[mira_talk] Re: MIRA internal error
- From: Torben Nielsen <torben@xxxxxxxxxx>
- To: mira_talk@xxxxxxxxxxxxx
- Date: Sun, 9 Feb 2014 21:57:06 -1000
It’s a Xeon system with 256 GB of ECC RAM.
It’s the first time I have ever attempted digital normalization. I did it
because it’s metagenomic and there is some wildly varying coverage. I’ll turn
it off. However, then I have to rerun the whole thing and I’m like ten days
into this run :-( Oh well. Stuff happens.
If it matters, this was a run that I had to restart earlier. It crashed and I
had to rerun it with the “-r” switch. After the next one is done, I can try the
“-r” again. But the odds of a wrong bit in an ECC DIMM is really low. Precision
guided cosmic rays :-)
Thanks, Torben
On Feb 9, 2014, at 21:47, Bastien Chevreux <bach@xxxxxxxxxxxx> wrote:
> On 10 Feb 2014, at 5:19 , Torben Nielsen <torben@xxxxxxxxxx> wrote:
>> MIRA just aborted on me :-( Here is the end of the log. Please let me know
>> if you want more than this. I’m working with large datasets and they
>> generate huge logs so I can’t e-mail it.
>
> Does the computer you are working on have ECC-RAM? If not, then there’s a
> very high probability it’s a RAM problem. The reason I a saying this: MIRA
> complains about a DigiNorm multiplier of “0”, while shortly after, during the
> dumping as MAF of the presumedly problematic data, the very same read says it
> has a DigiNorm multiplier of “2”.
>
> This makes no sense at all … except if a bit flipped in RAM during the
> readout and temporary storage on stack of the DigiNorm value.
>
> This or a very, very weird bug in MIRA.
>
> Unrelated to this: I do not recommend using DigiNorm for genome assembly.
>
> B.
>
>
> --
> You have received this mail because you are subscribed to the mira_talk
> mailing list. For information on how to subscribe or unsubscribe, please
> visit http://www.chevreux.org/mira_mailinglists.html
--
You have received this mail because you are subscribed to the mira_talk mailing
list. For information on how to subscribe or unsubscribe, please visit
http://www.chevreux.org/mira_mailinglists.html
Other related posts:
