Message boards : Questions and problems : New motherboard fried, not sure if BOINC is to blame
Message board moderation
Author | Message |
---|---|
Send message Joined: 28 Jan 18 Posts: 1 |
Hi guys. I've been a volunteer for a few years. Recently, I bought a Dell XPS 8920 with 16 GiB RAM, an i7-7700K CPU @ 4.20GHz and a GeForce GTX 1070. I decided to assign this machine to some GPU tasks, mostly from Collatz Conjecture and GPU Grid. I kept it on while sleeping and when at work, always crunching those tasks, leaving 100% of my CPUs and GPU available for use. Less than a month after buying the computer, I came home one day to find it turned off and it would not turn back on. Since its warranty was still valid, a Dell technician came, diagnosed the problem as motherboard failure and replaced it. Since the most hardware intensive software running on the machine at the time was BOINC, I started questioning whether this type of thing may happen because BOINC was running when the computer stopped. Maybe I configured BOINC wrong or something. I know that, in case the CPU overheats, the computer turns off before any possibility of hardware damage, but I never had a computer with a decent GPU and I don't know if the behavior is similar in this case. I also don't know if number crunching a whole day may cause my motherboard to fry. I was wondering if someone could clarify whether it is possible for BOINC to cause this type of hardware failure or if I should blame a faulty motherboard and jsut install BOINC again. Thanks in advance. |
Send message Joined: 29 Aug 05 Posts: 15555 |
Welcome to the boards. Sorry to read about your motherboard, but I can say with great certainty that BOINC was not to blame. BOINC by itself is just a managing program, it in itself does not add any stress to your system. You can easily test that by installing BOINC and not adding any projects, then you can leave BOINC on the system ad nauseam and it will only fail again if the next motherboard is also bad. In all my years of computing I only ever had one motherboard go bad (and when it went with a bang, it took my RAM and CPU along with it to the great beyond). This was due to bad capacitors, a thing that was a problem some years back. These days they're all military grade, and gold capped instead of cardboard. Usually when a motherboard goes it came with micro-fractures already that increased over time due to heavy use (read: heat & cooling cycles). The project applications will put a heavy stress onto the CPU & GPU, although that does depend on the project and whether they optimised their applications, and if so what kind of optimisation was used (integer mathematics, floating point maths, using simd, using just CAL/CUDA or OpenCL.) Then it depends on how much cooling your system has, and the amount of maintenance you spend on it. A case with 40 fans will grind to a halt just as fast as a case with 4 fans if you never clean them in a dusty (lived in, animal rich) environment. So go ahead and reinstall BOINC. But before you restart your projects, check your maintenance and decide if a little less intense usage is just as well. |
Send message Joined: 2 Jan 14 Posts: 276 |
My DVR has been crunching BOINC in the living room for about 8 years now, 24/7 and the only issue I've had with it is a hard drive and power supply. If you decide to come back to BOINC, I would monitor your CPU temperatures and make sure you're not sustaining high temperatures. You can usually get the maximum allowed temperature from the manufacturer's data (Intel Ark shows this as Tjunction or Tcase) and then I use a program like TThrottle to adjust BOINC on the fly to keep a set temperature. I usually set it 10-15 C below the max allowed temperature. You can also use that program to control unwanted fan noise. Since you're not crunching 24/7, I would definitely make sure your CPU and case cooling is adequate, but a motherboard failure that soon in a computer's life is more likely just a manufacturing defect. Nice that you got it covered under warranty. My Detailed BOINC Stats |
Send message Joined: 11 Oct 10 Posts: 13 |
You could monitor your temperatures to make sure your system is "overloaded". SpeedFan is good for the CPU and other system components. GPU-Z is good for the GPU. Both programs can create a log file, |
Send message Joined: 8 May 10 Posts: 90 |
Let me participate in this BOINC releated horor story sharing festival. Once I've rubbed through crappy DIMM. Indeed, it took 5.5 years, but, because I'm running 24x7, it is clearly fault of BOINC :D I'm counting for science, points just make me sick. |
Send message Joined: 23 Feb 08 Posts: 2492 |
Boinc did terrible things to my system. Because it made it warm it forced me to open the box and blow out the dust on a regular basis. Then it dried out the faulty electrolytic capacitors on the motherboard. Took about 4 years too. After that it was toast. Oh, on the other system, the brand new one, it so stressed the CPU it failed. That 10 days out of box. I know it was responsible for that. Never mind Intel making a bad CPU. PS the replacement CPU has been going 5 years now. |
Send message Joined: 27 Jun 08 Posts: 641 |
Hi guys. I've been a volunteer for a few years. Recently, I bought a Dell XPS 8920 with 16 GiB RAM, an i7-7700K CPU @ 4.20GHz and a GeForce GTX 1070. Dell makes good products and has good tech support and forums both "Owner forums" and Dell community. That being said, I discovered that my "Sensor Control" program does not regulate and only reports temps on my 1070 gpus. It does a fine job with the cpu sensor, disk sensor, motherboard sensor etc. It automatically scales up or down all the fans speed BUT GPU NEEDS PRECISON X OR AFTERBURNER to set its fan speed. Within hours of posting on the AlienOwners forum I got responses advising to use MSI's Afterburner. Dell uses MSI as OEM for graphic cards supposidly. Easiest thing to do for tech is test the power supply and if it is good, then simply replace the motherboard. You get to reinstall windows which probably fixes the original problem. They run a diagnostic on the motherboard and then sell it as refurbished. |
Copyright © 2024 University of California.
Permission is granted to copy, distribute and/or modify this document
under the terms of the GNU Free Documentation License,
Version 1.2 or any later version published by the Free Software Foundation.