Sign in with
Sign up | Sign in
Your question

Computer crashes and sometimes won't boot

Last response: in Components
Share
February 18, 2011 8:15:57 AM

Hello everyone.

At first, I'd like to address the moderators. I wasn't really sure where to post this, because I couldn't find a "Problems" section anywhere, so feel free to move this is necessary.


I bought brand new components about four months ago and assembled them myself. For the first month everything was working well, but all of a sudden the computer crashed. It wasn't a Windows-based crash, no. The screen froze, the sound went silent and I couldn't get the system to respond at all. I had to push the power button.

After shutting down, I rebooted the computer, but the screen stayed black and I couldn't hear the familiar beeps the computer makes when it starts. Nor could I hear Windows' welcoming sound. I tried again and again, but nothing more happened. Then I waited for 15 minutes or so and tried again and it worked. The crashes aren't directly related to the computer not booting, because sometimes after a crash the computer comes back alive without an effort, but booting problems never happen without a crash first.

For the next two weeks I didn't experience a single crash, but then it started to go downhill really badly. At first, I started experiencing crashes once a week, then couple of times a week, once a day and now I'm getting them even four times a day! This is really a pain in the ass, and not worth my well-earned money.

The reason why these crashes happen is a mystery. They don't seem to have a specific pattern. Below I've listed some examples.
- I plugged in a microphone
- changed a tab in Firefox to a HD Youtube video
- joined to a Counter-Strike game and pressed TAB to see the players
- turned on my music player
- rewound a DVD movie in VLC
- wrote a blog entry on the browser

I'm pretty sure that the problem is not caused by overheating. I've measured the temperatures after a crash from BIOS and with a program called HWMonitor in Windows, and they were fine. And once I actually managed to keep HWMonitor on while a crash happened and the temperatures were fine even then.

I can't give you much more clues, but here's something I've found interesting. Everytime the computer refuses to reboot, the DVD drive LED flashes either 'till infinity or not at all. If it flashes 4-6 times, everything is fine. Other peculiar observation includes the CPU fan. I installed a Scythe Mugen 2 fan to replace the default one, and it showed a CPU FAN ERROR at boot, so I turned the error message off from BIOS. Also, SpeedFan tells that CPU fan rotates at a stunning 400k-500k RPM.

My specs below:
Quote:
AMD Phenom II X6 1055T (2.8 GHz)
Asus M4A89GTD-PRO/USB3
2 x SanMax OCZ3G1333LVAM2G 2GB DIMM PC5300U
ATI Radeon HD 5850 (1440SP 725 MHz, 1 GB DDR5 4x1GHz)
Samsung HD103SJ (1TB, SATA300, 3.5", 7200rpm, NCQ, 32MB Cache)
Scythe Mugen 2 CPU fan
Antec 550W ATX12V V2.3


This is really important to me and I want to thank each and everyone of you, who even read this thread. I'm losing my sanity over this and I really don't want to spend anymore money on expensive tech support.

More about : computer crashes boot

February 18, 2011 8:36:04 AM

Welcome to THG.

Many of the symptoms you listed can be cause by memory errors, or video errors. Sometimes PSU, but the first culprit is always memory. The first thing I would recommend doing, is go into your BIOS and LOAD OPTIMAL DEFAULTS, then when it reboots, get back into the BIOS again and make sure Overclocking and Memory items are all set to AUTO.

As for the FAN error, perhaps the fan you bought isn't compatible will the Cool n Quiet specification, so I will recommend you disable Cool n Quiet while you are in the BIOS.

Which version of Windows are you running? I didn't see above and sorry if I missed it. If you are running Vista or Windows 7, go to the control panel and click the option to show you everything in small icons. Click on Administrative Tools then double click on Windows Memory Diagnostic. Click RESTART NOW and CHECK FOR PROBLEM. When it restarts let it do the diagnostic and do not interrupt it. Let it finish. If you don't have anything to do, you can watch it and it will tell you if there has been a memory error. If you are busy, you can walk away, let it finish and reboot. After Windows comes back it will tell you if it detected any memory related issues.

If it did, try doing the same procedure above, one memory stick at a time installed. then you can narrow down a memory problem.

If you don't have any errors, next thing to do is run a stability test. But before you proceed, I want to rule out your video card. Make sure you have the latest ATi Drivers installed. If you already have this installed, we'll move on.

Many people recommend many different temperature monitors, but I usually used Speedfan. I will start Speedfan and let it sit for a moment and check CPU temps. If everything is less than 45 C, then I will start the program Prime95. I will use the INTEL stress test and start it for all cores. You should monitor your CPU temps in the first 10 minutes of the stress test and make sure they are under 65-70C. If it is 60-65 you can let the stress test continue for a few hours or until you get an error. If CPU temp is above 70, better start checking your heatsink fan again.

If one of the worker threads stops in the Prime95 test, you could still have a memory issue that the Windows diagnostic missed. But if all temps are cool and you get no errors during a stress test and the heat is well below 65C, then start looking at your power supply.

If the power supply voltage is not stable on the CPU line, you could need a bigger one. Depending on the stepping of your CPU, the wattage could be between 125-140w.

If you are not familiar with Speedfan or Prime95, I suggest you google it and read up a bit. So you will know what the errors could be and how to keep everything in safe limits. Also reseating your RAM and video card can't hurt too.
m
0
l
Related resources
February 18, 2011 9:27:25 AM

^ Here here. That is why there will probably be memory errors if that's the case. I got a P5QL Pro that won't take most single-sided chips if they are DDR2-1066, but seems to take almost any cheapo DDR2-800 single-sided or double-sided.

Go to the Asus website and look at the memory QVL list for your board and check that your current sticks are on it, cause sometimes it's just not stable if they didn't verify it.

So test the memory thoroughly. Also, what is the exact model number of your power supply, cause if you have less than 140w continuous power available to the CPU, then the PSU could be your problem.
m
0
l
February 18, 2011 10:01:22 AM

Thanks for the fast replies!

I'll do a quick update now and a new one later on, because I still have to go through beanoslim's list for most common errors.

I loaded optimal defaults in BIOS and Memory and OC related topics are on AUTO now. I also turned off Cool n Quiet and what do you know, the fan speeds in HWMonitor turned back to normal (but are still a bit odd in SpeedFan).

Prime95 has been running for an hour now and temperatures are looking great. By clicking the links below you can see what I'm seeing right now.

http://oi51.tinypic.com/103wv42.jpg SpeedFan
http://oi53.tinypic.com/bfpcw5.jpg HWMonitor

I have the newest ATI Drivers.

I'm not sure what to look for in the power supply manual, which btw is really scarce, but it says on the front that "Antec TP-550 power supply supports _dual GPUs and dual/quad core CPUs_". I have a hexacore, could this be a problem?
m
0
l
February 18, 2011 10:11:00 AM

Just read the label on the PSU.

Looks something like this.

Output
+3.3@24A,+5V@24A,+12V1@18A,+12V2@18A,+12V3@18A,-12V@0.8A,+5VSB@3.0A

Tell us what the first one says, +3.3 how many Amps?
m
0
l
February 18, 2011 10:14:49 AM

Oh, ok.

+5V @ 25A, +3.3V @ 25A, +12V1 @ 20A, +12V2 @ 20A, +12V3 @ 20A, +12V4 @ 20A, -12V 0.8A, +5VSB 3A

---
OCCT starts now!
m
0
l
February 18, 2011 10:15:36 AM

Dual core and Quad core chips can be 95w-140w. Most Intel chips like Core 2 Quad are 95-115 while AMDs are usually 125-140w. So if that is a little weak, then it is possible your CPU is drawing too much from it. That could definitely cause stability issues. As for Dual GPUs... most modern GPUs that are mid to high end cannot be run on such a small PSU. My GTX465 recommends at least a 550w PSU for a single card. If I had dual GTX465 cards they recommend 700-850w in the Zotac manual.

m
0
l
February 18, 2011 10:18:50 AM

Well 25A should be enough for the 1055t.

You should pick a good tool for monitoring temps and voltages, see if your voltages drop or fluctuate a lot when running Prime95 and when it is idle. If the voltages are solid, then your PSU is fine. It still usually comes down to memory or hardware conflict that causes stability issues if your temps and voltages are spot on.
m
0
l
February 18, 2011 10:27:39 AM

^OCCT will monitor voltages if you set it to in the configue tab.
m
0
l
February 18, 2011 11:39:00 AM

The test is over. And to sum it up I've made a collage of voltage graphs that you can view from the link below. I can't claim to understand the graphs perfectly (or rather, I don't know what to look for), but to my eye it seems that there's nothing awful going on. I'll let you to decide.

http://oi53.tinypic.com/2wrojk8.jpg

CPU temperatures stayed below 45C, which should not be alarming.

I was actually feeling really positive about resetting the BIOS settings to default and turning off Cool n Quiet, but it turned out I was celebrating a bit too early. As I was writing this post for the first time, the computer crashed!

Now I'll go through the List.

EDIT:
Oh no, now it's going completely haywire... I was reading the list and another crash occurred.
m
0
l
February 18, 2011 12:15:06 PM

You ran a 1hr OCCT test and stayed under 45C, amazing.

+1 for memtest next, 10 passes, take out you mem sticks and clean the contact with a cotton bud dipped in alcohol and blow out the mem slots to make sure no hairs are in there.
m
0
l
February 18, 2011 12:52:33 PM

Should I run memtest with just one stick at a time or is it ok if I do it with both in at the same time?

EDIT: Oh, I must have missed Davcon's post.
m
0
l
February 18, 2011 1:13:44 PM

be11 said:
Should I run memtest with just one stick at a time or is it ok if I do it with both in at the same time?

EDIT: Oh, I must have missed Davcon's post.

Yeah bud Memtest 86+ is much better then the built in Windows memory test.
I had issues with bad ram and the Windows Memory Diagnostic found nothing.
Memtest86+ found over 2000 errors on the bad module.
m
0
l
February 18, 2011 1:22:55 PM

Quote:
That psu is fine. Did you make sure your voltages of your ram is on 1.8?

I'm not sure if I'm doing this correctly, but I downloaded CPU-Z by CPUID and it says that both memory slots are running @ 1.50 V. Should I change it?

I've just stumbled upon a new symptom. Both the mouse and the keyboard go unresponsive every now and then for a half a second or so even though everything else keeps running. They are both plugged into a usb slot.
m
0
l
February 18, 2011 1:34:06 PM

Just manually set the voltage in the bios, if it was me I'd set the default timings also 5-5-5-15 or whatever.

Seems strange to me that it passed OCCT as that stresses quite a bit of ram.

I'd go for both sticks 1st, 10 passes, if you get any errors try each stick on its own.

Don't forget to give them a good clean before testing.
m
0
l
February 18, 2011 1:36:54 PM

beanoslim said:
Just manually set the voltage in the bios, if it was me I'd set the default timings also 5-5-5-15 or whatever.

Seems strange to me that it passed OCCT as that stresses quite a bit of ram.

I'd go for both sticks 1st, 10 passes, if you get any errors try each stick on its own.

Don't forget to give them a good clean before testing.

Ok. I'll have to get back to you on this in a couple of hours (or in the worst case, tomorrow). I'm going to have to do some work now and since I just have this one computer, I can't be running a Memtest while at it.
m
0
l
February 18, 2011 2:03:52 PM

beanoslim said:
Oh dear what am I on, is this it?
http://www.ocztechnology.com/ocz-ddr3-pc3-10666-gold-ed...

You need to get the info from your ram and find out what the default timings and voltage should be.
Set them in the bios then run your memtest test.

Oh yes! Those are the right ones.

So, I'll set the timings and voltage from BIOS and then run Memtest? Understood!
m
0
l
February 18, 2011 2:14:13 PM

And clean the mem contacts and slots. :D 

Good luck.

You only need to set the 9-9-9-20 timings, no others and voltage to 1.7v.
m
0
l
February 18, 2011 5:20:50 PM

Wow. Just wow.

I changed to 1.7V 9-9-9-20 and holy cow, now everything's messed up. The crashing probably stopped, or at least it didn't happen during the three hour period I was working. But when I started watching a movie every program lost control. Crashes here and there, memory dumping, odd texture bugs. Fried memory 101.

Booted and got memtest running -> over 250k errors during the first pass.

Does this mean I have to buy new sticks? And if so, could you recommend me some GOOD ones since obviously my last choice sucked badly. I can't believe it. Just four months ago they were in mint condition. What could have made them burn out so fast?

And yeah, this baffles me too. The first time I ran memtest, before changing the settings, it didn't find any errors during five passes. Why not?
m
0
l
February 18, 2011 5:35:48 PM

Did you clean the contacts and slots?

Are they still under warranty then, send them back.

1st you want to try a few things like

Try each stick in each slot and see if you can get a memtest pass.

Try both sticks in the other mem slots to make sure its not a slot problem.

I would recommend some Corsair XMS3/Dominator or some Crucial Ballistix.

Wow indeed, thats some serious errors, you'll probably find its one stick causing the problems.

If they were mine I'd also bump the voltage up to 1.8v and try them just out of interest.
m
0
l
February 18, 2011 9:07:51 PM

Uh oh, I've made a mistake. Not sure if it was a big mistake or not, but I just disassembled everything and put it back together with the default CPU fan. I also noticed that my memory sticks weren't the ones you found, but these:
http://www.ocztechnology.com/ocz-ddr3-pc3-10666-gold-am...
OCZ DDR3 PC3-10666 Gold AMD Edition _Low Voltage_

This means they only require 1.65V to work, but I guess 0.05 volts won't really matter.
m
0
l
February 18, 2011 9:53:31 PM

I badly fear that my efforts were in vain. Memtest at 9-9-9-20 @ 1.65V produced 2500 errors, which is much better than 250k, but still a significant amount. I didn't swap the slots, just cleaned the sticks and blew into the slots. I'll try swapping them tomorrow.
m
0
l
February 19, 2011 12:26:58 AM

be11 said:
I badly fear that my efforts were in vain. Memtest at 9-9-9-20 @ 1.65V produced 2500 errors, which is much better than 250k, but still a significant amount. I didn't swap the slots, just cleaned the sticks and blew into the slots. I'll try swapping them tomorrow.

Time for a RMA the module is junk!
Contact OZC.
Run test stick at a time.
usual only one stick is bad in the kit but who knows you have OZC ram.
If only one stick is bad mark it with masking tape.
m
0
l
!