Cold computer booting problem

[voj]

Hello,

Sometimes I have such problem:

When I turn on my computer after a longer period of inactivity,
for example in the morning - after a night, it start (the monitor is on,
fans are spinning) but it doesn't boot, it doesn't show any bios info,
the keyboard doesn't blink.
Unfortunately my case doesn't have a beeper, so I don't know if it beeps,
I'll try to get one.
When I do the reset or turn it off and on again - it starts normally.
After a few hours of inactivity the problem doesn't appear, it must
be about at least 7 hours maybe - and it's not always there even in
that case.
After such situation, when I enter Bios it says "I failed to boot maybe
because of the overclock or voltage change in bios" - but I did
"load defaults" - and it didn't help.
The computer is about 2 months old, and the problem started
about one week ago.

I haven't done any hardware changes.

I haven't had any problems with my computer,
except with ati hd 5770, which I've read are driver related ones.

I can run memtest86+ for hours, I can run Furmark and Prime95
at the same time for hours - the system is rock stable.

I don't know what (and how) to check first, psu ? motherboard?
What would you recommend?

My config:

Motherboard: gigabyte ga-p55a-ud3 (bios F5)
CPU: intel core i5 (prime95 tested)
RAM: 2x2GB OCZ 1066Mhz (Memtest86+ tested)
Video Card: Sapphire Radeon HD 5770
PSU: OCZ modXtream 500W PRO
Monitor (connection used): 17" LCD SONY DVI
Hard drives: Samsung SpinPoint 750GB
Optical Drives: DVD-RW
Additional PCI cards: pentagram WiFi
Number of fans present: 3 case fans

 

[buzznut]

Well, I would suspect due to the error message you're receiving that there is something amiss in the BIOS. The ram you mentioned seems to be running at an odd speed, or perhaps that is a misprint. If using ddr3 ram, it could be 1333 Mhz, or 1066. I haven't heard of 1033Mhz.
Perhaps you need to change the ram speed in bios.

I would get out the motherboard manual and check the bios settings before doing anything else. Sometimes the factory defaults (reset BIOS) are not ideal for your hardware.

Its a strange problem, but not completely unfamiliar. My current machine was doing this thing where it would turn on but not boot, then a reset would correct this and it would boot normally. Basically it would boot normally every other time. It didn't bother me much, but it was a weird thing, ya know? Anyway, an update to the bios made that problem go away. I didn't know it was the bios at the time, but I had to update the bios for a new Phenom II cpu I dropped in.

Updating a bios is a serious undertaking, which is why I didn't recommend it right away. If you do go that route, be absolutely certain you know what you are doing- you could brick your machine.

 

[voj]

I'm sorry, my mistake, it's 1066 - I've corrected it.
I did a bios update, because it fixed the famous "ati big cursor issue", but it was some time ago and everything was OK after this change.

 

[roonj]

Want another take, the bios update seems logical as a failure to boot can be due to a component not initiating. So which components are capable of this, almost all, the fact that reset works sugests that some electical component is discharging when the system is off and is incapable of of recharging quickly enough from a cold boot (several hours of discharge time). This could be caused by a faulty capacitor or a leaking diode or even just a weak electical connection (two plugs not making proper contact. Good news is it's isolated meaning if the scenario I offer is right only the affected component is likely to fail.
If it's not the bios having device intepretation problems which an update can fix or just unplugging and replugging everthing in the component will eventually fail and or bios will provide a more specific identification.
I have no idea why I presented this sort of doomsday analogy, maybe the chili I had last night.

 

[voj]

Thanks,

to make the situation even more unclear - I only happens sometimes. For example today - no problem despite the fact it was pretty cold in my room.

 

[jitpublisher]

This has happened to me a few times over the years, and it is usually, not always, but usually either one of two things. Try bumping the voltage on the memory up a little, or it could be a PSU that simply is not functioning correctly.

 

[voj]

This has happened to me a few times over the years, and it is usually, not always, but usually either one of two things. Try bumping the voltage on the memory up a little, or it could be a PSU that simply is not functioning correctly.

I tried with mem voltage with no success. I have also observed something else:
1. I enter the deafault values in Bios manually, like PCI-e frequency to 100 (instead of auto) or memory multiplier to 8x (instead of auto)
2. The computer has a booting problem (it happens every lets say 10th starting up)
3. after next reboot bios says something like that: "the system has experienced boot failures because of overclocking or changes of voltages." But it doesn't make sense, because it's not overclocking or voltage change - these are the defaults which are set by 'auto' Bios later on during the booting process!
Maybe it's normal - I'm no expert, but maybe it could be clue?

 

[voj]

Can a bad ram stick cause this?
After removing one stick the problem wasn't appearing - and with the other one - it was still there.
I did few tests, so maybe it could be a coincidence, I'll do more, but are such things possible?
What's more, Memtest shows no errors (after 130 passes) BUT Prime95 in Blend test shows an error after a few hours!
Does anyone come across such situations?

 

[obsidience]

"Can a bad ram stick cause this?"

You are smart to look into this direction.

I've had a similar issue I've been battling the last 6 months and have spent a small fortune trying to resolve it. First system was a Phenom II with the older socket and a Gigabyte board with 8GB (2GBx4) of DDR2 memory. My current system is a Phenom II with newer socket and an Asus mobo and 8GB (2GBx4) of DDR3 issue. I've replaced the memory, board, CPU, power supply. I've added a brand spanking new 1500VA UPS and a power conditioner trying to rectify this problem to no avail.

What I currently know is that there are two types of cold boot issues. One in which the system will not post and another in which the system posts but may not load into Windows (bsod). After a number of reboots the system will post and will load into Windows. In some cases it takes a power down of the PSU and other's just reboots and time. No changes to voltage appear (CPU/HT/NB/Mem) to rectify the problem for me.

For my current issue the problem appears to be higher speed memory occupying all 4 memory banks. Apparently a number of memory, motherboard manufacturers don't support this. Even AMD has a document stating the problem:

http://gskill.us/forum/showthread.php?t=2977

I am hearing that this problem can occur with bad memory in two dimm slots however a majority of the issues are due to people using all 4 dimm slots. I've been able to verify the issue by burning a copy of the ultimate boot cd and memtesting after a cold boot (overnight) and watching it fail on test 5.

I've also heard that some people have been able to isolate the bad memory (usually 1 or 2 sticks) and get it replaced and it replaced and resolving the issue.

 

[voj]

Thanks,

computers can be frustrating

The thing I don't understad is why memtest shows nothing... I tried both versions, I tried goldmemory, some windows tools - nothing except the mentioned above Prime95. "So, it must be my Core I5" I thought - BUT there's no problem when Prime does his calculation in the first test - the one "without" RAM, the one which stresses CPU the most.

 

[voj]

Solved!
I had a faulty OCZ ram stick

 

[ekoostik]

How did you determine the OCZ ram stick was faulty? Did you get errors in memtest86+ or some other application? Or did you replace the RAM and it worked? If your replaced the RAM, what model RAM did you use that works?

 

[aced]

voj, thanks for letting us know that replacing the DRAM fixed the problem. But please let us know how you determined the problem, and what you replaced the memory with. This is a frustrating problem for many folks right now (myself included) so any help would be great.

 

[voj]

It was a strange issue:

1. When I used both sticks or the one with "1" at the end of s/n alone:
- my computer sometimes refused to start in the morning after night off, not always - but it was happening. Bios was beeping once - and was stopping, os wasn't starting, I couldn't enter bios. I had to reset, sometimes twice to get my computer starting OS.
- another issue - it was crashing after a few hours of Prime95 Blend test, stopping with "hardware failure" error.
2. When I used only one stick, the one with "2" at the end of s/n alone - there were NO PROBLEMS!
3. When I used completely different memory, like for example cheap Goodram (made in Poland ) - No problems.
4. What's most interesting - I could run memtest86+ for days (!!!) without any problem in every memory configuration described above!

I did RMA, my Reseller tested the memory and decided to send me a new pair. So, when I get them I'll confirm that new OCZs work fine. I

Hope this helps someone

 

[aced]

1. When I used both sticks or the one with "1" at the end of s/n alone:
- my computer sometimes refused to start in the morning after night off, not always - but it was happening. Bios was beeping once - and was stopping, os wasn't starting, I couldn't enter bios. I had to reset, sometimes twice to get my computer starting OS.
- another issue - it was crashing after a few hours of Prime95 Blend test, stopping with "hardware failure" error.
2. When I used only one stick, the one with "2" at the end of s/n alone - there were NO PROBLEMS!

I never did try just one stick at a time to see if I could recreate and avoid the problem, thus faulting a single stick. I figured it would only show up with both sticks involved but I never did confirm this. I should try this.

 

[aced]

voj, any progress? Replacing the memory worked for me.

 

[voj]

voj, any progress? Replacing the memory worked for me.

It's good to read it
I have a new pair of OCZ sticks (the same type) and everything seems to be OK now

 

[r_manic]

Best answer selected by r_manic.