prime95 2nd test error

[Lifthanger]

Hi.
My actual system is in my sig. The problem is:
the first stress test (fpu intensive), was stable until I stopped it after about 8 hours, so is the third test (ram intensive). But the second test stoppes due to an error after about 40 minutes. Which hardware component could be the problem? CPU and ram should be fine according to test 1 and 3. The board should do 200FSB flawlessly, as its built for it - the ram should do as well. CPU on die temperature doesn't even get over 50C and system is very unimpressed with a constant 32C.Also Voltage should be enough to handle this overclock. So what does the second test do? some gurus out there who can give me hand with this one?
Wanted to overclock this cpu to at least 2,5GHz.. I don't want to live on with an unstable 2,2GHz

edit : sometimes an error occurs just after 1 minute of testing.. this can't be ok

 

[dicecca112]

your cpu idle is 45C? What Heatsink are you using? 45C on idle seems high to me, especially if your system is only 33

 

[Lifthanger]

I'm using the slk900 with an 45cfm 92mm fan.
these temperatures are really good, I think the board reports a bit high.
my old athlon xp 2000+ was always between 55C and 60C. This board reports on die temperatures, which are about 10 to 15 degrees higher than socket temps.
Temps are no problem.. in fact, the opposite is the case, I'm really proud that I can stay below 51C under full load for hours

 

[dicecca112]

okay good. well to rule out the ram completly run memtest to be sure. Your right about the cpu voltage, that should eb more than enough. Did you try bumping up the Vdimm voltage?

 

[Lifthanger]

I've yet to try memtest86.. good idea thanks.
my problem is, that 2,85V is the max my board is willing to do.
But the ram is rated at 2,6V and its running in its specifications (speed).
But to be sure I'll run memtest tomorrow.

 

[PCGUY112887]

My 2600+ runs at 2.65ghz with 42*C idle... and my cooling isn't all that good. Front intake doesn't really take in any air .

 

[Lifthanger]

damn I know that everyone else has lower temperatures than me, but that's just because everyone else hase a different motherboard.. its like comparing two identical motors with different gearboxes

 

[Lifthanger]

all right I've run memtest at 410MHz and all was fine..
according to this
prime test 1 stresses cpu to the max and test 3 stresses ram, but both tests perform well, while the second one crashes at once or after half an hour.
could it be possible that the NB gives bad results? Who knows what the second test is doing? Should I ignore this error?

btw. at this link, you can read also that gigabyte idle temps under 50C are nice.

 

[dicecca112]

hmm interesting. How is the stability of the system?

 

[Lifthanger]

except one sudden reboot yesterday, all is fine.
system is runnig almost the whole day, stable.
maybe its fluctuation of the vcore values, but then again the first test ran for about 8 hours. it would be great to know exact differences between the tests.
second test has most power output, so maybe its the PSU, but it has 550W and very high currents on all voltages. how could I test it?

below are the rails of my PSU, unfortunately easytune and MBM5 show different vaules, maybe I should try to use a multimeter, but I've no idea where. Maybe I should do some research on this..

the values recorded by MBM5 are really low on the 5 and 12V rail..
according to MBM5 vcore is rippling by 0,07V

 

[Lifthanger]

I've measured my PSUs rails with an multimeter

12V(max 24A) : 11.89V
5V(max 40A) : 5.17V
3.3V(max 32A) : 3.39V

all values are under load and stable

is 24A on the 12V not enough?

 

[Lifthanger]

now the 2nd test ran for quite some time and 5 minutes after I stopped it my system rebooted 3 times... wtf

 

[oc_byagi]

HIGH FTT test (second test) uses the CPU most, and yes, it is the HARDEST TO PASS when you're overclocking (at least for socket A platform). It's very easy to get errors with minor overclock. Like my system, I was able to boot my XP-M up to 2.7ghz into windows, but I could only get it to stable at 2.5ghz in HIGH FTT test. I don't exactly know if you're giving enough voltages to cpu but seems like maybe you didn't have luck on the cpu. (btw, how is the mobo doing? I never heard anything about gigabyte socket A board, maybe you might want to take a look at some of the bios settings)
the fact that you passed the THIRD test pretty much rules out memory (also the memtest)
failling second test leaves you with combinations of either CPU, MOBO, or PSU
is there any replacement for PSU or CPU to test??? try upping the vcore since the mobo may be undervolting.
also, very important, use MBM5 and monitor the voltage lines WHILE running prime95. that way, it'll give you how powers are going through under intensive loads

 

[Lifthanger]

the board is one of my biggest problems. It won't let me go ANY higher on voltages. But I've read about them, that they should do 220FSB most of the time, which is more than enough, as I'd be happy with 210...
its an nforce 2 ultra 400, so it should do just fine..
but I'm not confident anymore.. Maybe I had no luck with MB, cpu,PSU and mem...
unfortunately I've no other CPUs and PSU around.

I'll prime for some time and monitor voltages with mbm5

thanks for your help

 

[oc_byagi]

well, you gotta tell me all you can tell me, if you're willing to find out the problem.
so how were monitoring the voltage line go? did you find any jumps or dips?
lower the clockspeed and test, that'll cancel out the CPU as well. (try something like 1.8ghz or 2.0ghz. 200x10, 200x9)
what brand is your power supply? (I'm guessing it's 550W from your sig)
Are the CPU temperatures you're seeing correct? check bios, mbm, motherboard utility (if you have), touch the heatsink, and fee the air being blown away.

 

[Lifthanger]

I'll do what you say and of course I'm very glad that you're helping me.

mutimeter check on molex got the same results as before.
while the 2nd test is running, the rails are behaving as follows:
all values according to mbm5
cpucore: no change - fluctuates between 1,67 and 1,74
ram: no change - fluctuates between 2,83 and 2,88
3,3V : is stable when idling, but while testing it start moving like mad, but only between 3,33 and 3,38
5V: doesn't move 4,08V all the time
12V: values are more constant under load, but they're 0,15V lower - 11,25V

my PSUs brand is super-silent (if this is a brand at all)
my board doesn't suplly lower multipliers, but I'll 9X at once

all utilities I'm using show exactly the same temperatures. I'm sure they are lower because the heatsink is not that warm. And my cooling efforts should make up for lower values. But I think is caused by the motherboard.. I've read in many forums that gigabyte ga 7n400 users are asking for help because of their high temperatures. One explanation was, that these boards just report 10C more to protect your system.

edit: all right, now im running at 9.5*205 which is below stock speed
I'll try prime95 now

edit: after reading through many posts (again) I can definitly say that my temps are great.. my temps are more than 10C lower than other users with the same board. Most of them hit above 60C under load - I stay below 52C.

edit: right at this moment the 2nd test is running for about 30minutes
what I can see so far is, that the airflow coming out of the PSU is much colder than with 2260MHz and the 12V line is more stable at a higher value..

edit: all right while last editing this post, prime failed with 896K FFT length after 30 minutes. this means my cpu cannot be the problem.
I'm glad and sad at the very same time... help

if noone's going to help me I'll kill one hostage every 30 minutes..
*takes screwdriver and aimes at mobos mosfet*
you're next goddamned PSU

 

[oc_byagi]

hahaha, sorry I was away for a while since I'm having minor headache today. (major lack of sleep over the weekend)
Well, seems like you're doing good so far, trying to rule out one part at a time. Since the PSU is a no-name brand, there's also a possibility there, since you've said the system was running nice and stable motherboard may not be a problem. well, let your cpu prime a bit more and post some results. (I say 2 hr minimum before making a decision to rule out your cpu).

 

[Lifthanger]

my cpu passed 8 hours of the first test at 11*205..
I'm at 200*9 at the moment and the 2nd test is failing.
I guess the first test would work as it even worked at 11*205..
maybe its really the PSU.. I've paid 65€ for it... damn it

 

[oc_byagi]

Yeah, it may well be it, you should look into brands like forton, antec, enermax, for quality power supplies.

 

[Lifthanger]

hm but what exactly is the problem with the PSU? the voltages and currents are just ok... maybe some mods could do the job.