Bad computer luck

[Brian Altenhofel]

Man, this week has been unlucky for me.

Thursday night, my CPU or motherboard in my desktop computer died suddenly. Not overheating, no power issues. Just died an instant death. So I ordered a replacement CPU and motherboard, and opted for water cooling (I have overheated it a few times in the past - likely a contributing factor).

I had a part of a project nearly finished, but I hadn't pushed the changes up yet. So I figured I'd work on another part over the weekend while suffering through the limitations of my laptop.

Then the hard drive in my laptop died a noisy death. Finally, that SATA/IDE->USB kit I bought on clearance several years ago (and had never opened) came in handy! It took several hours (because of all of the damaged sectors) to clone the old drive to a new Samsung 850 EVO SSD and to repair the recoverable corrupt files, but I'm finally up and running again.

A friend of mine told me to not come near any of his machines until I have clearance from a doctor that I am free of whatever virus I seem to have been carrying the past couple of days.

[Fulcrum]

Sounds like you got a visit from Murphy. Did he happen to leave any gremlins behind?

[Freelancier]

Sometimes it's not what you think....

Back in July, my main workstation stopped working... we had a power outage that lasted about 8 hours and when the power came back, the box did not and had some LEDs on the motherboard indicating that the motherboard had failed. This happened the day before I was to leave on a 10-day vacation and would need to access it remotely in order to support clients. Huge fire drill, took my kids' older PC, slapped my disks into it, fired it up, fixed the driver issues so that it could stay up. Contacted customer support for the system, they told me to call the manufacturer because it would be faster to get a new motherboard; motherboard manufacturer would not do a swap without getting my board first, so take the board apart, box it up ship it off. Crazy day, no work got done. Left for vacation, when I got back, the new board showed up that day, I put my workstation back together, worked great... until the next power outage. Same issue. Hmmm...

Did some more research on the motherboard. Found a suggestion, did that and the board came back up. The suggestion: remove the power cord, unplug all the USB connections (I have 6 USB devices on the system), then power it up and re-attached the USB devices. Weird, but it worked.

[Brian Altenhofel]

Did some more research on the motherboard. Found a suggestion, did that and the board came back up. The suggestion: remove the power cord, unplug all the USB connections (I have 6 USB devices on the system), then power it up and re-attached the USB devices. Weird, but it worked.

That sounds like either an undersized power supply or a motherboard that doesn't allow enough power to be routed to the USB ports for everything to be turned on at once. Just like with most electrical items, there is a spike when a device is first turned on (and it can be inconsistent, too). All six plugged in at the same time may be below the power limit while running, but may pull too much and trigger overload protection when all are in their initial startup at the same time.

Mine is definitely a dead CPU or motherboard. I've breadboarded it to make sure. There was a burnt electronics smell when I pulled it out, but I couldn't nail down the source. The processor doesn't show any outside signs of catastrophic thermal failure, and all capacitors on the motherboard are intact. The initial surge from power on will spin the fans for a split second as expected, but they don't continue, which points to the failure likely being related to the CPU since the motherboard uses the CPU to control the fans post-startup. If I had a known good CPU or motherboard to swap, I could be certain which one is the problem, but I don't.

To be sure it wasn't a power supply issue, I tested the power supply both idle and under load, and voltages were well within spec. I will be getting a new power supply anyway because the water pump and additional radiator fans will put me close to the limit of the existing power supply.

[Brian Altenhofel]

On the plus side, I notice a huge difference in my office without the computer running. I had thought maybe there was an issue with the duct work because my office was always the second-hottest room in the house though the air handler is in the closet in here (hottest is bedroom because of two walls of single pane windows). Now it's the coldest as expected.

[nealrm]

First, I think this thread should be deleted, purged from the systems hard drives and then the hard drive removed and crushed, just in case what Brian has is contagious.

If you don't have a UPS attached to your computer you should. Surge power strip are just a step above worthless. A good UPS will not only protect you from spike, but make sure that your computer gets a steady 120 volts. If keeping you system up is extremely important, look into having a whole house surge protector installed. If will handle the a much larger spike and protect the whole house.

I have to say the new liquid cooling kits do a great job and are almost as easy to install as the fans. You do need a case that will support the unit so I'm not sure how well they will work on store bought units. Mine keeps the temps at a new and cool 47C even under high loads for long periods. (I also have 7 cases fans)

[Brian Altenhofel]

First, I think this thread should be deleted, purged from the systems hard drives and then the hard drive removed and crushed, just in case what Brian has is contagious.

If you don't have a UPS attached to your computer you should. Surge power strip are just a step above worthless. A good UPS will not only protect you from spike, but make sure that your computer gets a steady 120 volts. If keeping you system up is extremely important, look into having a whole house surge protector installed. If will handle the a much larger spike and protect the whole house.

I have to say the new liquid cooling kits do a great job and are almost as easy to install as the fans. You do need a case that will support the unit so I'm not sure how well they will work on store bought units. Mine keeps the temps at a new and cool 47C even under high loads for long periods. (I also have 7 cases fans)

I run an APC Smart-UPS. I definitely notice how dirty the incoming power is here, especially in the summer. Pretty much if it's electronic I run a UPS or conditioner.

Finally got my desktop back up and running with watercooling.

866

Idle:

867

Previous idle was typically 8-10 degrees above Video 1. I haven't tested it under a heavy load, though right now it's running at ~22C installing Linux updates, indexing a large project in PHPStorm, and three active virtual machines.

I'm sure the root of my previous heat issues were my video cards generating so much heat that the CPU heatsink was trying to dissipate heat into already heated air and just could not keep up, even with four 90CFM case fans trying to move the hot air out.

[nealrm]

Something is not right with that temp reading. 16.5 C is 61 F. So unless you are running this in a meat locker I don't see how it can be correct. This type of cooling can't drop the CPU temp below room temp, so you should be seeing numbers around 32-40C at idle and 47-60 under load. It is possible that your problem is not too much heat, but that the system thought the CPU was cooler than it really was. Depending on the cooling configuration, that could result in the system powering down the fans.

Interesting setup on the liquid cooling. Did it come with the external attachment or was that a modification you made?

[Brian Altenhofel]

The kit came with the brackets for external mounting so that you can choose whether to install in or out of the case (or, in my case, have too small of a case).

I was using the wrong sensor for the previous picture. It turns out the onboard sensor on AMD chips actually gives a value according to some algorithm to tell the fans when they really need to run.

This is the correct idle:

868

After running mprime (Prime95) for a while:

869

Shortly after stopping mprime:

870

[nealrm]

That looks good.

[nowak1]

You definitely had a bud luck. But do not worry everybody sometimes has it. One time a I was making a big assignment for my university project and during writing (project was almost done) I went to kitchen to drink something. When I went back I saw my cat sitting on my backspace button. Everything was gone.