BRP3cuda Error 2
5 Jan 2011 17:15:24 UTC
Topic 195554
(moderation:
On one of my computers all BRP 1.04 WUs error out with:
... The system cannot find the file specified. (0x2) - exit code 2 (0x2) ... [15:19:03][4676][ERROR] Input file on command line ../../projects/einstein.phys.uwm.edu/PM0005_00171_206.binary doesn't agree with input file ../../projects/einstein.phys.uwm.edu/PM0005_00171_205.binary from checkpoint header. [15:19:03][4676][ERROR] Demodulation failed (error: 2)!
for example this one
Runtimes vary between a couple of minutes and more than an hour.
Windows 7 64 bit system, GTX 295 + 9600 GSO, driver 265.90. WUs error out on all GPUs.
BRP3cuda Error 2
)
Hi!
Thanks for reporting this. Looks 'interesting' (as in "I've never seen this before"). Each WU consists of 4 bundled tasks. If a computation is interrupted, the next time it is started it will pick on from where it stopped, using data from a so called checkpoint file. But in your case, the app tries to resume from a checkpoint file after starting a new sub-task. There should not be a checkpoint file at that point.
So I reported this to the developers, but as this is not happening for other users (or your other PCs), there must be something specific with this PC that triggers this problem. Is there anything unusual about that BOINC installation or the the PC itself?? A different anti-virus software than on your other PCs maybe?
Thanks for any feedback,
HBE
I think I've seen this on my
)
I think I've seen this on my machine as well (see the following WU: http://einsteinathome.org/task/212864764)
This happened right after a power outage. I turn the PC back on and the WU failed :(
This only happened once though, not like the OPs with every task failing, so it might not be relevant.
Indeed, the same thing.
)
Indeed, the same thing. Interesting that this happened after a power outage, might be useful to pinpoint what's going on here.
Thanks!
HBE
RE: Is there anything
)
Thanks for your reply!
The computer has been up and running for 2-3 weeks without a problem doing some CPU demanding work for me. No power outages, not even heavy HDD traffic going on when those problems happened...
I just enabled new Einstein tasks today and BOINC downloaded the new BRP app and WUs. It ran previous Einstein CUDA WUs without a problem. The BOINC client is the default version from the BOINC webpage.
I've installed "process tamer" to increase process priority to "above normal" and set BOINC's CPU usage to 90%.
Antivirus scanner is Microsoft Security Essentials which also runs on my other PCs.
Think i produced that error
)
Think i produced that error to here too: http://einsteinathome.org/task/214542259
System had a bluescreen. I am still trying to pinpoint why it has them. Its an older 8800GT that i did put back into service in my dads pc. Seems to get a lot of bluescreens with einstein, did get around one per day while crunching Primegrid.
Right now i am letting it run with underclocked memory and gpu to see if it is the cards fault. CPU is also overclocked, but so far was never an issue.
It seems i have to correct
)
It seems i have to correct myself. The Overclocking on that cute Pentium Dual E2180 from 2 to 3Ghz was the fault for the bluescreens and in return the Error 2.
The added 8800GT in the PCI-E slot (system did run before only with the integrated GFX) and added load from Einstein did put a lot stress on the system it seems. So not really Einsteins fault. I am trying to fix it now with more FSB voltage and a little more juice to the memory controller (yes, i know what i am dooing, overclocking is a hobby of mine, i know the risks *g*).
Thanks for letting us know.
)
Thanks for letting us know. Indeed, the E@H apps are quite good for stress testing :-)
CU
HB
And now it seems to crunch
)
And now it seems to crunch away smoothly. For the odd chance that someone runs into a similar Problem: Just raising the Vcore did the trick. Added GFX Card pushed the voltages just a lil bit down and the overclocked cpu got unstable cause of this. Should have figured that out earlier...
Else the cuda units run very nice on my 8800GT! Only double the time of some GTX 580 that got to validate my units and faster than most GTX 260. I also got to like the only 60-70% GPU load, means i can let it run on the card while pc is in use. If for future versions it would be possible to set a value of GPU usage it would be really awesome or have it be smart and use less GPU when a Video is playing etc. and use up to 95% while just browsing.
Anyhow, i gladly donate this old friend of mine (the 8800GT) to science!