ECM Errors !!!

[STE\/E]

I've had 2 Error's on the same box that look the same to me, any reason for these Error's ??? I don't think the Box has finished a Wu yet since I put it back on the Project ... Thanks

http://www.rechenkraft.net/yoyo/result. ... id=5163731
http://www.rechenkraft.net/yoyo/result. ... id=5163972

A Third one erred out on the same Box ... Running @ Stock CPU Speed

http://www.rechenkraft.net/yoyo/result. ... id=5164062

The Wu's seem to get almost to the end of their run & then error our for some reason ... The Box did finish 1 Wu so far ...

PS: Not absolutely sure but I may have found the Problem. The CPU Cooler Fan kept stopping because of a Faulty Wire Connection & the CPU was running about 20c Hotter than it should be running. I fixed the Connection & will let it run some more Wu's to see if that was really the Problem ...

[Erkan Yilmaz]

Hello PoorBoy,

from your results I see that it works already, or ?
info: same WU on a linux, amd, quad succeeded.

Happy ecm crunching

[STE\/E]

Fixing the CPU Cooler Fan seemed to fix the Error'son that Box, haven't had one since except for the one that was already in progress for over 5 hr's ...

But on another Box I've had several already too, the only other Box I've had any on, will check that one in a Bit to see if I can figure out what's wrong with it or you can tell me something why the Wu Errored ...

http://www.rechenkraft.net/yoyo/result. ... id=5164465
http://www.rechenkraft.net/yoyo/result. ... id=5164175

[Ananas]

If only the ecm_mp are affected and if you try to run 4 ECMs concurrently, it might be a RAM shortage.

[STE\/E]

I've quit running any YoYo Wu's on the Box for now, it has other issues that I haven't been able to figure out yet. The Box started out having Error's on the MP Wu's & then when I fixed the CPU Fan it seemed to run okay for a little over 1/2 a day and then it just started to lock up. I switched the Box to Edge Wu's which use a lot of Memory too but the Box still locked up. So I switched the Box to NFS Wu's which use a lot of Ram too & it run's fine.

I even tried doubling the Memory to 4GB running the YoYo & Edge Wu's but the Box still Locked Up, but it seems to be running fine with 2GB running the NFS Wu's which use just as much Memory as the YoYo & Edge Wu's Do. So it must just be something to do with the Wu's themselves that the Box just don't like ...

[Beyond]

I've probably run as much ecm as anyone so maybe can be of some help. Most of the WUs are no problem on any machine. The MP WUs however take a lot of ram. This box has undoubtedly run more ecm than any other on the project:

http://www.rechenkraft.net/yoyo/show_ho ... stid=33460

It was originally an X2, then a Phenom 9600, now a Phenom II 940 (same MB). Also have another box running ecm with an Athlon II 620 (previously also X2, then Phenom 9600). Both were originally XP and are now Win 7. The reason for listing all this is because they've run stably under all these conditions after I figured out how to run the MP WUs consistently. First, the MP WUs fail sooner than other types (and other projects) on machines that are OCed too far, so be SURE that your OC is solid if you choose to go that route. Also be SURE that your ram is good and running within parameters. Secondly, notice that I have set the swap space to 8GB (for machines with 4GB ram) on my machines running ecm. Until I increased the swap space to well above Windows default I had some WU failures. Note that MP WUs will sometimes stop for short periods waiting for ram, and then continue. Perhaps on MBs or CPUs that don't handle this swapping well they may be more susceptible to failure, but on my AMD / Biostar 790GX systems they run well.

[STE\/E]

None of my Box's except for the lone i7 I have are Overclocked, they used to be by about 50% (Quad Cores) but once I started adding the Capable GPU's for Milkyway & Collatz & GPUGrid I backed the CPU's off to their Stock Speeds because they became insignificant when compared to the GPU's I'm running. That plus it was a small effort to decrease the overall Power consumption in the Systems so the GPU's wouldn't be lacking for Power.

Anyway I have the Box running okay now, I swapped out 4GB's of Memory from another Box & put the 2GB's of Memory that were in it into the Box the 4GB's came out of. So right now both Box's are running the ecm_op's with no problems so far. For some reason that Box needs the full 4GB's of Memory while the other Box run's just fine with only the 2GB's in it. I ordered about $800 worth of Memory today to get all my Box's up to the 4GB Level, about half of them are already and I should have had the other half done already but was procrastinating on doing it.

Their all XP Pro 64-Bit Box's so they can & will use the full 4GB's of Memory if installed ...

[rroonnaalldd]

Host 30845 gets error 195 on many ecm_nr units. How to resolve?

[yoyo]

Host 30845 gets error 195 on many ecm_nr units. How to resolve?

Put more memory into this PC.

These ecm_nr units need 800MB each. Your computer has only 1GB. So running two of them in parallel on this computer will lead to problems. Normaly Boinc should handle this properly but this seems not to work properly.

yoyo

[rroonnaalldd]

Okay, thanks for the info. It would be helpfull to see the memory-req's of ecm_nr-units on the project-frontpage or better for all type of units on the app-page.

[technik007.cz]

Host 30845 gets error 195 on many ecm_nr units. How to resolve?

Put more memory into this PC.

These ecm_nr units need 800MB each. Your computer has only 1GB. So running two of them in parallel on this computer will lead to problems. Normaly Boinc should handle this properly but this seems not to work properly.

yoyo

I cannot believe it !!!
16 years later and I am experiencing the same problem on many of my arm boards. I have many of ecm_nr workunits which has failed. The other ones like ecm_cn are fine.

They fail with this error:

app exit status: 0xf
called boinc_finish(195)

One of funniest error which I found was only 1 workunit on Quad-Core waiting for free memory where nothing else than boinc was running.

[yoyo]

You haven't posted any link to such workunits. So I can't help and can't say what is wrong.
If it is a memory problem, than BOINC can't handle properly the memory requirements which are given to each workunit.