Hi folks. I am dealing with this problem for more than a year now. I bought a powerful machine in 2010 and wanted to share my massive computational power for science purposes.

Problem description:

When BOINC is running my system randomly freezes. I am not able to move the mouse, the actual music is running in a loop. Its not responding at all, I have to press "hardware reset".
Sounds like a hardware issue, right? Maybe it's the ram or cpu having a tiny error causing this after a couple of hours. So I was searching for this error over and over. Running prime 95 and memtest but I couldnt find any issue with ram or cpu. Then I thought the error might be located in the grapics card (used by boinc as well). Got a new one (warrenty case, said them it is causing my system to crash). The freezing still occures when boinc is running.
So, I can play the latest games for hours or run my system for weeks without any issues or freezes when boinc is not running. There is no hardware issure with my system.
My projects are "world community grid: clean energy" and "einstein at home". Is there any known issue at WCG causing these freezes? (I try running einstein only and see if there are any crashes but I think the clean energy project to be more meaningful.) I would love to run boinc all the time because my system is silent even on full workload. But I wont accept these freezes. They drive me nuts.
(I googled this for many hours and found that there might be an issue with the mainboard. But since the system is only freezing with boinc running, I dont think there is a problem with that now.)

Some Data:
BOINC Client 7.0.28 64 bit Windows
Windows 7 Home Premium 64 bit
Intel i7-930 4 x 2,8Ghz (no OC)
Nvidia GF 560 Ti 448 cuda cores
8Gb DDR3 7-7-7-21 Ram
Asus Rampage II Board

Sincerely,
Fire

There is no hardware issue with my system.

Actually, playing games doesn't stress your system as much as many projects running under BOINC can do. You only changed your videocard, but forgot the motherboard. A small micro-crack anywhere can have the effect you talk about. But I'll be honest, that is one of the most difficult problems to bring to light.

There's also this little thing called heat. Can also wreck a good system. Try running with Core Temp, perhaps that one or more of your cores overheat after so many hours.

---

The other component which you haven't mentioned, but which can be crucial in cases like this, is the PSU (Power Supply Unit).

It's not sufficient just to have one which is rated above the minimum required by your graphics card. It has to be in good working order (capacitors inside PSUs can bulge and fail, just as the ones on mothersboards sometimes fail).

And it has to be capable of delivering sufficient power at each of the voltages required by the components. For a GPU, the most important one is the "12 volt rail" (or rails). The GF 560 Ti 448 is specified to require 210 watts (18 amps @ 12v), and a minimum 550W system PSU.

Earlier versions of the 560Ti were notoriously susceptible to power supply problems, but yours is a later 'special edition' which should be free of those problems.

@Angeless: I know that games are not that stressful, but the error should occour even at 50-70% workload after some time. The crashes can occour even after seconds starting boinc. And I tried many different ways of stressing the system. My core temp is never exceeding 60°C, even after many hours and a normal room temperature. This should be totally fine for my i7.

@Richard: I considered my PSU to cause the problem and tested the real outputvoltages 3.3V, 5V and 12V and the accuracy of the voltage is perfectly fine close to the target voltages. I bought a high quality 700W PSU which should be (in theory, in the end I didnt buy a second GF card) able to supply 2 NVidia GF cards.

OK, you've checked the basics of the PSU - and that's about as far as my hardware knowledge goes.

But I was drawing on these two SETI posts (by a CUDA developer with extensive knowledge of the 560Ti) - you might like to check them out.

message 1240004
message 1240409

I made an interesting observation.
I totally kicked out the WCG project and the system is running with boinc stable since 2 days. This doesnot mean too much, the error is not happening periodically. A couple of days without freeze are possible but I start to think its directly related to these project work units (since I got the first freeze seconds after boinc was running WCG work units).
(Actually I got the nvidia 301 driver and the freezing occured with all (tested/installed) versions from 250 to 301...)
I might post this in the world community grid forum as well. Thanks so far!

I had a very similar system or screen freeze up, just after I installed and got using BOINC for SETI and Einstein, in February. The only solution to it was to reboot my PC. The problem was never really pinpointed; however, I solved it by just ceasing to use my graphics card for BOINC processing. The reason is, while poking around, checking cpu temps (I use Core Temp) and graphics stats (using the Nvidia Precision evga advanced graphics tuning software), I found that my graphics card was getting nearly maxed out by BOINC, showing GPU usage nearly 100%, with resultant very high GPU temps, etc. This bugged me; I felt I could lose the card if not my whole system if that continued, and with no built-in way to control it in BOINC.

So, no more GPU use (you cannot deselect GPU use in BOINC Manager, you have to go to your project account online). Since then, I have never had any freeze up problem at all, thank goodness. I could have tried TThrottle, but that was too much tuning for a newbie.

Note FYI: I run WU for Einstein and SETI. I tried Cosmology but it kept taking over, with High Priority WU's taking all the cpu. I also tried climateprediction.net, but it only gives out whopping giant WU's of 600+ hours long. Too much. They need to break these down, but get quite snooty at the suggestion.

----------------------------------------------------------------------------

David


Some Data:
BOINC Client 7.0.25
Gateway DX4850:
Windows 7 Home Premium 64 bit
Intel core i7-2600 cpu @ 3.4 GHz
GeForce GT 520 core clocked@881 MHz, 48 CUDA cores, 4 GB memory. Plan to upgrade this sometime.
8 GB system Ram
450W PSU
BOINC use at most 50% cpu time, 60% of processors.

---

I also tried climateprediction.net, but it only gives out whopping giant WU's of 600+ hours long. Too much. They need to break these down, but get quite snooty at the suggestion.

Hmmm, you think you're the first person to suggest that in CPDN's 8 years running under BOINC? As you aren't. And they do already have shorter models.

However, you cannot compare the running of a CPDN model to running a task of any other project. It's a climate prediction model that you are running. That can't be done in just an hour. The big program they have is over a million lines of code. Put things into perspective.

And even then, the initial time is an estimate. Your i7 will do approx. 8 days over a HADAM3P, not the 25 it initially says. Most probably even a lot less if you run only one of them. It has a deadline of about a year, so there shouldn't be a real problem here. If that's still too long for you, then just go do any other project that suits your needs.

---

Update: Next Freeze after 3 days running boinc Einstein@home only. Error is not related to WCG project. I cannot reproduce the error with any kind of stressing the system. It only occurs while boinc is running. I might try to run it without using my GPU (Temp is about 65°, totally fine even on full workload). That would be a pitty becaus I would loose something about a half of computational power.
Problem might be the Main Board but I will probably never know for sure.

I cannot reproduce the error with any kind of stressing the system.

What tools did you use to stress the CPU and GPU?

The highest load for CPU:
linpack -
used by LinX:
http://www.xtremesystems.org/forums/showthread.php?201670-LinX-A-simple-Linpack-interface/page34

... and IntelBurnTest:
http://www.xgamingstudio.com/forum/showthread.php?9-RELEASE-IntelBurnTest-v2-53


The highest load for GPU:
OCCT (can check for artifacts, on my GPU (AMD Radeon HD 6570) this tool (v3.1.0) gives the highest temp (72°C), @ Shader Complexity 7):
http://www.ocbase.com/

Furry Cube in GPU Caps Viewer (61°C for me):
http://www.ozone3d.net/gpu_caps_viewer/

FurMark (60°C for me):
http://www.ozone3d.net/benchmarks/fur/

(SETI apps heat less - to ~55°C)

All temps measured in the same month (May 2012) so similar ambient temp.

---

- ALF - "Find out what you don't do well ..... then don't do it!" :)

Just wanted to throw my random anecdote in here, in case it is helpful.

I had a similar problem where I was convinced BOINC was a problem, but even when I booted up without BOINC, I never quite fully solved my freezing problem, which sort of ruled BOINC out in my eyes,

Eventually, I remembered that my CPU and Memory were overclocked. I had done it almost 3 years ago, and prime95 was still fine, but maybe my hardware could no longer handle it, so I reset everything back to stock. If thats it, hey almost 3 years is a pretty good run for a stable overclock.

Boom, that was it. I'm now totally stable, no freezing, everything's fine. If you overclocked your system a while back, before blaming BOINC, try adjusting that first.

I had this exact same problem a few years back on a gaming laptop running BOINC, was always hard locking with the looping music etc.

In my case it turned out that the CPU was undervolted for some reason (never really knew why!). Used RMClock to boost the voltages up to the AMD specification and its never locked up since!

If you want to investigate this check what voltages RMClock says you are running at, and check AMD/Intel specs to see if it correct. If you try increasing voltages remember very small increments! Don't want to be blowing the CPU.


Edit: Link to RightMark CPU Clock Utility (RMClock) - http://cpu.rightmark.org/download.shtml

---

Although this is an older thread I'd like to make an addition to it.

A few weeks ago I suddenly had lockups on my Win7 PC. The first few times it happened while I was away from the keyboard and upon returning the screen was blank (no video output from the computer) and the monitor kept looking for a digital or analog signal.
A bit later I saw that the cursor began to "stutter" and moments later it would not move at all. A hard reset was all I could do. I was thinking about the last things that changed and suspected a program that recently was updated.
Since no one else reported problems on their forum I kept digging further, lowering boinc CPU use in % and number of threads. It looked better but still there were lockups, some even after 30 minutes after a restart. Workunits that failed were always Einstein never one of the other projects. After suspending the Einstein workunits I had no more problems.
The computer ran 24/7 for days until I re-activated Einstein and after a few hours again the PC grinded to a halt.
I deleted all Einstein WU's and suspended the project and so far no more problems.

Something in the Einstein WU must not like my setup. At least it's solved (for now). I'm back at 100% CPU :-)

I have been running SETI for years (10+) and without problem.

But lately, my system becomes unstable. It freezes randomly. After some 6 months of observation, I finally nailed the cause of the problem. It's the app I least suspected since I have been running it for years.

Before the freeze (mouse moves but no response from system, then finally stuck at one point), the display would go blank for a few seconds before the display driver recovers. But for the last 2 months, it just died.

Until a solution is available. I have to stop running SETI and the other app.

I am running Win 7/64 with i5 3330 on a Z77 board with Nvidia Gefore GTX 460 display card.

M Tan

I have also had this problem. It seems to exhibit itself by shutting down my usb ports - my keyboard screen reports No Connection. This started to occur after the last BOINC update.
Removing BOINC has removed the problem so the culprit is clearly the BOINC software.
Please resolve this; I have had more problems since using BOINC than I did using the original software for SETI.

I too have been running Boinc >10 years, with AMD processors, I'm now running the fx8350 since January and since then the pc crashes regularly while Boinc is installed. The PC's only stable when Boinc is uninstalled ANY clues, as the whole drive for me upgrading is more processing power, now using GPU's as well.
System :- MSI 990XA-GD55, AMD Vishera 8 core 4.0GHz & 16Gb 1600MHz Vengence, 2X ASUS HD6670 1Gb DDR5 GFX cards Win7 64Bit.

the pc crashes regularly

BOINC with one or more projects attached is a good stress-test and burn-in tool. Any PC not being able to withstand the force is just not a stable PC to begin with.

Which projects are you running?
Any errors in these projects?
Any (Windows) error messages when it crashes?
Blue screens? What do they say?
Have you got (mini-)(memory-)dumps (check C:\localdumps and c:\windows\minidump)?
Have you tested your memory, with either Windows its memory test or memtest86+?
Are you checking for heat? The FX series are quite a hot series by themselves, they require good cooling.
Are you all up-to-date with BIOS updates and (motherboard chipset)driver updates?

---

the pc crashes regularly

BOINC with one or more projects attached is a good stress-test and burn-in tool. Any PC not being able to withstand the force is just not a stable PC to begin with.

Which projects are you running?
Any errors in these projects?
Any (Windows) error messages when it crashes?
Blue screens? What do they say?
Have you got (mini-)(memory-)dumps (check C:\localdumps and c:\windows\minidump)?
Have you tested your memory, with either Windows its memory test or memtest86+?
Are you checking for heat? The FX series are quite a hot series by themselves, they require good cooling.
Are you all up-to-date with BIOS updates and (motherboard chipset)driver updates?

Can your PSU still supply all voltages within specification when under full load?

Hi Ageless,
I'm running SETI, Rosetta & Einstein on both PC's the other one is a Quad cord (Phenom Steady as a rock)
Only when it crashes, sometimes get a computation error
Yes with BSOD?, usually point to hardware issues.
Mini-dumps...LOADS...
Heat? - NO. I'll apply additional cooling
Updated Bios, tried several versions of Boinc, now trying the newly released 7.1.1
Updated Windows7 64bit, fresh installation, minimal installation just the basics to get pc running.
replaced mobo, new HD, new GFX card & cards, new memory, checkec modules individually, all seemm ok, sorry for all the typo's in a hurry to get this in b4 it crashes again...

Good point Richard, I have a gadget to monitor supplies, I'll check.