra storage. Also, I doubt there is a need for liquid cooling since the computer is 5 years old and wasn't very high spec at the time.

Admittedly, the case isn't particularly clean nor is the heatsink free of dust but I attempted to clean it with compressed air before I replaced the power supply and managed to remove some dust. I also completely cleaned the exhaust fan. If there was any change in temperature, I would have expected a slight improvement. The dust is old news though I suppose my attempt at cleaning could have pushed some dust closer to a sensor or something.

I've never taken the heatsink off for a more thorough clean for fear of repeating an incident in which the heatsink fell off inside the case the second time my computer was delivered (it took 3 attempts for the vendor to deliver a working one).

As for fans, my heatsink is a thermaltake blue orb mini with what seems like an 80mm fan (possibly slightly smaller) and my exhaust fan is 120mm. As mentioned in my previous post I installed a front case fan; this is also120mm but it had no effect and is blocked by the front of the case though there is a small gap in the bottom of the front through which it can probably draw some air. There is a 80mm fan slot on the left side of the case in front of the heatsink. I didn't know whether this should suck or blow air which is why I got a front case fan. The case's air flow seems good as I've noticed any effect to temperatures from taking the side off.

Thankfully, 65C is usually as high as coretemps get and it's been doing this for a few years. I've seen the occasional 67C and that might be cause for panic since 72C is meant to be the absolute limit for this processor. Improving this was the reason I got the front case fan. GPU temps aren't really a concern since it's limit is meant to be 100C though it has definitely got hotter over time. I use EVGA precision to give it a much more aggressive fan speed profile.

I think the motherboard manufacturers own softare, Asus PC Probe gives the same readings as HWMonitor anyway does hence my using that. Nontheless I'll retry it and Speccy once my malware scans are done.

 

I have had mixed results with side mounted fans, but generally I don't use them because they tend to disrupt the desired "flow" of air moving through the case. The exception would be if your side fan uses a tube to direct air directly onto the CPU fan or GPU.

In any case the side panel should blow cool air in, then observe temps to see if helps or not.

For sure, keeping heatsink clean of heat trapping dust is essential.

The problem with removing a heatsink fan assembly from the CPU is you ABSOLUTELY MUST then thoroughly clean the mating surfaces and properly apply a fresh new layer of TIM (thermal interface material). You should NEVER EVER reuse TIM once cured, and the bond is broken.

 

If you are stating that the room is warm, the next question is what is the actual physical temp in the room. If it is say same as outside temp of 85, the temp of the computer will spike up higher, because all it is doing is pulling in warm air.

Electronics do not like warm spaces. I keep our house no warmer than 72 during the day, 68 at night, and have not had any issues with any electronics in our house spiking with high temps. Suggest to start cooling the house, so the electronics are happy. Really for most of the time my laptop runs around 110, may spike to 132 if it is doing something that will push the cpu to work harder. My Lenovo ix2-4 NAS runs around 165 for the CPU, fan around 129, due to that is the nature of that device. Rated to max out at 215º f, before it shuts itself down. That unit is in the basement and right now it is 69º f down there. I am not worried about the temps on that unit, because it was designed to handle high temps.

 

Yeah, before Intel changed things with their Core 2 Duos and 130W CPUs were the norm, I had to install a window air conditioner in my office because the 3 computers in that room were pumping out some serious BTUs. Before the windows AC, I would have to set the house to 65°F to keep that room below 75°F. Not good when the power bill came in.

Sadly, some folks think a fan cools a hunk of metal (heatsink) like it cools living tissue (people). Not so. Only living tissue is affected by the "wind-chill effect". A fan can only cool an object to the temperature of the air it is blowing. If the room is hot, the fan is blowing hot air.

 

Since it's been an INSANE amount of time since I last posted in this thread (10.5 months!) this post will include a recap so no one has to read the rest of the thread. (along with all the new information obviously)

• Last year I replaced the power supply in my PC as the old one had blown out. Since then the sensors HWMonitor call Auxtin and Systin have been showing temperatures that quickly spike up very high (the highest I've seen 105C) and very low (the lowest I've seen is 2C, odd since it idles around 45C!) when using the GPU e.g. games and HD videos. The temperatures quickly return to normal when the offending task is closed e.g. when I tested a game today one reading changed from 73C to idle of 42C almost instantly. I would like to know whether these readings are real as the excessively high temperatures would be a problem that needs to be solved.
  
  
• The case's airflow is good as opening the case has no apparent effect on temperature. The CPUs heatsink has 80mm fan and there are 120mm fans on the front and back. Liquid cooling isn't needed as the computer is more than 5 years old and wasn't high spec for its time.
  
  
• Speedfan concurs with HWmonitor and was used for the graphs in previous posts. It was suggested that I try Speccy and the motherboards manufacturers own temperature utiliity, in this case ASUS PC Probe 2. Speccy showed different readings but they spiked up and down just the same and within the same range. I couldn't get ASUS PC Probe to run; there was a way to get it to run as I tried it a few years ago but I can't remember or find it again.
  
  
• I was asked to provide pictures of the inside of the computer. I got them relatively quickly but never got around to posting them as photobucket didn't work on my PlayStation 3 browser and didn't want to use my PC as I thought it was infected. I was kind of correct though it wasn't serious or anything to do with the temperature as the (almost) malware wasn't downloaded until months after I noticed the temperature problems. Here are the pictures:
  
  Computer and monitor
  Inside the PC
  Close up of the heatsink
  
  
• As you can see in the third picture there was a ton of dust (I had no idea how bad it was). Most of the gap since my last post was from deciding not to post again until I had taken off the heatsink, cleaned it, and re-applied thermal compound since that was probably going to be the next step anyway. Obviously, this hasn't solved problem since I would have had no need to write this post otherwise.
  
  
• I was also asked about room temperature. I bought a cheap thermometer between the PC tower and monitor and it has not reported any temperatures higher than 26C in the summer months.

 

No, they are not real. How do I know? Because 2°C = 35.6°F and there is no way your room temperature is 3.6° above freezing (fans cannot cool objects cooler than the ambient/room temps). And 105°C = 221°F which is well above the boiling point of water, and way too hot, even for many GPUs. While some CPUs may be able to tolerate those extreme temps for short periods of time, it is not likely they will be stable.

I use and recommend CoreTemp for my real-time CPU monitoring.

5 years old has nothing to do with it. In fact, CPUs from back then tended to run hotter than today's CPUs. The fact is, liquid cooling should NEVER be "needed" (except, perhaps, with extreme overclocking). Remember it is the case's responsibility to provide a sufficient supply of cool air "flowing" through the case. And with your case supporting at least 1 120mm fan in front and 1 fan in back, that "should" be plenty (assuming you keep the system clean of heat-trapping dust) and normal clock speeds.

As for your images, two thing. (1) Clearly your CPU heatsink needed cleaning. I do NOT, however recommend pulling the heatsink as that introduces the potential for physical damage and ESD problems, as well as the need for thorough cleaning, then a fresh and clean application of new TIM (thermal interface materials). Instead, I recommend using an air compressor properly fitted with a suitable moisture and particulate filter, or use a can of compressed dusting gas. Either way, do it outside. And note those cans are NOT full of compressed "air" but rather hazardous gases that should not be inhaled. (2) I would try to "dress your cabling". That is, try to tuck away some of those extra power cables under your optical drive. You want the impact your cables impose on that desired front-to-back cool-air flow to be as minimal as possible.

Side note (totally unrelated to any heat issues): It looks like you only have 1 stick of RAM installed. For many years now, most computer motherboards have supported either dual or triple channel memory architecture (most commonly, dual). Since your board has two slots, I feel safe to assume that board supports dual channel.

Ummm, sorry - freaking out for a second - it is raining hard in my front yard, but not a drop has fallen on my deck in my back yard. :confused

Anyway, back on track. Dual channel allows the system's memory controller to utilize RAM at nearly twice the speed. This means 2 sticks of 4Gb will perform at nearly twice the rate of 1 stick of 8Gb. The "catch" however, is that both sticks in each pair MUST have identical specs (preferably the same model number from same maker).

Does your computer perform well otherwise? Are you experiencing sudden shutdowns, lockups or rebooting?