AMD Torture Test FAILS!

[fibersnet]

Hello, this is sort of a complex problem.

System setup:

AMD XP 3200+ Barton
Abit NF7 rev2
2 x Kingston 512MB

Full Load CPU Core is 54C

I run mprime in torture test mode. All is fine until 4 hours 28 minutes and then it gives an 'unexpected rounding error'. Which basically means the CPU returned a faulty result. How do we know? It is being compared against known results.

Okay, I run it again thinking it is some kind of fluke. B00M, right there 4 hours 28 minutes and it gets the same error.

What would cause this system to produce the same error in the exact same time?

Thanks.

 

[MDE]

Try to reinstall a fresh copy of Prime95 (download it again) or bump up the RAM or CPU voltage.

 

[fibersnet]

Okay, Just did that, running it again.

I bumped cpu voltage to 1.7 ram should be next?

@ full load:

benchmark@energy:~$ sensors
w83627hf-isa-0290
Adapter: ISA adapter
VCore 1: +1.63 V (min = +1.62 V, max = +1.78 V) ALARM
VCore 2: +2.62 V (min = +1.62 V, max = +1.78 V)
+3.3V: +3.28 V (min = +3.14 V, max = +3.46 V)
+5V: +4.92 V (min = +4.73 V, max = +5.24 V)
+12V: +11.80 V (min = +10.82 V, max = +13.19 V)
-12V: -11.95 V (min = -13.18 V, max = -10.88 V)
-5V: -5.05 V (min = -5.25 V, max = -4.75 V)
V5SB: +5.48 V (min = +4.73 V, max = +5.24 V)
VBat: +3.47 V (min = +2.40 V, max = +3.60 V)
fan1: 3813 RPM (min = 3497 RPM, div = 2)
fan2: 0 RPM (min = 4927 RPM, div = 2)
fan3: 0 RPM (min = 1400 RPM, div = 4)
temp1: +36 C (high = +19 C, hyst = -79 C) sensor = thermistor

temp2: +54.0 C (high = +75 C, hyst = +70 C) sensor = thermistor

temp3: -48.0 C (high = +75 C, hyst = +70 C) sensor = thermistor

vid: +1.700 V (VRM Version 9.0)
alarms:
beep_enable:
Sound alarm disabled

 

[Budman]

Test your Ram.

http://www.memtest.org/

 

[fibersnet]

Okay, memtest86 it is.

 

[fibersnet]

Update:

memtest86
25 hours testing, 51 tests, 0 errors.

 

[dszd0g]

Originally posted by: fibersnet
What would cause this system to produce the same error in the exact same time?

If you redownloaded the test, it is probably unlikely to be the test that is bad. You can try running the same download on a different machine to verify this.

Do you know what stepping your Barton is? It would be nice if someone with the same stepping on this forum could verify that the AMD torture test works for them and is not a flaw with the stepping. If one checks the errata for AMD and Intel CPUs the errata tends to be rather long.

I would guess that it is a bad CPU. Just because the CPU overall functions fine does not mean that their is not a flaw in the silicon. It could have even passed the same test if/when AMD tested it or if you ran the test right after you bought the system. Over time or if more than the rated voltage was applied a defect in the silicon could occur. If the problem is a rounding error then you need to decide how critical that is to you, if you use it for your personal finances would it matter to you if it was a penny off here and there?

 

[imported_Phil]

Try underclocking it and see if you get the same error. That'll tell you if it's a genuinely faulty CPU that was put in the wrong speed bin at the factory.

 

[vailr]

If not running with the latest motherboard bios version, then try updating the bios. For Intel CPUs, at least, an updated bios can include "updated CPU microcode", which can compensate/bypass known CPU errata. AMD motherboard bioses may also include a similar microcode update.

 

[fibersnet]

Thanks for the help

Oh, I kept on running memtest86 to verify the memory.
It has been running 49 hours now, 0 errors.
So now I am going to replace the hard drives. (You never know)

This seems much like a logic error in the processor perhaps?
Getting the same error at the exact same time eliminates overheat issues etc...

fibersnet@energy:~$ cat /proc/cpuinfo
processor : 0
vendor_id : AuthenticAMD
cpu family : 6
model : 10
model name : AMD Athlon(tm) XP 3200+
stepping : 0
cpu MHz : 2194.704
cache size : 512 KB
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 1
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 mmx fxsr sse syscall mmxext 3dnowext 3dnow
bogomips : 4341.76


I am just going to install windows. Hopefully this 'bug' will go away.
I already RMA'ed the CPU in the first place because I thought the CPU did not meet up to 3200 spec.

I should have saved the stepping of my old am3200.