F
FKS
Is there a good program to test an Intel motherboard? I checked out Intel's
web site but couldn't find any info.
web site but couldn't find any info.
You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
J
John Doe
FKS said:Is there a good program to test an Intel motherboard? I checked out
Intel's web site but couldn't find any info.
Good question, IMO, mainboard makers should make one. But that would
cost one penny per mainboard sold (a wild guess).
R
Rebecca
FKS said:Is there a good program to test an Intel motherboard? I checked out
Intel's web site but couldn't find any info.
Itel provides diagnostics for their Dual and Quad XEON Server Motherboards
(the boards and CPUs incorporate Boundary Scan design throughout so they
_can_ be diagnosed). Nothing for their Gamer entry-level boards. Same with
most vendors.
F
Franc Zabkar
Good question, IMO, mainboard makers should make one. But that would
cost one penny per mainboard sold (a wild guess).
How hard would it be to allow access to the POST routines? For
example, could the motherboard/BIOS manufacturers implement a BIOS
based, user controlled debug monitor that could exercise some or all
of the POST routines in a continuous loop?
- Franc Zabkar
D
DaveW
NO, there is no such program.
F
Franc Zabkar
Good question, IMO, mainboard makers should make one. But that would
cost one penny per mainboard sold (a wild guess).
IIRC, some old 386/486 motherboards had low level formatting (for
older MFM HDDs) and hard disc diagnostic utilities built into the
BIOS. I'd also like to see something like Memtest86 incorporated into
a BIOS diagnostic module.
- Franc Zabkar
R
Rebecca
Franc said:IIRC, some old 386/486 motherboards had low level formatting (for
older MFM HDDs) and hard disc diagnostic utilities built into the
I only remember that on some 286 machines. Dell did it.
BIOS. I'd also like to see something like Memtest86 incorporated into
a BIOS diagnostic module.
Turn off Quickboot in BIOS. Most have a pretty good Memory Test, but most
users disable it.
F
Frank McCoy
In alt.comp.hardware.pc-homebuilt "Rebecca"
Actually, the memory test in just about any BIOS is almost worthless.
Better than *no* memory test, but not much.
I never once had the so-called "memory test" in BIOS find a problem;
even when the problem WAS bad memory! I have several bad memory sticks
left in a special box, that pass every BIOS memory test with flying
colors; yet cause Windows to crash intermittently, or not load at all.
If you want to check your memory, memtest86+ does a fair job.
http://en.wikipedia.org/wiki/Memtest86
http://www.memtest.org/
Turn off Quickboot in BIOS. Most have a pretty good Memory Test, but most
users disable it.
Actually, the memory test in just about any BIOS is almost worthless.
Better than *no* memory test, but not much.
I never once had the so-called "memory test" in BIOS find a problem;
even when the problem WAS bad memory! I have several bad memory sticks
left in a special box, that pass every BIOS memory test with flying
colors; yet cause Windows to crash intermittently, or not load at all.
If you want to check your memory, memtest86+ does a fair job.
http://en.wikipedia.org/wiki/Memtest86
http://www.memtest.org/
F
Franc Zabkar
In alt.comp.hardware.pc-homebuilt "Rebecca"
Actually, the memory test in just about any BIOS is almost worthless.
Better than *no* memory test, but not much.
I never once had the so-called "memory test" in BIOS find a problem;
even when the problem WAS bad memory! I have several bad memory sticks
left in a special box, that pass every BIOS memory test with flying
colors; yet cause Windows to crash intermittently, or not load at all.
If you want to check your memory, memtest86+ does a fair job.
http://en.wikipedia.org/wiki/Memtest86
http://www.memtest.org/
I've had much the same experience as you. I believe the reason that a
BIOS based memory test fails to catch many RAM faults is that it reads
a memory address soon after writing it. This does not allow the data
in a faulty memory cell to decay.
- Franc Zabkar
F
Frank McCoy
In alt.comp.hardware.pc-homebuilt Franc Zabkar
ignore addressing problems. If you write to a specific address and
write the SAME data to many others, that's no real assurance that the
address you were actually writing to was the one you wanted, or the data
you got back came from that address.
That's one (of many) reasons I recommend a memory test that writes
*different* and preferably *random* data throughout memory, then returns
after the write and checks to see if the data stored matches the pattern
put.
It's amazing how many memories fail that sort of test that pass all
others.
I *think* from what others have told me, that memtest86 does that sort
of testing; though only after several passes of less sensitive tests.
I personally would make it about second or third; after all ones and all
zeroes, to test the more obvious faults.
But I, at present, have no faults to find with that program.
That, and the simple test they use with fixed data patterns completelyI've had much the same experience as you. I believe the reason that a
BIOS based memory test fails to catch many RAM faults is that it reads
a memory address soon after writing it. This does not allow the data
in a faulty memory cell to decay.
ignore addressing problems. If you write to a specific address and
write the SAME data to many others, that's no real assurance that the
address you were actually writing to was the one you wanted, or the data
you got back came from that address.
That's one (of many) reasons I recommend a memory test that writes
*different* and preferably *random* data throughout memory, then returns
after the write and checks to see if the data stored matches the pattern
put.
It's amazing how many memories fail that sort of test that pass all
others.
I *think* from what others have told me, that memtest86 does that sort
of testing; though only after several passes of less sensitive tests.
I personally would make it about second or third; after all ones and all
zeroes, to test the more obvious faults.
But I, at present, have no faults to find with that program.
F
Franc Zabkar
That, and the simple test they use with fixed data patterns completely
ignore addressing problems. If you write to a specific address and
write the SAME data to many others, that's no real assurance that the
address you were actually writing to was the one you wanted, or the data
you got back came from that address.
I've just had a look at the BIOS listing for the original IBM AT.
Test #19 of the POST includes a rudimentary address test. It generates
three possible error codes:
201 - data compare error or parity
202 - address line 0-15 error
203 - address line 16-23 error
Among other things, test #19 writes xFFFF to the first and last
addresses within each 64K block, and x0000 to all other addresses.
IMHO a traditional "data=address" test would have been much more
thorough.
- Franc Zabkar
F
Frank McCoy
In alt.comp.hardware.pc-homebuilt Franc Zabkar
But then, I originally wrote my monograph on testing with random
patterns (and why to do so) about the same time IBM came out with their
PC; so I can't really fault them for not reading it back then.
However, I *can* fault all the BIOS writers ever since for not doing
something far better, when such tests became far more common in the
testing world. But that (an improved memory test) just doesn't seem to
be a priority to ANY of the people writing BIOS software.
"If it ain't broke, don't fix it," seems to be the standing attitude.
Only the problem is: The memory-test in the IBM PC was broke from the
day the machine was introduced. ;-{
Or, as I have said many times, a random-pattern test.I've just had a look at the BIOS listing for the original IBM AT.
Test #19 of the POST includes a rudimentary address test. It generates
three possible error codes:
201 - data compare error or parity
202 - address line 0-15 error
203 - address line 16-23 error
Among other things, test #19 writes xFFFF to the first and last
addresses within each 64K block, and x0000 to all other addresses.
IMHO a traditional "data=address" test would have been much more
thorough.
But then, I originally wrote my monograph on testing with random
patterns (and why to do so) about the same time IBM came out with their
PC; so I can't really fault them for not reading it back then.
However, I *can* fault all the BIOS writers ever since for not doing
something far better, when such tests became far more common in the
testing world. But that (an improved memory test) just doesn't seem to
be a priority to ANY of the people writing BIOS software.
"If it ain't broke, don't fix it," seems to be the standing attitude.
Only the problem is: The memory-test in the IBM PC was broke from the
day the machine was introduced. ;-{