random reboots after upgrading to x2 4200+ from a64 3400+

[vexingv]

i had been running an A64 3400+ (venice core, 1600 mhz HT) system for nearly a year now without any problems and the system is not overclocked. with the recent deal at newegg, i upgraded to an x2 4200+ (toledo core). however, since installing the new CPU, i have been experiencing random reboots. there is no BSOD at all and nothing is logged in the event viewer. usually, the mouse becomes sluggish/unresponsive for a second or two before the screens blanks and the system restarts. i have experienced these reboots during a variety of tasks such as playing videos on VLC, watching OTA HDTV via pci tuner card, doing regular desktop/office work, after playing CS:S, listening to music through itunes, using skype via bluetooth headset, and even while idling (another client was accessing files from the PC via samba).

here are my system specs:

• amd64 x2 4200+ @ stock speed; arctic cooling freezer pro 64
  2x1gb ddr400 dual channel using auto/spd settings (pqi 3-4-4-8; crucial 3-3-3-8)
  ecs kn1 lite (nforce4 s939); onboard lan and sound
  BFG nvidia 7600gt
  500 wtt ultra xfinity psu
  hauppague hvr1600
  pci firewire card
  400gb seagate SATA
  liteon dvd-rom, iomagic dvd-rw
  attached via usb/usb-hub:
  • logitech elite keyboard, logitech g5 mouse, xbox 360 wireless receiver, 8-in-1 flashcard reader, laser printer, bluetooth dongle

i really don't know what is going on and have tried several different things to no avail:

• --temporarily increased VDIMM to 2.7V
  --rearranged/used different DIMMS slots
  --switching from dual channel ddr to single channel ddr (though my sticks were not part of a kit, 3-4-4-8 and 3-3-3-8- PC3200, they had been running fine for the past 3 months in dual channel mode)
  --rearranged pci card placement (firewire and tv tuner)
  --made sure that windows is not set to restart during an error
  --i have the latest drivers installed
  --installed the amd dual core optimizer and windows dual core performance hotfix

i don't think that it is a heat issue either as during stress testing the cpu can is usually around 60 C, but that is within the operating range of 49-71 C as stated by amd. the case temp is usually in the low 40 C. the bios is not set to restart/shutdown when it reaches high temps. also, it has restarted within minutes of coming out of standby (meaning the case couldn't have even gotten very hot that quickly)

not sure that it is my psu either, as i think it has more than enough power (it is 500 wtt, SLI certified, and AFAIK has decent specs). up until a few months ago it was even powering 5 IDE hard drives (instead of only 1, like it does now). also voltage, as measured by speedfan, doesn't seem to fluctuate much.

as to stress testing:
-the system restarted at various times during testing with Stress Test Prime (Orthos). a few times it would restart within 30 minutes of testing. other times it would run for 1 hr, 2 hrs, 3 hrs, 8 hrs, or 13 hrs before rebooting. another time it ran for 24 hours before i terminated the test myself as i thought 24 hrs was more than sufficient testing time.
-i have only run memtest for a couple of times a few hours each time and no errors have appeared.
-running 3dmark06 demo in loop mode and haven't experienced any reboots

i formatted and reinstalled windows xp from scratch this morning as a last attempt (i have read some posting about windows not properly recognizing the dual core cpu after an upgrade) but i still experienced these reboots. these restarts are unpredictable and very frustrating.

what could be wrong?
i'd rather not, but might i need to raise the vcore on the cpu?
with all the shipping involved, i would rather avoid an RMA, but is it possible that i have a faulty CPU (though wouldn't a bad CPU simply not POST at all...) that needs to be RMA-ed?

thanks.

update 10/7
i received my replacement cpu late thursday and popped it into my existing windows xp installation to stress test w/ orthos. i don't know whether it can be attributed to better skill/experience placing the heatsink/thermal paste or cooler weather, but temps while idle and under load were noticeably cooler (34-37C, 50-55C) than before. testing with orthos, the system rebooted after about an hour. i let it run again and went out for the night. when i checked back on friday afternoon, the logs indicated that the system had rebooted w/i an hour of my leaving.

on a whim, i set up a partition w/ vista (installed all peripheral drivers, but didn't bother to install my bluetooth driver/software yet) and was able to stress test for 13 hrs before i terminated the test myself. thinking everything was OK, i removed vista in favor of XP b/c i lose fullscreen video mirroring, my scanner doesn't have any vista drivers (vuescan also doesn't work either), and i get a significant performance hit in games. i created a new winxp partition and installed all of my drivers (all of my peripherals). the winxp system would reboot nearly every 30 minutes in normal use or within 10 minutes of running orthos. i got fed up and reluctantly reinstalled vista again (this time with software/drivers for my bluetooth dongle). the system rebooted twice while i was using bluetooth (skype via bluetooth headset). this leads me to believe that something may wrong with the drivers for my bluetooth dongle (or perhaps other peripherals or add-in cards).

another theory: in my research, i have come across many posts where ppl have upgraded from single core to dual core (or just X2 users) and experience BSODs, freezing, or reboots. they all seem to believe that the culprit is USB on NF4 chipset mobos and and its incompatibilities with dual core cpu's. nvidia has not addressed the issue in its drivers. a stopgap solution is to disable onboard USB functions and use a pci usb card.

to rule out hardware, i stress tested using a bootcd that included mprime (the linux equivalent of prime95). i ran two instances of mprime (forced each one to a specific core on the cpu); this ran without any errors for nearly 11 hours before i ended it myself. this leads me to believe that the hardware (mobo, cpu, ram, psu) itself is stable and capable or running the X2, it's a matter of the drivers or peripheral hardware incompatibilities. currently, i'm 4 hrs into orthos testing with a new winxp install. this time, i ONLY installed NF4 ethernet and smbus, AC97 sound, and forceware video drivers. all my USB peripherals, besides the mouse and keyboard, have been unplugged and i have not installed the drivers for the tv tuner card, which is still plugged into the mobo. i'll leave it for a few more hours and see what happens. i may start adding peripherals one at a time or add everything but the bluetooth dongle. or i may go back to my original single-core winxp installation and unplug all the USB peripherals and test. another option would be to test that winxp installation (via remote desktop) with onboard USB disabled.

in summary:
-cooler temps on new cpu
-replacement CPU still reboots in old WinXP installation (with all peripheral drivers installed)
-orthos ran for 13 hrs (before manual test termination) on a fresh Vista install (all peripherals installed except for bluetooth driver/software)
-computer rebooted twice while using (or immediately after using) skype with bluetooth headset in Vista
-was able to use ubuntu for 2 hrs (i was trying unsuccessfully to run mprime; i'm a linux noob) w/o reboot
-able to run 2 instances of mprime (using T Mod bootcd) for close to 11 hrs before manually terminating torture test.
-currently 4 hrs into orhtos testing on a fresh XP install (only NF4, video, and sound drivers installed; no usb peripherals plugged in)

plans
A. hopefully current test runs well for a few more hours:
-install tv tuner drivers and install usb peripherals + drivers one by one (this is going to be very tedious so i may just skip to plan B)
B. go back to my existing Winxp partition (the one i have been using w/ the single core) and:
1. unplug bluetooth dongle and remove bluetooth driver and stress test
2. unplug all usb devices (except keyboard/mouse) and stress test
3. disable onboard USB and stress test

hopeful solutions:
1. removing bluetooth dongle and drivers will resolve reboots
2. find out which is the incompatible peripheral/driver
2. buy usb pci card (if it is an onboard USB/NF4 chipset issue)
3. give up on the x2 and go back to single core

See any possible incompatibilities in my peripherals list below?
my usb peripherals:
connected directly to the mobo:
-logitech elite keyboard
-logitech G5 laser mouse
-samsung ml-1710 laser printer
-usb hub
connected to externally-powered 7-port Belkin usb hub:
-xbox 360 wireless receiver
-dazzle 8-in-1 flash card reader
-targus 25-in-1 flash card reader
-various flashdrives
-Iogear GBU321 bluetooth dongle (i also have a spare Belkin F8T013 bluetooth dongle that i have never used but may try instead)

add-in cards:
-hauppauge hvr1600 tuner card
-firewire card (automatically installed w/o drivers)

RESOLVED!
i received my Syba usb pci card thursday 10/11. i now have my usb hub connected to my computer via the pci usb port. connected to the hub are the 360 receiver, bluetooth dongle, 2 multi-format card readers, and the occasional flash drive. the devices connected directly to the motherboard usb are the keyboard, mouse, and laser printer. i have not run into any issues since i've set it up this way.

thanks again to all those who have participated in this thread.

 

[5t3v0]

Do you have the latest BIOS for the mobo?

 

[cputeq]

I would triple-check to make sure your CPU HSF is attached properly.
Also, try slightly slowing down your RAM timings on both sticks, and set them at the same timings.
The last thing I would try is disconnecting all your stuff except the essentials and do some testing (HD, CPU, RAM, K/B + M), though that may be a bit extreme.

It sounds to me, though, you're dealing with a bad CPU with an intermittent failure. Just because it's bad doesn't mean it won't boot.

 

[5t3v0]

Unlikely. CPUs dont generally fail intermittently. They usually work or the dont. OP, can you post the readings from your Health Status screen - to check all voltages & temps are correct. Also what is the BIOS version? You need to have version 1.1E for X2 support.

 

[Viditor]

Check that you have the new bios here.

Fixes include:
-Fixed Use USB storage large files size copying to HDD or USB HDD failed issue
-Support FX-60 VID for 1.35v CPU
-Fixup system will hang while Setup SATA RAID Enabled
-Support AMD FX-60 Dual Core processors
-Update BIOS for AMD new E4 CPU

 

[vexingv]

i do have the latest 1.1e bios; i upgraded immediately when i purchased the mobo.

my HSF is attached properly and its fan is connected (otherwise temps would be much higher than 60's C). at first i think i had placed too much arctic alumina (put all over the cpu heatspreader), but after looking at some instructions, i took it off and applied thermal paste only on the center where the core resides.

the memory timings are the same in the BIOS; the 3-3-3-8 crucial stick defaults to the slower 3-4-4-8 timing of the pqi dimm. what should i lower it to improve stability?

before going to sleep, i ran orthos again. looking at the speedfan log, it only ran for 1 hr and 20 minutes before rebooting.

screenshots:
speedfan_readings.jpg: this is the first 5 lines of speedfan readings and the last 5 lines of speedfan taken during the orthos test i ran over night. afaik there isn't any real deviations.

bios pc_health_status.jpg
dimm_timing.jpg
advanced_chipset_features.jpg (i've played around with the "spread spectrum" but that hasn't changed anything)

right now i'm about to put my old 3400+ back in and stress test that to make sure it is my mobo that has gone awry (but i can't remember if i actually stress tested the 3400+, then again i have been running perfectly fine for nearly a year). at this point, i think i'm about to RMA the CPU as i've only got 7 days for a return and 30 days to replace it. i'm really fed up with stress testing too as i've got other things i need to take care of...

 

[GeezerMan]

Is the CPU voltage at 1.3?
Try 1.35

 

[Markfw]

Power supply ? The new cpu takes more than the old. I am not farmiliar with that PSU, so I have no idea if its any good.

 

[bailw]

use this to read your temperature. http://www.thecoolest.zerobrains.com/CoreTemp/

It is definitely not the psu, pretty sure the system is unstable. Have you try to reinstall the OS?

 

[vexingv]

right now i'm 30 minutes into orthos stress testing on the old 3400+. the biggest difference that i can tell from the speedfan readings is that the core temperature tops out at 43 C, which is nearly 20 C less than the x2 4200+ core temp at load. is this temperature difference within norm? otherwise maybe i did get a faulty/poor cpu...

also here's a screenshot of the bios pc health page with the 3400+ installed. bios_pc_health_3400.jpg

 

[vexingv]

didn't refresh the page so i didn't there were more replied....

the x2 4200+ was using the set to "normal" Vcore setting in the bios. from the speedfan readings and cpuz that put it between 1.3 and 1.35 V. i don't OC so i don't see why i would need to increase it. also, i'd rather not increase the cpu voltage considering that it is already topping out at 63 C with the default voltage. in addition, last night i also installed the AMD XP drivers which included some quiet and cool function to adjust the cpu clock speed and voltages i think. didn;t matter since i still had reboots both before and after.

yes, i spent yesterday morning reinstalling windows and i still have reboots.

i'm going to stress test the 3400+ until the evening and see how it fares. and then try the x2 4200+ again.

bailw is coretemp better than speedfan? i went with speedfan since it was updated more recently than core temp.

 

[GeezerMan]

Your board may be undervolting. Software that reports temps and volts are often not accurate. I setup a 939 board awhile back that cpuz reported OK ram volts, but actually was undervolting.

 

[myocardia]

Set your vcore manually in the BIOS to 1.35v, the minimum for any X2 @ stock speed. If it's still restarting, it's because of your psu, which seems to be undervolting your +12v rail by a fairly large amount.

 

[vexingv]

Originally posted by: GeezerMan
Your board may be undervolting. Software that reports temps and volts are often not accurate. I setup a 939 board awhile back that cpuz reported OK ram volts, but actually was undervolting.

do you suggest i increase the voltage for RAM and CPU? or just one of them

if so how much should i increase it for each?

RAM voltage it is set at 2.63 V right now but can be changed to 2.55V, 2.63V, 2.71V, 2.79V, 2.87V, 2.95V, 3.03V or 3.11V.

the cpu voltage is set to default but can be raised in 25mV increments.

 

[vexingv]

Originally posted by: myocardia
Set your vcore manually in the BIOS to 1.35v, the minimum for any X2 @ stock speed. If it's still restarting, it's because of your psu, which seems to be undervolting your +12v rail by a fairly large amount.

maybe a dumb question, but what should the +12V rail be outputting? (12 V??) also i noticed that my PSU has dual +12V rails.

as i stated in the previous post my mobo sets cpu voltage by "normal" and then 25 mV increments thereafter. maybe i should start at the lowest and move up?

 

[myocardia]

Originally posted by: vexingv
maybe a dumb question, but what should the +12V rail be outputting? (12 V??)

As close as possible to 12v. 11.45 in the BIOS is quite low.

as i stated in the previous post my mobo sets cpu voltage by "normal" and then 25 mV increments thereafter. maybe i should start at the lowest and move up?

Just set it to 1.35v, then forget about it; that's where it's supposed to be.

 

[GeezerMan]

Yes, the 12V should be 12 volts. It's easy to check that with a multimeter attached to one of the molex connectors. Check that first

 

[vexingv]

Originally posted by: myocardia
Just set it to 1.35v, then forget about it; that's where it's supposed to be.

well there is no 1.35V setting; just "normal" and increments of 25mV. i went and increased by 25 mV. in one of my earlier pictures the bios was showing 1.37 V anyway. and cpu-z shows 1.36 most of the time but sporadically drops to 1.344 V.

Originally posted by: GeezerMan
Yes, the 12V should be 12 volts. It's easy to check that with a multimeter attached to one of the molex connectors. Check that first

i don't have a multimeter...

popped the x2 4200+ in again and running orthos now, we'll see how far i get.

also should the temperature of the two individual cores (as measured using coretemp) differ? i'm showing a 5 to 8 C difference between the two cores.

 

[vexingv]

i looked at the amdcompare site and the operating voltage is shown b/t 1.3 and 1.35 volts. also don't most processors these days throttle voltage or clock speed to run cooler when not being taxed?

 

[GeezerMan]

Temp difference between the cores is normal, I used to use the same power supply as yours, and speedfan would report about the same 12 volts as yours, too low, but a multimeter would report 11.97 volts. Of course, that was just a sensor problem in my case, your problem may be different. Get a decent multimeter, they are inexpensive.

 

[GeezerMan]

Your CPU will throttle back if you have cool n quiet enabled in the bios and in the power settings, and the driver installed

 

[myocardia]

Originally posted by: vexingv
well there is no 1.35V setting; just "normal" and increments of 25mV. i went and increased by 25 mV. in one of my earlier pictures the bios was showing 1.37 V anyway. and cpu-z shows 1.36 most of the time but sporadically drops to 1.344 V.

In that case, Normal should be fine, since software always shows slightly less than the cpu is actually receiving. That means that the reason you're getting these shutdowns is because your psu is overheating; they just shut down, with no warnings, when that happens.

 

[vexingv]

just returned from a reboot. orthos ran for about 40 minutes. this was with the 25mv increase in cpu voltage. doesn't really seem to be any difference. is there any way to definitively know it's the PSU? ultra has a minimum 3 yr warranty on most of their products so i should be able to get a replacement if needed. i'm still thinking of rma-ing the cpu though....

 

[vexingv]

ran orthos again @ stock voltage and this time it rebooted within 2 to 3 minutes of starting.

 

[myocardia]

There's nothing wrong with your cpu, your power supply just doesn't have the power to run what you're trying to use it to run. That's why those Ultra psu's are so cheap; they're very, very overrated in their output capabilities.