Core 2 processor issue.

[Fossaw]

One of my systems has an MSI P35 Neo2 motherboard with a 2.4 GHz Core 2 Quad processor, and all the usual other bits and pieces. It had been running 24/7 processing BOINC jobs, with a very mild overclock, around 2.6GHz, but with a very serious, big Zalman heatsink, since the Core 2 came out, about 2006/7.

Earlier this year, it started having trouble, it would crash once a week or so. The interval between crashes shortened, to daily. Now, at stock speed, it will not run for an hour with four BOINC jobs going.

It would seem to be dying. I seek opinions now on whether to replace the processor, or rebuild the machine with an i7. Obviously, the chip replacement is considerably cheaper than the requisite motherboard, processor and memory of the upgrade, but the performance upgrade is not huge, so perhaps the processor chip will suffice.

Opinions.

[Mario F.]

Well, I would say there is always a gain from keeping your hardware up-to-date, even if for that specific task the performance increase won't be significant to justify the investment. You have more options open to you if you later plan to repurpose that machine and you won't have to buy a third processor.

Another issue you may want to give a thought is the price/performance ratio of a 2.4 GHz Core 2 Quad and a newer i5 or i7. It is likely that an older i5 or i7 SandyBridge offers a better price/performance ratio than the even older Quad core. Processors tend to evolve like that and in the domain of discontinued processors, newer ones tend to be cheaper relative to their capabilities than much older ones. Of course the more recent processors still in production are all outside this trend. Their price is still inflated.

The need to change the mobo does factor in, particularly if this will force you change other components too. But only you can really answer your question. Make a comparative analysis and in particular measure the cost/performance ratio of the overall purchase. Use those results to weigh your decision with how advantageous will be for you to maintaining a more modern system.

[Nominal Animal]

Have you monitored the CPU temperature and/or case air temperature?

It could be something as simple as a borderline seated CPU heatsink. Removing the heatsink assembly and thermal paste from both the CPU and the heatsink, and applying new thermal paste (not too much!) could help.

It could be due to reduced airflow (fans aging) in the case, causing significant rise in the case air temperature, reducing the cooling ability of the heatsinks (both CPU, and various heatsinks on the motherboard). In my experience, the fans either work well, or they make noise and eventually stop altogether; making sure the case is not caked with dust, and the case air temperature is not too high above room ambient is enough. I do update my fans every few years, though, because there are new ones with less noise at higher airflows.

If it is none of the above, it is most likely some capacitors or voltage regulators on the motherboard aging, and failing to work under prolonged load. I've had one voltage regulator explode on a Gigabyte motherboard; the symptoms were eerily similar to thermal issues, but the temperatures were well within acceptable range. This would, of course, mean that changing the CPU would not fix the issues; you'd need to change the motherboard, too.

In general, I don't recommend using a motherboard for more than 5-7 years anyway. The electrolytic capacitors start to dry out, and they're usually not the most reliable kind either. It also seems that added airflow near the processor, on the motherboard surface, is needed to keep the voltage regulators (usually as a bank between back connectors and the CPU socket) working under load for years on end.

[Fossaw]

Mario:

I agree, to a certain extent, however, if you are running several machines, keeping all of them up to spec becomes a constant job. What I tend to do is juice up my oldest machine each autumn, it would probably be this one actually, a Pentium 3, but it is totally reliable 24/7/365 for what it does. If I was to seriously do up the Core 2 system, this one would have to wait. The jump from Pentium to Core processors for science work is very appealing.

NA:

I tend to be careful, not fanatically, but I believe, sufficiently, to be confident in my maintenance routines to discount dust and crap in/on the machine. I have equipment running here with electrolytics more than 40 years old. Cheap junk caps fail, sure, but reasonable quality components on good boards go on for years.

[Mario F.]

I agree, to a certain extent, however, if you are running several machines, keeping all of them up to spec becomes a constant job.

Oh absolutely. I didn't mean doing this on a constant basis. Only periodically and, frankly, only depending on your use case. I my own case I don't even consider replacing a machine before I'm forced. I tend to squeeze every little bit of juice of it until it either breaks or I'm forced to upgrade because of some new software. Because of that, my machines tend to become 5 or more years old. My current main computer is a 6 year old i5 760 Lynnfield (quad-core). I have yet to make any replacement to it. Memory, disks, GPU, all still the ones that came with it. But because of that I do tend to make a substantial upgrade when I finally decide to replace my main. I always refuse to buy the latest, but I always go for the one before the latest or the one before that.

I tend to be careful, not fanatically, but I believe, sufficiently, to be confident in my maintenance routines to discount dust and crap in/on the machine. I have equipment running here with electrolytics more than 40 years old. Cheap junk caps fail, sure, but reasonable quality components on good boards go on for years.

Electronics always have been great. With good maintenance and handling, our computers can last decades even after having experimented a period of a few years with intensive use. I still have a working 16 year laptop and my oldest working PC is 20. Ebay is full of stuff 2 or 3 decades old that is still working and has never seen a maintenance. But where we lost is in the shells. The quality of the materials that shell the electronics has been reduced. Often we have perfectly functional computers with corroded cases, laptops working perfectly but with a keyboard dead or with missing keys, fans that stop working, corroded rubber components, plastic that has gone brittle, etc. However not even 20 years ago we used to produce lifelong materials; 20 year old keyboards that still work today, mice that don't seem to want to stop working, computer cases seemingly immune to time or heat, etc.

[Fossaw]

I got a used 3GHz Core 2 on an ebay auction, will stick that in, (as soon as I can get some Arctic Silver...), and let it run.

Cheers folks.

[MutantJohn]

This reminds me, my gf finally ordered all the parts so we can build her a computer. It should be pretty decent too. I think she's getting like a 6 core AMD and a GTX 960 and a SSD. Should be a pretty great machine. I was also going to help her build it because I built mine like 4 - 5 years ago.

Turns out, Microsoft sells bootable USBs for installing their OS now. I thought that was pretty neat. Last time I installed Ubuntu, I had to manually make a USB bootable and then back again. It was a giant pain.

[Elysia]

Making a bootable USB drive with windows on it is very easy today.

[Fossaw]

I put the new old (!) processor in and believe I have reassembled the system to it's former state, the heatsink is fiddly and takes a suprising amount of effort, but apart from another fan that was in the way, I reckon it is back to where it was with the new CPU chip.

The system doesn't start, nothing at all, no mobo "Hi I'm an MSI" message, nothing at all. The PSU, which isn't very old, seems to be happy firing up for a few seconds, then shutting down and trying again, and again, and again...

So what gives here?

[Nominal Animal]

If I make any suggestions, will you promise to not get offended? This list is not my suggestions to you in particular, but more like the list of things I do when having that happen to myself.

Sniff the motherboard for any wiff of magic blue smoke. The scent is pretty unique. Check the voltage regulators between the CPU and the back panel connectors, too. (I've had one literally blow up, creating a hole on the top of the chip package, on an otherwise good-quality Gigabyte motherboard.)

Check the motherboard electrolytic capacitors for leakage. (I don't know of any motherboard manufacturer who actually uses the highest-quality electrolytic caps. Quick search indicates others have had issues with MSI motherboards. Many manufacturers do use Japanese ones, but it's easy to fall for counterfeits; I don't really blacklist a manufacturer if they happen to have used a bad batch. I've never replaced any myself, though; I've just replaced the motherboard, instead.)

Unseat the CPU, and verify all the legs are straight. (I've had *one* leg of a CPU twist on me, once. It was a normal CPU ZIF socket with a locking lever, and the CPU was correctly oriented, too. Don't know how it could have happened, but it did. Managed to fix it, though.)

Make sure the CPU heatsink is correctly oriented, fully seated on the CPU, and not partially on top of the socket plastic. (Last time I noticed the heatsink was a bit wobbly, I had somehow managed to get one side on top of the socket plastic, without noticing. Caught it because the heatsink holes didn't line up properly.)

Check the seating on the motherboard power cable(s). (Verify you didn't forget the 4-pin P4 power connector if the motherboard uses one.)

Verify you didn't put the power-on LED into the reset button pins. (I'm very careful, but goofs happen. You may laugh, but it could happen to you, too.)

Verify there are no shorts to ground on the backside of the motherboard. (I've dropped a screw there without noticing, but caught it before power-on: it rattled when I turned the chassis back upright. Now I check.)

[Fossaw]

I don't get offended by people trying to help me. The brain damage I received when my heart stopped for 46 minutes has meant I need help to do some things. Before my heart attack, things were different, but the medics forced me back into this miserable existance, I need help some times.

No smoke or smells. Looking around the board, and clearing some dust bunnies etc. nothing looked out of the ordinary. The capacitors look normal.

The site you linked shows many problems with MSI boards, but the board I have does not appear when searching. Of course, that just means nobody has reported or questioned about it, not that it is a solid piece. All my other systems have ASUS boards, I don't remember why I put an MSI in there.

The CPU does not have any legs, there is a large array of contact points that sit on an array of raised metal points under the chip. It has a couple of protusions which fit into matching slots in the socket to prevent misorientation.

The heatsink is mounted correctly, and firmly bolted down. It is a great cooler, but an awkward cuss to work with. Even if it were absent, the system should do something, but all I get is the fans starting running for varying length short periods, then stopping.

When dismantling the system, I wrote down each cable as I took it out, and returned each in reverse order when reassembling. They are all seated as they should be.

The flock of twisted pairs from the front panel are connected to a small chasis, (with one exception), which then connects to the mobo. I removed the chasis, not the cables, I know, these can be fiddly to work with. The other twisted pair has a different kind of plug and connects to a pin array to the side of the chasis, it only fits one way round.

I did not remove the mobo from it's mounting, there is nothing back there that was not there before. Certainly, nothing rattles etc. when setting it back upright and into it's place.

[Mario F.]

A short somewhere or a damaged PSU seem more likely.

Searching for a short, one thing you can do is trying to turn the computer on with only the barebones stuff connected to the PSU (the HDD and the CPU fan). If your CPU has embedded video, remove also the GPU. Disconnect all leds and any remaining fans. Disconnect DVD, floppies, whatever. If it it seems to work, then you just start plugging stuff back in from most important to least important until you find which one is shorting.

Before going through all that trouble you may wish to put that PSU on another computer and see how it behaves. That will rule out (or in) the PSU immediately.

[Fossaw]

Before going through all that trouble you may wish to put that PSU on another computer and see how it behaves. That will rule out (or in) the PSU immediately.

Yes, I can try swapping the PSU with a PSU in another machine. I seriously doubt that this is the source of the trouble though. The PSU in there is relatively new, and I routinely put a more powerful PSU in machines than they are likely to need. This gives me a margin to work with.

[phantomotap]

The CPU does not have any legs, there is a large array of contact points that sit on an array of raised metal points under the chip. It has a couple of protusions which fit into matching slots in the socket to prevent misorientation.

O_o

I love the LGA sockets.

The PSU in there is relatively new, and I routinely put a more powerful PSU in machines than they are likely to need. This gives me a margin to work with.

We've seen plenty of PSU claiming big watts fail quickly due, if I recall, to lousy capacitors or overshared rails.

*shrug*

I guess it would suck in any event because of the implication, but I have to wonder if you've tried seating the old processor.

Soma

[Fossaw]

I've been avoiding trying, it is such a pain in the arse to do.