My PC freezes randomly even in BIOS

K

keylog keylog

My computer freezes randomly when in Windows (windows 7 pro 64 bits) and even in BIOS. It can freeze 4 times in an hour or not freeze for many hours when it's used or not, completely randomly.

It has the following hardware:
CASE: Xigmatek Midgard Case (CPC-T55DB-U01)
PSU: CoolerMaster M520 Real Power Modular
CPU: Intel Core I7-950 (s1366, 3.06GHz, 8MB, Box)
MOTHERBOARD: Gigabyte GA-X58A-UD3R (s1366, DDR3, X58), REV 2.0
RAM: OCZ 6GB DDR3-1333 Ram Kit OCZ3G1333LV6GK - (240 Pin DIMM)
VGA: Palit GTS 450 Sonic (PCI-Express, GDDR5, 1GB) - (128 bit)
HDD: Western Digital Caviar Black WD1002FAEX 1TB HDD (64MB Cache, 7200RPM, Sata III)
DVD-RW: LG LS GH22LS50 DVDRW + / - (Sata, Bulk)

I have used BurnInTest, memtest86 and unigine programs to test the hardware and no errors are found.

I booted with the memtest86 cd-rom and I tested it for 9 hours without errors and my system didn't freeze during the tests. But in BIOS it freezes. Does this fact means that in Bios the system loads something else that makes my computer to freeze ???

Any suggestion appreciated.
Thanks in advance.
 
R

Rob

My computer freezes randomly when in Windows (windows 7 pro 64 bits) and even in BIOS. It can freeze 4 times in an hour or not freeze for many hours when it's used or not, completely randomly.

It has the following hardware:
CASE: Xigmatek Midgard Case (CPC-T55DB-U01)
PSU: CoolerMaster M520 Real Power Modular
CPU: Intel Core I7-950 (s1366, 3.06GHz, 8MB, Box)
MOTHERBOARD: Gigabyte GA-X58A-UD3R (s1366, DDR3, X58), REV 2.0
RAM: OCZ 6GB DDR3-1333 Ram Kit OCZ3G1333LV6GK - (240 Pin DIMM)
VGA: Palit GTS 450 Sonic (PCI-Express, GDDR5, 1GB) - (128 bit)
HDD: Western Digital Caviar Black WD1002FAEX 1TB HDD (64MB Cache, 7200RPM, Sata III)
DVD-RW: LG LS GH22LS50 DVDRW + / - (Sata, Bulk)

I have used BurnInTest, memtest86 and unigine programs to test the hardware and no errors are found.

I booted with the memtest86 cd-rom and I tested it for 9 hours without errors and my system didn't freeze during the tests. But in BIOS it freezes. Does this fact means that in Bios the system loads something else that makes my computer to freeze ???

Any suggestion appreciated.
Thanks in advance.

This sounds like an intermittent hardware problem and I suspect
the memory. The BIOS uses some areas of RAM which Memtest can't
address.
Try swapping the DIMM modules between slots. This should then
put any bad module into a memory area which Memtest can get at.
There are other possibilities, but this is easy to try as a
first step.
HTH
 
D

David W. Hodgins

My computer freezes randomly when in Windows (windows 7 pro 64 bits) and even in BIOS. It can freeze 4 times in an hour or not freeze for many hours when it's used or not, completely randomly.
PSU: CoolerMaster M520 Real Power Modular
VGA: Palit GTS 450 Sonic (PCI-Express, GDDR5, 1GB) - (128 bit)
I booted with the memtest86 cd-rom and I tested it for 9 hours without errors and my system didn't freeze during the tests. But in BIOS it freezes. Does this fact means that in Bios the system loads something else that makes my computer to freeze ???

My guess is lack of power.

Check the label on the psu. One article I found suggests it's only rated
for 19 amps on the +12V rail, which I doubt is enough for a GTS 450.

Regards, Dave Hodgins
 
P

Paul

keylog said:
My computer freezes randomly when in Windows (windows 7 pro 64 bits) and even in BIOS. It can freeze 4 times in an hour or not freeze for many hours when it's used or not, completely randomly.

It has the following hardware:
CASE: Xigmatek Midgard Case (CPC-T55DB-U01)
PSU: CoolerMaster M520 Real Power Modular
CPU: Intel Core I7-950 (s1366, 3.06GHz, 8MB, Box)
MOTHERBOARD: Gigabyte GA-X58A-UD3R (s1366, DDR3, X58), REV 2.0
RAM: OCZ 6GB DDR3-1333 Ram Kit OCZ3G1333LV6GK - (240 Pin DIMM)
VGA: Palit GTS 450 Sonic (PCI-Express, GDDR5, 1GB) - (128 bit)
HDD: Western Digital Caviar Black WD1002FAEX 1TB HDD (64MB Cache, 7200RPM, Sata III)
DVD-RW: LG LS GH22LS50 DVDRW + / - (Sata, Bulk)

I have used BurnInTest, memtest86 and unigine programs to test the hardware and no errors are found.

I booted with the memtest86 cd-rom and I tested it for 9 hours without errors and my system didn't freeze during the tests. But in BIOS it freezes. Does this fact means that in Bios the system loads something else that makes my computer to freeze ???

Any suggestion appreciated.
Thanks in advance.

I would start by using a search engine, and look for

GA-X58A-UD3R freezing

and see what pops up. Sometimes, it's a BIOS issue, in
which case, a different BIOS version might give relief.
(The BIOS can even influence the OS operation, as
SMM continues to run in the background and when SMM
interrupts the system, the OS can't even tell it is
happening.)

(SMM is sometimes used for adjusting the number of operating
phases of the VCore regulator...)

http://en.wikipedia.org/wiki/System_Management_Mode

Looking at the customer feedback here, I can see a few issues.

http://www.newegg.com/Product/Product.aspx?Item=N82E16813128423

You could try testing with one stick of RAM at a time in the
system, and see if the freezing stops or not. Three sticks
means three tests to run.

Freezing issues are not easy to fix. For example, some people
do load testing, and the system never freezes when load tested.
But if you leave the thing idle for five minutes, it freezes.
That can be... very frustrating. And while it's tempting to
assume SpeedStep or Cool N' Quiet VCore adjustment is doing it,
that isn't always the case. Usually, there will be some
ACPI adjustments in the BIOS you can make, to stop SpeedStep
changes (like disabling all the C states, disable EIST and so on -
on my current motherboard, you have to change a *lot* of settings,
to stop SpeedStep completely). But even if you try that, there
doesn't seem to be a strong correlation between freezing and
VCore changes. The engineers at Intel and AMD have done a
good job of margining their designs, and picking appropriate
load lines for their processors.

And the fact it is freezing in the BIOS, is a great observation,
as it gets the OS off the hook :) It means there is a real
hardware issue.

It could also be a defective power supply, but then the question
you'd ask, is why there isn't a stronger correlation to power draw.
For example, if it froze every time you entered a 3D game, I'd
be more tempted to blame the power supply. Apparently the GTS 450
only draws 11W at idle, and 84W in Crysis, so in many (non-gaming)
situations, it isn't a major contributor.

Swapping in a different supply, is the easiest way to check for
that kind of thing (defective power supply). Mainly because,
it's too hard to test the power supply with easily available
tools. Swapping it out, is the next best thing. A power supply
can have faults in its dynamic behavior, that you can't check
with a multimeter.

Paul
 
J

John McGaw

My guess is lack of power.

Check the label on the psu. One article I found suggests it's only rated
for 19 amps on the +12V rail, which I doubt is enough for a GTS 450.

Regards, Dave Hodgins

Given the OP's symptoms that seems less than likely. True, the video
subsystem can draw a lot of power but it doesn't do so all the time. In the
BIOS it will be drawing virtually nothing. If the problem came up during 3D
game play then I might go along with the PS theory but that is far from
random as was described.

That said, I don't really have a solid theory which might solve the
problem. The closest I've come is a Q6600 computer which had memory which
was just slightly bad/out of spec/mistimed/incompatible/screwed/whatever
and which would cause BSODs when it felt like it but which could pass
burnin testing and memory testing for extended periods without a glitch.
 
K

keylog keylog

I updated the bios to X58A-UD3R FF from X58A-UD3R FA two days before.

Today my computer froze again inside Windows and then for the first time itcould not boot at all, giving a short beeps sound. According to the motherboard's manual this was "Continuous short beeps: Power error".
I just started to think that I found the problem and it must the PSU unit that is fault.

But I decided to open the case and check the memory modules. The computer was able to boot again only when I put off all the three memory modules and put them back again. (I put off one memory module at a time and tried to boot the computer each time). After I put the memory modules back again the computer boots ok.

Now I am confused again. Do you have any other speculation??
Thanks again.
 
K

keylog keylog

I updated the bios to X58A-UD3R FF from X58A-UD3R FA two days before.

Today my computer froze again inside Windows and then for the first time itcould not boot at all, giving a short beeps sound. According to the motherboard's manual this was "Continuous short beeps: Power error".
I just started to think that I found the problem and it must the PSU unit that is fault.

But I decided to open the case and check the memory modules. The computer was able to boot again only when I put off all the three memory modules and put them back again. (I put off one memory module at a time and tried to boot the computer each time). After I put the memory modules back again the computer boots ok.

Now I am confused again. Do you have any other speculation??
Thanks again.
 
D

David W. Hodgins

Now I am confused again. Do you have any other speculation??

It's an old system with dirty connectors. Remove everything, clean
all connectors with Isopropyl alcohol, and then try it again.

If it still freezes up, then look at replacing the psu with one that
has enough amps on the +12V rail.

Btw, the reason I suspect the lack of amps on the +12v rail, is that
I've seen it before. The system would randomly lock up, or reboot.
What finally got me to suspect it was a power problem, was that with
three usb devices connected, it would not boot. Any two of the three
was ok, but all three, no way. Replacing the 500w psu with 24 amps on
the +12v rail, with a 750w unit with 50w on the +12v rail fixed the
problem.

Regards, Dave Hodgins
 
P

Paul

keylog said:
I updated the bios to X58A-UD3R FF from X58A-UD3R FA two days before.

Today my computer froze again inside Windows and then for the first time it could
not boot at all, giving a short beeps sound. According to the motherboard's manual
this was "Continuous short beeps: Power error".

I just started to think that I found the problem and it must the PSU unit that is fault.

But I decided to open the case and check the memory modules. The computer was able
to boot again only when I put off all the three memory modules and put them back
again. (I put off one memory module at a time and tried to boot the computer each
time). After I put the memory modules back again the computer boots ok.

Now I am confused again. Do you have any other speculation??
Thanks again.

I don't really know what a "Power Error" is. The thing is, the power supply has
its own "Power Good" kind of signal. And the motherboard regulators also can
have status signals, indicating their respective outputs are close to full
voltage. The motherboard won't come out of reset, unless those Power Good
signals are all asserted. So to a certain extent, if the power levels were
bad, the motherboard would just give a black screen (stuck in reset).

The motherboard has the option of reading system voltages with the Hardware
Monitor. To check that, enter the BIOS, and go to the Hardware Monitor page
and check the voltage readouts there. The BIOS may highlight values it is
unhappy with (values which it considers out of range). The ATX spec may allow
+/-5% deviation from the ideal value. The hardware monitor used crude measurement
techniques, and a multimeter may give better results, when you actually need
to verify voltages.

Things I'd try:

1) Re-seat processor. Remove the heatsink, clean off the paste, pull the
processor out of the socket. Look for damage, burnt pins or the like. Be
careful, not to get paste or residue into the socket pin area. Inspect
the gold pads on the CPU. If the socket is working properly, there should
be tiny scratch marks on each pad (proving there was contact with the "points"
in the socket. Reapply thermal paste, reinstall the heatsink/fan. If you don't
have thermal paste, you'll have to pick up a tube of it.

The purpose of re-seating the processor, is to try to solve a contact problem.

*Don't* use abrasives on the processor. Similarly, there is *no* cleaning
procedure for the LGA1366 socket. If the socket is fouled with debris, it's
a writeoff. The processor could be cleaned with isopropyl alcohol, but if
you use abrasives, the super-thin gold plating would be ruined. Gold is too
expensive now, for companies to be applying 50 micron coatings.

2) Strip system to minimum hardware, and test a bit at a time. The beep
codes received, should correspond to the error type. For example, if
you get "short beeps" when all three RAM sticks are removed, and also
get "short beeps" when some RAM is present, perhaps the short beep code
is actually a memory error ? With no video card, and some good RAM present,
you should receive a different code for the missing video card. Be careful
to always shut off the power supply at the back, before swapping any hardware.

You could also consider testing with a PCI Port 80 Test Card, but only
do so if you already own one. They're not a good investment, due to the
inability to get accurate tables of code values, to allow interpreting
the displayed value. I don't see a table of values in your manual for
example. (Some motherboards come with a Port 80 two digit LED display,
soldered onto the motherboard, to perform this test function.)

3) Enter the BIOS and check the settings. It could be, you haven't set something
that the hardware needs. On an LGA1366, there isn't much room for Vdimm adjustment,
due to the limited 1.50 to 1.65V suggested operating range. If you have purchased
memory which needs more than 1.65V, you'll need to set the memory to more
slack timing settings, to have it work well. Normally, a three DIMM "kit"
for an LGA1366 system, will stay within the 1.65V limit.

Things we can't really resolve that way:

1) Unknown kind of AYX power supply issue. Swap in another one (even one with a
lower power output rating) for a quick check. Does the system become more
stable ?

2) Consider the possibility that the motherboard is defective.

Paul
 
K

keylog keylog

Thank you Paul,

In fact, when my system was unable to boot, I saw exactly the behavior you mentioned in your first paragraph: The motherboard gave a black screen with a "short beeps sound" that was repeated (loop) after automatic reset (stuck in reset as you mentioned)...
 
P

Paul

keylog said:
Thank you Paul,

In fact, when my system was unable to boot, I saw exactly the behavior you mentioned
in your first paragraph: The motherboard gave a black screen with a "short beeps sound"
that was repeated (loop) after automatic reset (stuck in reset as you mentioned)...

The beep is generated under BIOS code control. If the BIOS is not working,
there can be no beeps.

To make a sound, requires the BIOS code to run.

When the computer is completely silent, and black screen, then that's a hint
that perhaps no BIOS code is running.

Beep sounds are used, whenever video is unavailable. For example, if there
is no system RAM present, and a video card is plugged in, the code to drive
the video card likely needs a bit of RAM to work with. Since the video card
can't be operated that way, the only choice is to beep.

There is some BIOS code, which is register based, and needs no RAM to work.
The processor reads instructions out of the BIOS EEPROM, and uses the registers
in the processor for storage. This is how the BIOS can beep the speaker,
when no RAM or video card is present.

But other BIOS code, eventually requires the reserved RAM areas to be
present and working.

When the computer knows (or thinks it knows) video is working, the
error message can be printed on the screen. For example, on my older
computers, if I unplug the PS/2 keyboard and mouse, I'll get a
BIOS message printed on the screen telling me that. There is no
need to beep, because the motherboard knows I can see the screen,
and the PS/2 error stops the system and prevents booting (unless
that kind of error detection is turned off).

If the video is available, and the POST sequence finishes, the speaker
will beep once. This is similar to a "lamp test" feature on other
electronic devices. The single beep is there, to prove to you that
the computer case speaker is working. The single beep may also
corroborate that the computer thinks everything is running - for
example, if you got a black screen, and a single beep, then the
computer believes it has successfully programmed the video card,
so then you'd go looking for a broken or disconnected monitor as
the reason for the black screen.

*******

You have "short beeps sound", but it has yet to be proven that
the error message associated with that is "Power Error". What it
could be is a RAM error. Remove all RAM and see if the symptoms
are the same or not. By comparing beep codes for various
test cases, to the sound you heard, you may be able to deduce
what the error beep pattern means.

For example, if one of my computers beeps twice or three times,
I know it is RAM or video. But I don't know which. And I seldom
believe what it says in the manual, as they may have switched the
values in the actual design.

In that regard, POST behavior doesn't come with iron clad
guarantees. What is written in the manual, is done by a
technical writer. And the tech writer, doesn't always have
access to the design engineer. I know this, because of
how the management at my company did things. Sometimes, they'd
coax the tech writer, to "go away" :) And that forces the
tech writer, to "make up stuff" and "be creative". And it's
done that way, because the engineer may be working on the
next motherboard, and management wants his time spent
strictly on the next new design. Leaving the tech writer
in a "vacuum". All in good fun of course.

Paul
 
E

Edwin T.

yeah dude..sometimes mem sticks need to be removed and replaced after
sitting around for a few years or with fluctuating temperature changes.(if
your computer is not within normal temp environment) Just follow the last
poster and clean em
 
S

ShadowTek

I booted with the memtest86 cd-rom and I tested it for 9 hours without errors and my system didn't freeze during the tests.

I've had bad memory escape detection on a full 24 hours of memtest. I'd
suggest 48 hours as a minimum.
 
P

Paul

ShadowTek said:
I've had bad memory escape detection on a full 24 hours of memtest. I'd
suggest 48 hours as a minimum.

Memtest isn't that good. You need a stress tester type application,
something that causes more stress than memtest does. And then that
test interval is put to better usage.

Some day, somebody will write a tool like memtest (something that
boots from floppy or CD), and applies stress like Prime95 or some
of the more modern stress testers, and without an OS being present.
That would maximize the amount of memory addresses that could be
stress tested.

Also, the memtest test cases, aren't of equal value when it comes
to equivalent stress. How many times does an error show up in
Test #1 versus Test #5. If you're going to leave it baking for
24 hours, then using the test most likely to complain, would be
the best choice. Some of the tests seem to be relatively useless,
which means you might want to run them for one pass for completeness,
but when "baking", run just the test that is most likely to flag
the memory.

If the source code for memtest is still available, you might even
be able to modify it according to your purposes. I made a three
line mod to the program years ago, and managed to get the build
process to work. I haven't tried anything like that lately, so
don't know how easy it is now.

Paul
 
L

larry moe 'n curly

ShadowTek said:
I've had bad memory escape detection on a full 24 hours of memtest. I'd
suggest 48 hours as a minimum.

Unlike MemTest86, memtest isn't very good because it runs only from
Windows, unless you pay HCI Design $15 for the self-booting version.

I'd also try Gold Memory because I've seen it find errors in a few
minutes that took MemTest86 several hours, and vice-versa.

Does anybody know of a recent evalution of memory diagnostics? The
newest one I've seen is RealWorldTech's, which is over a decade old.
 
Top