I had my computer re-built and upgraded — new motherboard, processor, case, fans…and it ran great for a little over a week. My house line got improved and I have a dedicated 20amp circuit just for the computer and the massive UPS for it.
For the last week, I’ve had random freeze-ups that seem to be caused by overheating. At least the computer won’t turn on immediately after freezing up, then comes on after a few minutes….usually. I planned to get an extra fan, but today, it’s refusing to come back on at all.
My usual helpers are unavailable. Would anyone nearby be competent and willing to help me troubleshoot it?
Toggling all internal resets fixed it for now.
Rebuilt? Hey, did they happen to replace your power supply with a new one of equal or greater power rating? Thats the first thing that would come to mind. Also your RAM (memory sticks), upgrade their or left alone?
scott
Old RAM, new 1050W power supply.
Do you have a box fan or other large fan? If so take the side of the case off and aim the fan at it and crank it up to max then use it for a while. If it doesn’t lock up after a few hours it’s probably not overheating.
Having said that if the the CPU fan is the one that came with the processor it shouldn’t be overheating.
“Having said that if the the CPU fan is the one that came with the processor it shouldn’t be overheating.”
Wrong.
I don’t have a lot of experience with AMD hardware, but Intel’s stock heatsinks are absolute garbage. They are designed to keep the majority of processors cool enough that they last until their warranty expires. No more, no less. That means that they generally don’t have a lot of headroom; The one that came with my i7 920 (SLBCH) couldn’t keep up with the CPU at 100% load (read: 100% load = 100 C/TJ-MAX temps). Aside from this, their mounting hardware is ALSO garbage.
Aside from getting a high-quality power supply (and high quality parts in general), the most important thing to do is to get a CPU HSF unit that is superior to the stock unit.
Interestingly, that is completely contrary to my experience.
I use stock i7 coolers for both development (where parallel builds use all the CPU it has for 10-15 minutes at a time) and gaming machines (and not just me, all the dev machines at the place I work).
Nary a problem with heat, and ones I installed on 920-series chips that are well past warranty are still turning just fine.
Might have been bad batches of i7-920 fans or bad factory thermal paste, maybe?
The attachment hardware is much, much improved from the Core-2 or P4 days. They used to be nightmares, but I’ve never had a problem with an i-series cooler not mounting easily and well.
Oleg,
This may sound odd but I and at least one other person have had issues similar to this with of all things SATA cables. It seems the cable is “intact” but not transmitting data like it’s supposed to (bad contacts? Who knows) in at least 3 separate instances at different locations changing out one of these cables has resolved what had been a basically insoluble problem.
SATA cables are cheap and easy to rule out. I begin by disconnecting things like DVD drives and seeing if it corrects the problem. If it doesn’t it’s a good guess that cable isn’t the problem and it can become the “known good cable” if you don’t have a spare around.
Do with this info what you will. Good luck on resolving your issues.
Missed the “Refusing to come back on at all”
Ignore my last.
Please try unplugging the CPU and plugging something like a toaster or large fan in the outlet so see if your UPS is functioning correctly. Your PC may be drawing enough power to shut it down. If you were getting overheating problems, you should have gotten a thermal event warning.
UPS is fine.
I doubt the UPS is the problem; I’m assuming that he’s running a mid-range workstation, and that he’s not running multiple high-end graphics cards. As such, I doubt the PC is pulling more than 250-350 watts from the UPS at peak load. Very few quality UPS units would have a problem supplying that much power.
As far as overheating producing a warning, that would depend on precisely what component was overheating.
If the CPU or graphics card were overheating, then a warning *should* be displayed in most circumstances. However, if it were the PSU overheating, then I wouldn’t expect to see any kind of warning.
open case and look at the fans. if they don’t start moving as soon as you turn the power on and stay on replace the fans. if fans are okay take off the cpu fan and check for heat sink grease. if there is no or not much grease on the back of the finned aluminum block that may be the problem, or the clips holding it in place could be loose. otherwise it could be a general power problem. the house you’re in could have a bad ground system and not be a true ground. get a battery back up system if everything in the computer seems fine.
p.s. the facebook link is not working
The video card seemed rather hot.
Ground was checked ten days ago and fixed.
How nearby are you to 37013?
I have had this same set of symptoms after a homebuild, there was insufficient thermal paste between the CPU and heatsink. Once I cleaned and reapplied the paste to the heatsink the issue went away.
Insufficient? I just read an article that points out that the correct amount of heat sink compound is a very small amount. Just enough to fill the irregularities in the surfaces, no more. If you can get any to squeeze out, you have too much. Too much will cause overheating. (So will too little, of course.)
Generally speaking, it’s impossible to have too much thermal paste as long as the heatsink is installed correctly.
You want to put a small amount (somewhat larger than the size of a grain of rice) in the center of the CPU, and then put the heatsink straight down on top of it, and bolt it down. The tension with which the heatsink is applied is sufficient to spread the thermal transfer paste to where it needs to go. The goal is to get as thin a coating as possible without introducing any air pockets, which is why this method is ideal.
As a sidenote, a good aftermarket heatsink with sturdy mounting system which includes a good backplate, is vital if you want your computer to last.
Rule out the RAM first. If you have another working computer and CD burner grab http://www.memtest86.com/ and give that a go.
Dave
This is a great tool to use when debugging. Bad RAM can cause a system to behave erratically. Therefore, whenever your computer is behaving in an erratic manner checking the RAM is a good first step.
In my experience, this tool can often be found as a boot option on the live disks of most linux distros I’ve played with.
Oleg, any time you have problems feel free to email me, I would be glad to assist.
Agreed with NJDave, make sure the memory isn’t faulty.
Do note: I have seen memory that was definitively bad take more than 24 hours of continuous memtest before it finally started failing those tests. It would then continue to fail as long as the test still ran, but if you rebooted, it could be anywhere from minutes to hours of testing again before it showed faulty: replacing it resolved the issue.
Might not still be your problem in this case, but I’d be sure to let a memtest run continuously for at LEAST 8 to 12 hours before you determine that the RAM isn’t the problem.
“For the last week, I’ve had random freeze-ups that seem to be caused by overheating. At least the computer won’t turn on immediately after freezing up, then comes on after a few minutes….usually.”
When you say that it “freezes up” do you mean that it dies suddenly, or that the graphics displayed on the screen freeze in place and remain fixed until you cut power to the machine?
If the latter, I’d focus on the graphics card; If the former, I’d probably investigate the PSU, and maybe the UPS as well (there are some PSU/UPS combinations that don’t play well together) before moving on to other components.
What graphics card are you running currently? How long have you been running it? Assuming that it’s the same one you were running previously, was anything done to remove dust from the heatsink? Does the fan on it appear to be working consistently?
Additionally it would be a good idea to keep an eye on the temperatures of various system components in real-time; GPU-Z is generally a reliable application for monitoring the temperature of a graphics card…
I’m going to recommend not an extra fan first; but ordering something like Artic Silver for your CPU fan. What kind of thermal paste you use will FAR FAR FAR outweigh everything else as far as overheating. If you just have the standard thermal paste, you can be look at as much as 12 degreesC in difference.
http://www.maximumpc.com/article/features/geek_tested_17_thermal_pastes_face?page=0,1
When people tell me their computers are locking up from overheating; in order of recommendations from me:
1) Change to better thermal paste
2) Put heat sinks on your ram
3) another fan in the FRONT pushing the air into the box with the fan in the REAR blowing out.
The heatsink itself, along with how it is installed (and the paste along with it) tend to matter a lot more than the paste you use.
The temperature variation between most of the high end thermal pastes is generally 0-4C depending on who is paying for the testing; Heatsink install technique can cause significantly more variation in results.
An ideally mounted heatsink will be installed with even and proper tension, and won’t have any air voids in the paste. In my experience, a mounting system with a rigid backplate is required, as well as application of the thermal paste directly to the center of the CPU’s integrated heat spreader. Thus, when installing you use the pressure between the CPU and the heatsink to spread the paste between the two surfaces. This method tends to minimize or eliminate the initial formation of air pockets between the two surfaces, and ensures that the paste gets where it needs to be.