| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: BOINC Agent Support
Thread: Blue screens, SiSoft Sandra Hardware reports |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 5
|
|
| Author |
|
|
Dirk Gently
Senior Cruncher England Joined: Mar 1, 2005 Post Count: 153 Status: Offline Project Badges: 
|
I built a machine 6 months ago. It has been stable and crunching for WCG (both projects) and Climate prediction.
----------------------------------------Machine is based on Gigabyte GA-8195X Royal Motherboard with P4D 830 dual core processor at 3GHz and 1MB RAM. All WHQL drivers. Windows up to date. Around the time HP2 came along, I got the odd blue screen, often with a memory related cause such as "bad pool header". I recently started crunching for MalariaControl.net as well. This is in Beta phase. These are small WUs at the moment - they whiz through. But, if BOINC decides to crunch 2 MC WU's at the same time, the PC is VERY likely to crach - often within minutes. Crunching 1 WCG and one MC WU results in fewer crashes, but more frequent than if I only crunch WCG. Threre is no way to stop BOINC crunching 2 MC WU's, other than limiting ALL projects to 1 core. I set about some diagnostics. MEMTEST86 ran all night with no failures. I used the Lite edition of SiSoft Sandra, and the Environmental Monitor, reports seriously out of spec +5v rail of 6.85V. The -5V and -12V rails are all over the place. A text snapshot is shown below - it does not include the -12V rail for some reason, but the -5V is shown. The graphs for both are very erratic:- Analysing... Date & Time : 03 July 2006 13:33:20 Power / Aux Temperature : 64.0°C (Min 59.0°C; Avg 62.9°C; Max 65.0°C) Chassis Fan Speed : 2766rpm (Min 2679rpm; Avg 2720rpm; Max 2766rpm) CPU Voltage : 1.31V (Min 1.22V; Avg 1.32V; Max 1.34V) Aux Voltage : 2.08V (Min 2.08V; Avg 2.09V; Max 2.10V) +3.3V Voltage : 3.25V (Min 3.23V; Avg 3.24V; Max 3.25V) +5V Voltage : 6.85V (Min 6.85V; Avg 6.85V; Max 6.85V) +12V Voltage : 11.54V (Min 11.42V; Avg 11.54V; Max 11.98V) -5V Voltage : 0.07V (Min -4.65V; Avg -0.45V; Max 0.63V) Standby Voltage : 4.95V (Min 2.12V; Avg 3.81V; Max 5.16V) Battery Voltage : 4.08V (Min 4.08V; Avg 4.08V; Max 4.08V) CPU Core Power : 94W (Min 80W; Avg 95W; Max 98W) It is very possible that Gigabyte have not implemented all of the voltage measurements that tyhe chipset is capable of - and the results may be nonsense. Does anyone have any experience of SANDRA voltage monitor in general, and the Gigabyte motherboard in particular? |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Did you check the +5V with a DMM? If it agrees with Sandra then you know. If it doesn't agree then.... hmmm.... try a 2nd DMM?
----------------------------------------[Edit 1 times, last edit by Former Member at Jul 4, 2006 7:51:38 AM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The trivial explanation is that you seem to have an unstable power supply.
Nothing in the WCG code can cause a STOP error, but the additional load on the CPU may plausibly push the power supply over the edge. MEMTEST isn't capable of detecting all error conditions, so don't rule out other hardware issues. That temperature looks a bit toasty. Probably not dangerously so, but a bit warmer than I would be happy with. |
||
|
|
Dirk Gently
Senior Cruncher England Joined: Mar 1, 2005 Post Count: 153 Status: Offline Project Badges:
|
Thanks for the replies.
----------------------------------------I have checked with a multimeter and a calibrator (UKAS CAL). The 5V rail measured via a HD power connector is 5.02V and does not vary when CPU load changes from 0 to 100%. The +12v rail reads as 12.01V (Sandra shows 11.54V !!). The others are a little more difficult to get at! I did not measure before because it was mainly SANDRA I was unsure about. I ran it on a freinds machine and got a similar result. There was a thread on a help forum showing exactly the same problem with same motherboard. They got normal readings with a multimeter too! Unfortunately, they did not post the outcome! It is possible that the motherboard reinverts the 5V rail anyway - it has a fancy additional regulator board. They also may not have bothered to wire up the voltage sensing lines and SANDRA is reporting nonsense. I am waiting for an answer from Gigabyte on this one. I suppose my first easiest course is to try another PSU! I dont fancy prodding around the motherboard itself! The processor temp - that is with a replacement heatpipe cooler, not INTEL's offering! You should have seen it before! I have cleaned the heatsink and it is down to 56 deg. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
It doesn't sound like much is wrong with your power supply. It's either Sandra not reading the SMBus properly or the SMBus itself or a bad regulator on the board.
----------------------------------------You definitely don't want to probe pins that small when the board is powered up but if you can find a good solder jockey and solder wires to the pins then you can power up and read V off the wires. I've done it but I have the right tools and steady hands. Don't try it unless you have a good iron, extremely small tip, good binocular magnifier, solder sucker and lots of experience soldering. [Edit 1 times, last edit by Former Member at Jul 4, 2006 1:56:37 PM] |
||
|
|
|