PC fails Prime95 at stock

[ajgoodfellow]

Hi

Having recently got my FX-62 back from RMA my PC still isn't working properly (the old chip definitely was faulty though and it'd only run at 2GHz)

At 2.8GHz, 1.4V Prime95 will fail on both cores after about 15 minutes. If I increase the voltage to 1.45V Prime95 will fail after about 25 minutes.

The chip gets hot but not too hot (65 degrees C). This is with a Zalman CNPS9500-AM2 cooler at full speed (very noisy unfortunately!) and an Antec Amber 120mm case fan.

I'm using an Antec Neo HE 550W PSU and the voltages are rock solid under load at 3.30V, 5.05V and 12.20V. I'm using Geil PC6400C4 memory at stock (400MHz 4-4-4-12 2.15V) on an MSI K9N SLI Platinum motherboard (nForce 570 Ultra)

The PC worked perfectly with my temporary X2 3800+ @ 2GHz stock and it was also far quieter as the Zalman cooler could be on low speed!

I'm running Vista Business 64-bit. The PC is stable apart from in Test Drive Unlimited which caused me to try Prime 95

Does anyone have any ideas?

 

[ajgoodfellow]

65 degrees is too hot IMHO, I think you need to re-seat the cooler. Take it off clean off all the old thermal paste, examine the base of the cooler for any defects and lap if neccesary then re-fit using some decent thermal paste.

/Edit - My 4200 X2 @ 2900Mhz sits at 31c Idle and 46 with both cores under load.

Thanks for your replies. It does sound hot to me and I've reseated it but I'm not sure why it's still so hot. Maybe the FXs are hot CPUs? I'll look into it though as I like a quiet PC and the Zalman at full whack definitely isn't quiet!

I've been trying Orthos Beta with Small FFTs and that's been going for an hour now on both cores without any problems (touch wood) and the temp's currently at 63 degrees.

I had been using Large FFTs on Prime95 when it failed - perhaps it's a memory issue?

 

[ajgoodfellow]

I haven't actually. I might test it using Memtest and see if that flags up anything

I'm worried about the temp though - I put the side of the case back on and it climbed to 80 degrees so I promptly removed it! I'm sure it used to be fine ages ago when I first built the PC with the old FX chip

Which way around should I have the Zalman cooler? Should the fan on the Zalman cooler be facing the case fan or should it be the other way around? There seems to be a hot spot behind the cooler at the moment. I've currently got the fan on the Zalman cooler facing the case fan (they're both on the same side)

Currently at 1hr 25 with Orthos Beta on small FFTs so maybe it's not the CPU, it's the memory

 

[ajgoodfellow]

I've just had a look and I can't see any that are touching it. I'm not sure that it is mounted properly as I can twist it slightly from side to side. I've reduced the voltage back to 1.4V on the fly (using the MSI Dual Core Centre) and that's reduced the temp to 58 degrees. I wonder whether that'll reduce it's stability at all

 

[ajgoodfellow]

Antec Sonata (the original one)

 

[ajgoodfellow]

To the OP, seeing as you had prior problems, it may not be the chips, could be the power circuit on the motherboard not being up to scratch for the load of the FX-62. Have u tried a different board? The increase of stability with increase of voltage leads me to think its power related somewhere along the line.

Thanks for the reply - that's a good idea

I was using an ASUS M2N-E before which worked well until I started to have some problems with the system. I changed the CPU which solved most of the problems but it still wasn't completely stable so I changed the motherboard to the MSI K9N SLI Platinum

 

[ajgoodfellow]

It passed an hour of S&M - I just hit go and it tested both cores of the CPU then the available ram (not much in Vista!) but the temp didn't go above about 55 degrees - far cooler than the 60 degrees+ under Orthos Beta

I've increased the voltage to the memory to 2.2V from 2.15V and I'm currently running the Large FFTs test on both cores (two copies) using Prime 95

Good point about the temp. I'll reseat the CPU - perhaps I used too much Artic Silver - and I'm looking at how I might improve the airflow as as far as I can see I can't mount a fan in the front of the case. I'll have a look on the Zalman website too to see if they suggest which war around to mount the cooler - if it's the wrong way then it might be fighting against the case fan and disrupting the airflow

 

[ajgoodfellow]

Although that seems to have helped it's still crashing

Looking at the Zalman website it looks like I've fitted it the wrong way round! I'll try re-fitting it and see if that hopes with temps and hopefully stability

 

[ajgoodfellow]

I've reseated the CPU cooler the right way round and that has helped with the temps - down to 53 degrees on large FFTs in Prime95 so far better!

Unfortunately it still doesn't seem stable. I've tried it at 1.4V core and 2.2V memory but it still fell over. I left the machine with it going at 1.45V core and 2.2V memory but I came back an hour later to find that it had rebooted!

I'm currently watching it at 1.45V core and 2.2V memory to see what happens. If it still doesn't work then I'll try the 2T suggestion

EDIT: Just had a thought. Would an overheating northbridge cause any of these problems? Unfortunately the MSI board is pretty poorly designed in this area and it only has a thin heatsink over the northbridge (no fan or heatpipe) which gets very hot

EDIT: Nope, just fallen over again. I'll reboot with 2T timings and give that a go

 

[ajgoodfellow]

Hope not! It only hit 80 degrees once but it did live around 70-75 degrees at one time which is too hot

It currently seems to be far more stable at 2T though - thanks again for the suggestion.

Prime95 has been running Large FFTs for over an hour now (far longer than before - touch wood!) without failing at 1.45V core 2.2V memory, 4-4-4-12 2T. The CPU temperature has stabilised at 56 degrees

 

[ajgoodfellow]

An update - the Large FFT test has been running for 2 hours 30 minutes without falling over so it looks like the 2T has fixed it

Thanks for all of your replies

 

[ajgoodfellow]

Hi!

Thanks for your replies - sorry for my late response!

2.2V could well be a little high as I've now found out the problem. I wouldn't like to reduce it below 2.15V though as that's the stock voltage for my memory IIRC

Not sure about the memory slot issue though - I'll definitely give that a go though as that could help