My computer doesn't like to think too hard
June 18, 2008 3:25 PM Subscribe
Why are my high CPU load applications crashing?
Lately whenever I run applications which uses a lot of CPU, they randomly crash in 0-2 minutes. These are application crashes and not system crashes.
They are extremely consistent and reproduceable so I can't even finish the tasks. I have been able to do these tasks in the past with no problems.
I've seen this in:
1) Transcoding my ripped CDs to flac
2) Compressing large folders
What can be the problem? I consider myself tech savvy am would be willing to try anything.
One thing that may be related is I had bad ram (memtest failed) last week and I replaced it with the same model of ram from Frys (memtest passed). I also tried eliminating the pagefile but the application crashes still persist. I'm on Windows XP SP2.
Lately whenever I run applications which uses a lot of CPU, they randomly crash in 0-2 minutes. These are application crashes and not system crashes.
They are extremely consistent and reproduceable so I can't even finish the tasks. I have been able to do these tasks in the past with no problems.
I've seen this in:
1) Transcoding my ripped CDs to flac
2) Compressing large folders
What can be the problem? I consider myself tech savvy am would be willing to try anything.
One thing that may be related is I had bad ram (memtest failed) last week and I replaced it with the same model of ram from Frys (memtest passed). I also tried eliminating the pagefile but the application crashes still persist. I'm on Windows XP SP2.
Make certain the memory is properly seated. If it is, try testing it without the new stick of memory.
posted by stavrogin at 3:40 PM on June 18, 2008
posted by stavrogin at 3:40 PM on June 18, 2008
Response by poster: Mike1024: How can I test the cpu? I wouldn't want to buy a new CPU and find it's not the issue. I've never had a CPU failure so I don't know what to check.
stavrogin: There is just one stick of memory and it passed memcheck. If it wasn't properly seated I would have imagined it would fail the memcheck.
posted by lpctstr; at 4:12 PM on June 18, 2008
stavrogin: There is just one stick of memory and it passed memcheck. If it wasn't properly seated I would have imagined it would fail the memcheck.
posted by lpctstr; at 4:12 PM on June 18, 2008
Are you sure the software isn't buggy? Can you run the same application on the same data on a different computer and have it run to completion?
posted by Class Goat at 4:15 PM on June 18, 2008
posted by Class Goat at 4:15 PM on June 18, 2008
Response by poster: Class Goat: yes I have tried the software on different computers. It works fine. 1) and 2) are also different programs. Both are widely used. 1) is flac, and 2) is 7-zip
posted by lpctstr; at 4:23 PM on June 18, 2008
posted by lpctstr; at 4:23 PM on June 18, 2008
Best answer: Heat maybe? If you're on a laptop, try getting more airflow. Try checking your CPU sensors.
posted by a robot made out of meat at 4:48 PM on June 18, 2008
posted by a robot made out of meat at 4:48 PM on June 18, 2008
It really sounds like bad memory, which can still pass a memtest. If you can't swap out modules, try running the test multiple times and using different programs. Even if it doesn't fail, exchange the memory at fry's for a new stick.
posted by stavrogin at 5:25 PM on June 18, 2008
posted by stavrogin at 5:25 PM on June 18, 2008
Could also be a faulty power supply.
posted by stovenator at 6:11 PM on June 18, 2008
posted by stovenator at 6:11 PM on June 18, 2008
Best answer: In my experience there are a couple real possibilities, most have been stated here.
1. Heat. I have a huge HUGE heatsink on my prescott 2.8 and was getting heavy-load crashes. A bit of arcticsilver totally cleared up the issue. Get in there with some canned air and clean it out...consider taking off the heatsink and cleaning the top of the chip with some alcohol on a qtip, then reapplying thermal grease.
2. Ram. Could be a bad stick that's really only reacting under heavy load/heat. Memtest86 is great, but won't always highlight the issue, but it's a good starting point. A Ubuntu live disc has the memtest86 function built in...pop in the drawer and reboot. This *could* also be a paging issue, but I sort of doubt it. The fact that you changed the stick *could* indicate a bad slot...you *could* try putting no ram in each slot, one slot at a time, and see if you can still reproduce the error. I once had a ram stick that took me a year to diagnose. Memtest86 passed it, and linux actually can skip bad sectors in drives/ram sticks so you don't know they're there...that was a nightmare.
3. Something horribly wrong in your windows install. If all else fails, backup/reinstall/nuke from orbit. I've given up and gone this route several (dozen) times.
posted by TomMelee at 7:25 PM on June 18, 2008
1. Heat. I have a huge HUGE heatsink on my prescott 2.8 and was getting heavy-load crashes. A bit of arcticsilver totally cleared up the issue. Get in there with some canned air and clean it out...consider taking off the heatsink and cleaning the top of the chip with some alcohol on a qtip, then reapplying thermal grease.
2. Ram. Could be a bad stick that's really only reacting under heavy load/heat. Memtest86 is great, but won't always highlight the issue, but it's a good starting point. A Ubuntu live disc has the memtest86 function built in...pop in the drawer and reboot. This *could* also be a paging issue, but I sort of doubt it. The fact that you changed the stick *could* indicate a bad slot...you *could* try putting no ram in each slot, one slot at a time, and see if you can still reproduce the error. I once had a ram stick that took me a year to diagnose. Memtest86 passed it, and linux actually can skip bad sectors in drives/ram sticks so you don't know they're there...that was a nightmare.
3. Something horribly wrong in your windows install. If all else fails, backup/reinstall/nuke from orbit. I've given up and gone this route several (dozen) times.
posted by TomMelee at 7:25 PM on June 18, 2008
Compression software generally uses very little memory. Are you seeing crashes with memory hungry apps like web browsers, etc?
What's the ambient temperature?
posted by ryanrs at 8:04 PM on June 18, 2008
What's the ambient temperature?
posted by ryanrs at 8:04 PM on June 18, 2008
Best answer: It's very unlikely to be the CPU itself. Question: did you see these crashes before you got the memory error and replaced the memory, or only after?
posted by madmethods at 8:58 PM on June 18, 2008
posted by madmethods at 8:58 PM on June 18, 2008
Response by poster: Thanks for all the input, everyone.
TomMelee: I tried using a different memory slot and the same errors occured. I agree it seems like it is probably the heat, as a few others have mentioned
madmethods: I had the crashes before I replaced the memory as well, but I thought they would be fixed when I replaced the memory.
ryanrs: The crashes don't happen with anything except high-CPU applications. Browsers, watching 4 videos at once, or playing games don't crash.
ryanrs: the room temperature is about 24 deg C
unmake: yes, with the memtest86 boot CD
From the suggestions, it's probably a heat problem. The only strange thing is Everest says the CPU is 30 degrees C no matter what is running. I have never seen the temperature be anything else and I have checked it many many times. I'll reapply some thermal heat and comment again if I've resolved it.
posted by lpctstr; at 9:16 PM on June 18, 2008
TomMelee: I tried using a different memory slot and the same errors occured. I agree it seems like it is probably the heat, as a few others have mentioned
madmethods: I had the crashes before I replaced the memory as well, but I thought they would be fixed when I replaced the memory.
ryanrs: The crashes don't happen with anything except high-CPU applications. Browsers, watching 4 videos at once, or playing games don't crash.
ryanrs: the room temperature is about 24 deg C
unmake: yes, with the memtest86 boot CD
From the suggestions, it's probably a heat problem. The only strange thing is Everest says the CPU is 30 degrees C no matter what is running. I have never seen the temperature be anything else and I have checked it many many times. I'll reapply some thermal heat and comment again if I've resolved it.
posted by lpctstr; at 9:16 PM on June 18, 2008
Best answer:
(Though I might be over sensitive to this because I've just replaced a laptop over this exact issue. Temperature sensor reported a temp of 0c so the fan never kicked in. Sensor was built into the motherboard...)
posted by Ookseer at 9:54 PM on June 18, 2008
The only strange thing is Everest says the CPU is 30 degrees C no matter what is running.That's suspicious. It might be that you have a bad temperature sensor that is not triggering the correct cooling. Open up the case, point your largest fan into it at full blast and see if you continue to get crashes.
(Though I might be over sensitive to this because I've just replaced a laptop over this exact issue. Temperature sensor reported a temp of 0c so the fan never kicked in. Sensor was built into the motherboard...)
posted by Ookseer at 9:54 PM on June 18, 2008
Response by poster: Uh oh, I think I really messed up...
I unseated my heatsink and applied thermal paste and put it back on. I started the computer and went into the bios and it seemed like everything was okay. Then I saw the temperature was 121 degrees C and I realized the heatsink fan wasn't spinning. I forgot to power it. I immediately powered down the computer, and plugged in the heatsink. However, when I boot now, it tries for a little bit (doesn't POST), and then shuts down and powers up again automatically (doesn't POST but stays on). But nothing appears on the screen.
I'm really panicking now. Any ideas what I should do?
posted by lpctstr; at 11:56 PM on June 18, 2008
I unseated my heatsink and applied thermal paste and put it back on. I started the computer and went into the bios and it seemed like everything was okay. Then I saw the temperature was 121 degrees C and I realized the heatsink fan wasn't spinning. I forgot to power it. I immediately powered down the computer, and plugged in the heatsink. However, when I boot now, it tries for a little bit (doesn't POST), and then shuts down and powers up again automatically (doesn't POST but stays on). But nothing appears on the screen.
I'm really panicking now. Any ideas what I should do?
posted by lpctstr; at 11:56 PM on June 18, 2008
Response by poster: If it helps, the cpu is a intel e6750, and the motherboard is an ecs g33t-m2.
posted by lpctstr; at 12:17 AM on June 19, 2008
posted by lpctstr; at 12:17 AM on June 19, 2008
Most modern CPUs will run at reduced speed when they get too hot, rather than damaging themselves. Give it a few minutes and try turning the system on again.
While you wait, verify that the cooler is on the right way - I once flipped mine through 180 degrees.
posted by Mike1024 at 12:37 AM on June 19, 2008
While you wait, verify that the cooler is on the right way - I once flipped mine through 180 degrees.
posted by Mike1024 at 12:37 AM on June 19, 2008
Response by poster: Mike1024: I had hoped so too but I waited about 30 minutes and started it and the same thing happened. It starts for a bit, shuts down, and then starts again but nothing happens. The cooler seems fine and is spinning now.
I think I'll have suck it up and to go to Frys in the morning to pick up a new CPU. I only hope nothing else was damaged as well.
posted by lpctstr; at 12:51 AM on June 19, 2008
I think I'll have suck it up and to go to Frys in the morning to pick up a new CPU. I only hope nothing else was damaged as well.
posted by lpctstr; at 12:51 AM on June 19, 2008
How long was it running without the fan? I've run an Athlon XP with the fan unplugged (accidentally, of course!) for 30 seconds and other than it eventually locking up, it wasn't an ongoing issue. Intel CPUs are supposed to be (and in my experience, are) much better in that respect.
You probably were messy with the paste and now some of the resistors on the CPU are shorted. If you clean it up, it might fix the problem. Either that or you applied way too much paste and it's not conducting heat properly.
posted by wierdo at 1:54 AM on June 19, 2008
You probably were messy with the paste and now some of the resistors on the CPU are shorted. If you clean it up, it might fix the problem. Either that or you applied way too much paste and it's not conducting heat properly.
posted by wierdo at 1:54 AM on June 19, 2008
Oh, and I used to not believe in power supply failure as a cause for random issues like this, but I'm officially a convert. It seems to be happening to me regularly lately. It could be that the power supply is the root of all your issues, including the current not booting.
I had a server that would run anywhere from 30 seconds to 4 hours between crashes, regardless of load. It was the power supply. In the past I always figured if it could get through all the thrashing and CPU usage of loading the OS, the power supply probably wasn't the issue.
posted by wierdo at 1:58 AM on June 19, 2008
I had a server that would run anywhere from 30 seconds to 4 hours between crashes, regardless of load. It was the power supply. In the past I always figured if it could get through all the thrashing and CPU usage of loading the OS, the power supply probably wasn't the issue.
posted by wierdo at 1:58 AM on June 19, 2008
Response by poster: Well, things keep getting worse.
I went out and bought a new CPU and the same problem occurs. So for now I'm ruling out that the overheating damaged the CPU.
On the other hand, my next culprit is the motherboard, which would be really tedious to change.
Does anyone know what the "components power up, components all die 5 seconds later, components power up again" pattern means? Maybe it's the motherboard trying to reset the CPU to default settings when the first boot fails but that's just a guess.
posted by lpctstr; at 11:30 AM on June 19, 2008
I went out and bought a new CPU and the same problem occurs. So for now I'm ruling out that the overheating damaged the CPU.
On the other hand, my next culprit is the motherboard, which would be really tedious to change.
Does anyone know what the "components power up, components all die 5 seconds later, components power up again" pattern means? Maybe it's the motherboard trying to reset the CPU to default settings when the first boot fails but that's just a guess.
posted by lpctstr; at 11:30 AM on June 19, 2008
Look at the components in and near the CPU socket. Do they appear damaged? Are any of the capacitors bulging? Does anything look scorched? If not, it's probably not the motherboard, although it could be. I'm betting power supply, but that may just be because that was the cause of my most recent perplexing problem.
posted by wierdo at 2:03 PM on June 19, 2008
posted by wierdo at 2:03 PM on June 19, 2008
Are you 100% positive that you've got the chip seated fully and absolutely, and that the sink is on the right way, and that your ram is fully seated?
posted by TomMelee at 2:07 PM on June 19, 2008
posted by TomMelee at 2:07 PM on June 19, 2008
Response by poster: Good news, I am sane again.
I bought a replacement motherboard and put everything on that, and it boots now! 2 POST beeps have never sounded so beautiful.
CPU-intensive apps work again as well. I also learned that the everest temperature sensor is bogus.
So I guess when the CPU overheated, it fried the motherboard. Now I have an extra CPU to sell on craigslist and a broken motherboard (though under warranty).
I marked everyone who was "correct" as best answer
posted by lpctstr; at 2:51 PM on June 19, 2008
I bought a replacement motherboard and put everything on that, and it boots now! 2 POST beeps have never sounded so beautiful.
CPU-intensive apps work again as well. I also learned that the everest temperature sensor is bogus.
So I guess when the CPU overheated, it fried the motherboard. Now I have an extra CPU to sell on craigslist and a broken motherboard (though under warranty).
I marked everyone who was "correct" as best answer
posted by lpctstr; at 2:51 PM on June 19, 2008
This thread is closed to new comments.
posted by Mike1024 at 3:39 PM on June 18, 2008