• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

I am (still) puzzled as to why my P3 Tualatin is so unstable

E

Electrode

Diamond Member
I've had a Pentium 3-S 1.26 GHz CPU for awhile now, and ever since I first set it up, it has been strangely unstable.

Right now it is at 1376 MHz, and seems stable enough. BIOS reports it's running at 44 degrees idle. Vcore is 1.55 volts (I set it to 1.60). It boots just fine, and left to itself it'll run SETI@home forever. But once I start running apps it starts going to hell. Mozilla will usually start, but sometimes it will segfault. UT2K3 crashes on start, sometimes before it changes the resolution, sometimes after. glxgears runs for a few seconds, and then locks up the system requiring a power cycle. Quake 3 is most interesting: it starts, shows the intro movie, and gets to the menu. But once there, I see a lot of flickering and various visual artifacts, and then it crashes out to the desktop after a few seconds.

As the title says, I am puzzled. The CPU doesn't seem to be overheating, the video card isn't overclocked at all, the RAM isn't cheap no-name stuff, there's a good ammount of airflow through the case (thanks to a 60mm hair dryer fan at the back of the case), and the memory is set to the lowest CAS settings available, despite being rated much higher.

Specs:

CPU: Intel Pentium 3-S Tualatin 1.26 GHz
RAM: 2x256mb Crucial PC133
Motherboard: Soyo TISU
PSU: Enhance 350W
Video card: Visiontek Geforce2 Ultra
Sound card: Creative Audigy (it didn't mind a 190 MHz FSB, 145 MHz shouldn't be a problem...)
Networking: 802.11b PCMCIA card in a PCMCIA-to-PCI adapter, Linksys Ethernet card
OS: Linux

EDIT: Forgot to mention that this happens at rated speed too, although it lasts quite a bit longer before stuff starts going wrong (UT2K3 will run for 2 matches or so, but hang during the third)
 
The BIOS didn't volunteer much info. I'll see if I can get lm_sensors (the Linux equivilant to MBM) to work tomorrow, but for now, here's what the BIOS said:

Vcore: 1.55V
VTT: 1.23V
3.3v: 3.24V
+12V: 13.12V
VBat: 2.88V
 
Alright, I just installed lm_sensors and ran it. This is the first board it has ever worked correctly on. 🙂 More importantly, I have found the cause of the stability problems. Under full load (SETI running, just got done compiling lm_sensors), the PSU is really falling short:

VCore 1: +1.20 V (min = +1.53 V, max = +1.87 V) ALARM (undervolting and overclocking don't mix. hell, even rated speed isn't possible with this kind of voltage!)
+3.3V: +6.48 V (min = +2.96 V, max = +3.60 V) ALARM
+5V: +2.02 V (min = +4.50 V, max = +5.48 V) ALARM 🙂Q)
+12V: +13.04 V (min = +11.36 V, max = +13.80 V)
-12V: -5.19 V (min = -15.86 V, max = -13.40 V) ALARM
-5V: -2.04 V (min = -10.13 V, max = -9.44 V) ALARM
Stdby: +5.61 V (min = +4.50 V, max = +5.48 V) ALARM
VBat: +2.96 V
fan1: 6958 RPM (min = 3000 RPM, div = 2)
fan2: 7848 RPM (min = 3000 RPM, div = 2)
fan3: 0 RPM (min = 3000 RPM, div = 2) ALARM
Temp1/MB: +22°C (min = +20°C, max = +60°C)
Temp2/CPU: +41°C (min = +20°C, max = +60°C)
Temp3: +33°C (min = +20°C, max = +60°C)

Thanks for your advice bgeh! Looks like I need a new PSU. At least I know my brand new $184 CPU isn't at fault. 🙂
 
I always knew that Intel had some pretty broad tolerances for their voltages but that's incredible!

Nice catch!

My first guess was gonna be ram, then voltage.
 
An update:

My new Enhance 350W PSU (30A +5 rail, 220W/330W max loads) came yesterday, and I just installed it. It seems to have helped, but only a little. What I did:

1. Install new PSU
2. Go into BIOS and set FSB to 150 MHz, Vcore to 1.500V
3. Boot
4. Notice that, as the kernel is starting (hasn't even gotten to init yet) there are 'stray pixels' underneath some characters
5. Reset, since the kernel paniced about halfway through early initialization
6. Back into BIOS to set FSB to 145 MHz, leaving Vcore at 1.500V
7. Boot
8. Start SETI@home, insert lm_sensors modules and run sensors

Sensors shows some improvement, but none where it counts:

electrode@collective:~$ sensors
it87-i2c-0-2d
Adapter: SMBus I801 adapter at 5000
Algorithm: Non-I2C SMBus adapter
VCore 1: +1.45 V (min = +1.53 V, max = +1.87 V) ALARM
VCore 2: +1.23 V (min = +2.25 V, max = +2.75 V) ALARM
+3.3V: +6.58 V (min = +2.96 V, max = +3.60 V) ALARM
+5V: +2.02 V (min = +4.50 V, max = +5.48 V) ALARM
+12V: +12.96 V (min = +11.36 V, max = +13.80 V)
-12V: -6.80 V (min = -15.86 V, max = -13.40 V) ALARM
-5V: -5.20 V (min = -10.13 V, max = -9.44 V) ALARM
Stdby: +5.64 V (min = +4.50 V, max = +5.48 V) ALARM
VBat: +2.89 V
fan1: 6887 RPM (min = 3000 RPM, div = 2)
fan2: 7758 RPM (min = 3000 RPM, div = 2)
fan3: 0 RPM (min = 3000 RPM, div = 2) ALARM
Temp1/MB: +21°C (min = +20°C, max = +60°C)
Temp2/CPU: +46°C (min = +20°C, max = +60°C)
Temp3: +35°C (min = +20°C, max = +60°C)
Click to expand...

Note especially that the 3.3V rail is getting walloped, and the 5V rail still sits at 2.02V. Now, either lm_sensors is lying to me (entirely possible) or there's something wrong with the motherboard. I'll see if I can dig out my multimeter later today and see exactly what's going on.

EDIT: Just checked a molex connecter with the multimeter and it showed almost exactly 5 volts.

So the PSU is supplying 5 volts, but the motherboard's sensor only reports 2.02 volts. It's still unstable under heavy loads such as UT2K3. Might the motherboard be at fault?
 
Ordinarily, I would have to remain silent with shock, but I noticed the Quake3 problem.

I've seen this before.

The flickering polygons usually occur on GF2 cards when the AGP bus is either running out of spec, or the AGP voltage is far too low.
I've seen this occur on a Geforce2 Pro, when AGP is set to 2x, with AGP bus running at > 86 mhz,with older BX boards (newer ones would be a bit more tolerant). Quake3 would basically start showing flickering triangles, especially on the models and gun, then the entire video card would just shut "off", with the PC locked up with repeating sounds, and the monitor in power save mode. GF4 cards are far more tolerant of this, although Sidebanding must be disabled.

Yours, however, seems to just crash. but the artifact effects are similar.

There is no reason why the 3.3v and 5v should be so low. Since you exchanged PSU's, the blame has to be on the motherboard.

You said that you are using sensors, instead of the BIOS, to measure the temps? The sensors are by the motherboard connector right?

I'd try a new motherboard...or RMA that one....
 
You must log in or register to reply here.
Back
Top