Prime95 failure on A7N8X-E Deluxe

[LooseScrew]

Prime 95 fails on every test with my settup. I checked the memory for 24hours, cant remember the
prog name. PC is rock solid otherwise. Can anyone tell me what else I can check? According to the
author my system is unstable.

A7N8X-E Deluxe
AMD 2600+
1 gig PC2700 ram
Sony CDR
BFG TI4200 128m
440w PS
20gig boot drive
80gig secondary
Thermaltake copper heatsink and fan on CPU

Where should I begin?

TIA

 

[BigBadger]

What temperatures are being reported for the CPU while it's on full
(Prime95) load? What voltages are being reported while on load? Failure
could be caused by overheating (say >60C), or by low voltage on the CPU or
on the 3.3V, 5V or 12V power rail..

Some folk say you can ignore Prime95 failure is the system appears otherwise
stable...I don't agree with them. A stable system should be able to run any
(non buggy) software without failure. In my experience if it won't run Prime
you will eventually run into problems with some other program.

 

[Kylesb]

Are you overclocking?

--
Best regards,
Kyle
| Prime 95 fails on every test with my settup. I checked the memory
for 24hours, cant remember the
| prog name. PC is rock solid otherwise. Can anyone tell me what else
I can check? According to the
| author my system is unstable.
|
| A7N8X-E Deluxe
| AMD 2600+
| 1 gig PC2700 ram
| Sony CDR
| BFG TI4200 128m
| 440w PS
| 20gig boot drive
| 80gig secondary
| Thermaltake copper heatsink and fan on CPU
|
| Where should I begin?
|
| TIA

 

[LooseScrew]

System is not overclocked.

Temp for CPU is 48-49C

MB is 24C

3.3v is measuring 3.232

5v is measuring 4.75 < is this significantly low?

12v is measuring 12.288

Vcore is at 1.632

This is a 420w PS, not an expensive one though. It came included with my alienware style case.

 

[LooseScrew]

BTW My PC fails Prime95 torture test on the very first test EVERY time. At least its consistent in
its failure.

 

[BigBadger]

Your temps look fine (assuming they are taken under load). The 5V looks on
the lower limit of what is acceptable (about +/- 5%), this could be part of
the problem. Your Vcore is a little below what it should be, 1.65V is the
core default voltage. As your temps are fine I would increase the voltage a
little, try 1.675 or 1.7V and see how it performs in Prime, also keep an eye
on the core temperature after the voltage increase.
One other point: have you got the auxiliary ATX power connector (square 4
pin connector) plugged into your board? (if the board has one you should
have it connected)
--
*****Replace 'NOSPAM' with 'btinternet' in the reply address*****

System is not overclocked.

Temp for CPU is 48-49C

MB is 24C

3.3v is measuring 3.232

5v is measuring 4.75 < is this significantly low?

12v is measuring 12.288

Vcore is at 1.632

This is a 420w PS, not an expensive one though. It came included with my alienware style case.

 

[LooseScrew]

I will give this a try. As far as I know this MB doesn't have an auxiliary power connection, I could
be wrong but I didn't see one.

 

[Arnie Berger]

Prime 95 fails on every test with my settup. I checked the memory for 24hours, cant remember the
prog name. PC is rock solid otherwise. Can anyone tell me what else I can check? According to the
author my system is unstable.

A7N8X-E Deluxe
AMD 2600+
1 gig PC2700 ram
Sony CDR
BFG TI4200 128m
440w PS
20gig boot drive
80gig secondary
Thermaltake copper heatsink and fan on CPU

Where should I begin?

TIA

The first thing I would do is find a computer where Prime 95 passes
everytime. Just to certify that it is possible to pass. Also, I assume
that this is a benchmarking program, so what does fail mean?

Also, you quote voltages, how did you measure them? The only voltage
readings I would trust are those measured with a good digital
voltmeter at the pins of the processor socket. I would trust the
software tools to show me variations in voltage, but not the absolute
values.

Finally, there has got to be other CPU-intensive, freeware
benchmarking programs that you could run as an alternative to Prime
95. Try some of those. If your memory runs memtest reliably, then
that's a good sign. Since this is a CPU intensive test, and the CPU
seems to be crapping out, the only possibility besides something
marginal in the CPU itself is heat. So, I would try to run it slower
in order to generate less heat. The easiest way to do this is to back
off the FSB clock rate. See if you can drop it 5% or so. Does it still
die? If so, how much longer does it take before it goes belly-up. If
it doesn't die, then that's a good clue. Perhaps its time to reinstall
the heatsink and put a good thermal grease there.

HTH

Arnie

 

[BigBadger]

Prime95 is has a pure stability testing application built in....and yes it's
possible to get it to run without error. It's far and away the best program
to test basic CPU/RAM system stability although it does not test graphics.

 

[SPRITE1001]

Heres a question, I ran memtest, no problem. I run CPU intensive programs
reguarly stressing ram, hard drive, and of course the CPU, all without fail.
However when I run Prime 95 it crashes. I dont mean I get an error withint
prime95, I mean the program tanks, shutsdown, etc. This is the only thing my
system has a problem with. Anyone have an idea?

Thanks
~Arie

System
Asus K8V Deluxe
AMD 64bit 3200+
Corsair PC3200 1Gig (512x2)
Hitachi SATA 80gig (system)
IBM ATA 120gig
Lite-On CDRW
Windows XP PRO (post SP1)

 

[LooseScrew]

Mine seems to give an error on the first iteration every single time. Meanwhile a nearly identical
PC next to it runs flawlessly. Only difference is slightly slower PC and Ram. I may start to pull
parts from the error free once to try to isolate the problem if I dont find out something soon.

 

[Ed]

Heres a question, I ran memtest, no problem. I run CPU intensive programs
reguarly stressing ram, hard drive, and of course the CPU, all without fail.
However when I run Prime 95 it crashes. I dont mean I get an error withint
prime95, I mean the program tanks, shutsdown, etc. This is the only thing my
system has a problem with. Anyone have an idea?

Thanks
~Arie

System
Asus K8V Deluxe
AMD 64bit 3200+
Corsair PC3200 1Gig (512x2)
Hitachi SATA 80gig (system)
IBM ATA 120gig
Lite-On CDRW
Windows XP PRO (post SP1)

Strange, three AMD64 systems here on MSI boards running WinXP Pro, prime95 runs just fine, you download and running the
latest version of prime95?

Ed

 

[LooseScrew]

Yes p95v238.exe is the latest version.

Strange, three AMD64 systems here on MSI boards running WinXP Pro, prime95 runs just fine, you download and running the
latest version of prime95?

Ed

 

[LooseScrew]

From the mersenne forum I found

QUOTE

FATAL ERROR: Rounding was 0.4997747507, expected less than 0.4 << The error I get.

It means your computer is making errors while running Prime95.

the most common problems are old or wrong drivers, especially sound card and video drivers.

Update all drivers and Win2K. Make sure they are the correct drivers.

If this fails, slow down your memory settings in the BIOS. Use "SPD" or "Auto".

In an attempt to educate me what would sound or video drivers have to do with a rounding error?
Thanks!

If they use floating point instructions and change the rounding mode or other
settings in the floating point unit and don't restore the settings when done.

MMX also uses the same registers and can messup floating point registers
and the fp status register if they aren't cleaned up after use.

In that case wouldn't running Prime95 in "safe mode" be a better test of hardware stability?

Yes, safe mode should be better for a hardware test of CPU and RAM.

It wont stress the hard disk normally (unless windows starts swapping virtual memory to disk because
too much RAM is allocated).
Also wont stress the network card.

It should also give the best performance for Prime95.exe for prime searching.
There wont be as many processes requesting CPU time so Prime95 can get more.

Prime95's priority could be increased to provide extra workload, this will make the system less
responsive so it IS NOT recommended.
Especially if the priority of both threads is raised.




Going to try this in Safe Mode Next...

 

[Paul]

Mine seems to give an error on the first iteration every single time.
Meanwhile a nearly identical PC next to it runs flawlessly. Only
difference is slightly slower PC and Ram. I may start to pull
parts from the error free once to try to isolate the problem if
I dont find out something soon.

The other posters have suggested some reasons it may be failing.

I would add to their suggestions, that you use MBM5 (Motherboard
Monitor) or the Asus Probe program (one of those programs, not
both programs at once), to monitor your supply voltages. Depending
on how long you can convince Prime95 to run, you may be able to see
MBM5 do an update of its readings, as the Prime95 program is
running. It is possible your PS voltages dip more than they should,
just when the Prime95 program starts to run.

Another weak link, could be the Vcore circuit. When the Prime95
program runs, the Vcore circuit is put to the test. If there
is any weakness in the circuit, the voltage could droop too low
for the processor to maintain error free operation.

To keep the Prime95 program running a bit longer, you could
try dropping the FSB frequency a bit. The test conditions won't
be quite as severe, but it may make it possible to get some
readings from MBM5 while Prime95 is running its test
simultaneously.

Finally, there is always the possibility the processor is
defective. Sometimes, noise problems in the processor only
show under the most strenous conditions. Normally, factory
testing is very good at finding these (as random instruction
sequences used at the factory are even better than Prime95,
for finding bad chips), but the failure condition on the
processor could be stress or aging related, rather than
an "infant" fault.

BTW: The comment about "random instruction sequences" refers
to the difference between sequences of instructions emitted
by compilers, versus those that a human can craft by hand.
Compilers don't emit code that exercises even a fraction
of the whole instruction set, and there will be plenty of
branches and loads mixed into code sequences. A human,
on the other hand, could put 20 floating point instructions,
one after another, which would do a great job of heating up
the FPU block etc.

Paul

 

[LooseNut]

Tests I have now tried (one at a time) without any luck:

Underclocked RAM

Underclocked FSB

Upped Core Voltage to 1.65

XP in safe mode

One stick of SpekTek 512mb PC2700 instead of 2

One stick of Crucial 256mb PC2100

I'm 100% sure its not the RAM now. When it comes to hardware I suppose the CPU is the next possible
culprit but I loath swapping CPUs around. Still I WILL try it after I isolate possible software
problems. I am not totally convinced that running in Safe Mode eliminates and software problems. I
should have a extra drive laying around that I can set up win98 se2 pretty quickly on and try prime
95 on that. Too bad there isn't a dos version.

BTW the test fails immediately, before the cpu can even start to stress.

 

[Kylesb]

<snip>|
| BTW: The comment about "random instruction sequences" refers
| to the difference between sequences of instructions emitted
| by compilers, versus those that a human can craft by hand.
| Compilers don't emit code that exercises even a fraction
| of the whole instruction set, and there will be plenty of
| branches and loads mixed into code sequences. A human,
| on the other hand, could put 20 floating point instructions,
| one after another, which would do a great job of heating up
| the FPU block etc.
|
| Paul


Although I think your comments are generally top notch Paul, I believe
all compilers emit code that includes a fraction of the whole
instruction set.

 

[Nom]

Prime 95 fails on every test with my settup. I checked the memory for
24hours, cant remember the prog name.

Check it again, with MemTest x86

PC is rock solid otherwise.

If it can't run Prime95, then there *IS* a problem

Can
anyone tell me what else I can check?

You need run your memory at SYNC/100% speed. Set all other timings to
Automatic.
Raise the memory voltage a little (from 2.5v to 2.7v).
What brand is your RAM ?
Raise the CPU voltage a *little* (from 1.6v to 1.65v for example).

Any luck ?

 

[Paul]

<snip>|
| BTW: The comment about "random instruction sequences" refers
| to the difference between sequences of instructions emitted
| by compilers, versus those that a human can craft by hand.
| Compilers don't emit code that exercises even a fraction
| of the whole instruction set, and there will be plenty of
| branches and loads mixed into code sequences. A human,
| on the other hand, could put 20 floating point instructions,
| one after another, which would do a great job of heating up
| the FPU block etc.
|
| Paul


Although I think your comments are generally top notch Paul, I believe
all compilers emit code that includes a fraction of the whole
instruction set.

Sorry for my English

My observation is based on a little "looking under the hood"
I did years ago. I'm a hardware guy, so take the comments with
a grain of salt.

I think at the time I was writing a simulator, and had noticed
a big difference between the code being produced by a commercial
compiler and the code coming from GCC. I looked at the intermediate
file that GCC produces, and it seemed to generate a "pseudo code"
for the assembler that was very RISC-like, on a CISC platform.
If I looked in the code, there weren't any complex table
indexing instructions being used, there were discrete address
calculations being done and so on. Maybe 1/3rd of the instruction
set was being used in the code I looked at - hard to say for
sure.

I didn't do a frequency analysis, to see what instructions were
being used what percentage of the time, but it seemed in that
case, that the compiler wasn't using any strategy of finding
the "best instruction for the job" while generating code.
(Not that I expect a human could write such a compiler anyway.)

My second piece of evidence, was an errata issued for a processor
that was in production. In the lab, the designers of the processor
discovered that if certain sequences of floating point instructions
were used, noise in the FPU portion of the chip corrupted the
computed results (that is, switching noise upsetting the Vcore
local to the gates in that section of the chip). The manufacturer
noted that "no compiler will generate this sequence" and as a
result, no fix was created for the bug in later steppings of the
processor. This again implies that manufacturers make certain
simplifying assumptions about what the code that runs on the
processor looks like, and don't necessarily guarantee every
pathological instruction sequence - a fact I find scary, as all
it would take is one "assembler kook" to break the thing.

HTH,
Paul

"assembler kook" - a person who is convinced he can write
better code than a compiler, and will produce a 6" thick
undecipherable uncommented assembler program listing to
prove it

 

[LooseNut]

Thanks for the input. I have isolated the RAM as not being the problem. The PC main RAM is Spekteck
PC2700 512mb (2 total). I put in a Crucial PC2100 256mb stick and was greeted with the same
results. After some research I have found that not only hardware can be the culprit. Some software
drivers can alter the "...floating point instructions and change the rounding mode or other
settings in the floating point unit and don't restore the settings when done". So once I isolate
software as the possible culprit I plan to do a swap with CPUs next. I will keep you all posted.
From searching past usenet messages and forums I know other people have problems as well so I hope
this ends up helping out someone else as well.