𝕏

Posts by NATE1

1) Message boards : Number crunching : Strange cpu behaviour (Message 144)
Posted 20 Apr 2015 by NATE1
Post:
Please give us the device number. (via URL)


http://denis.usj.es/denisathome/show_host_detail.php?hostid=1413 (Intel Dual Core)
http://denis.usj.es/denisathome/show_host_detail.php?hostid=285 (AMD Six Core)

Edit: it's a FPU problem? AMD Bulldozer core has shared FPU..

that and

http://www.extremetech.com/computing/100583-analyzing-bulldozers-scaling-single-thread-performance

a bit dated but still valid....
2) Message boards : Number crunching : Extremely slow download? (Message 82)
Posted 9 Apr 2015 by NATE1
Post:
Registered in past 24 hours 2161

care to guess how many of those 2161 new users are spam bots???????????
3) Message boards : Number crunching : failed downloads (Message 80)
Posted 9 Apr 2015 by NATE1
Post:
check your files and directory on the boinc project server

may have to put a longer backoff on the client, too many request/connections at once...also your allocated amount of disk space/memory may be low, due to DB bottleneck, in order to allow the server time to create the files in download directory.
once the http errors start the server is over loaded trying to updated the DB
and created the next taskname_x in the run. x being 0, 1, 2, 3, etc..



<core_client_version>7.4.42</core_client_version>
<![CDATA[
<message>
WU download error: couldn't get input files:
<file_xfer_error>
<file_name>300_sim_2636.conf</file_name>
<error_code>-224 (permanent HTTP error)</error_code>
<error_message>permanent HTTP error</error_message>
</file_xfer_error>

</message>
]]>
4) Message boards : Number crunching : failed downloads (Message 73)
Posted 9 Apr 2015 by NATE1
Post:
yes getting a lot of them now

is the server ip address static or dynamic

boinc project server ip address have to be static..

thanks..
5) Message boards : Number crunching : Extremely slow download? (Message 72)
Posted 9 Apr 2015 by NATE1
Post:
Hello everyone,
I recived very often, more than 300 times, the message that the download of the tasks failed.

Greetings

Harald


don't think their IS dept was ready for that big of a load on their servers all at once
Users #
With credit 105
With recent credit 96
Registered in past 24 hours 2157
Computers #
With credit 210
With recent credit 180
Registered in past 24 hours 263
Current GigaFLOPS 37.72
6) Message boards : Number crunching : errors Too many errors (may have bug) (Message 70)
Posted 9 Apr 2015 by NATE1
Post:
not started by deadline - canceled
timed out - no responce
error while downloading

edit: also aborted by user

should never be counted as a true error.
it just causes tasks to be wasted and then you will have to send them back to the Que manually (if you really need the data)
(on the flip side, the host just ran a task for nothing)

fyi, thanks.......


name CRLP2011EPI_001_30_2334
application Carro-Rodriguez-Laguna-Pueyo Epicardial Model (Carro et al. 2011) for human ventricular cells
created 8 Apr 2015, 12:20:06 UTC
minimum quorum 2
initial replication 2
max # of error/total/success tasks 3, 10, 6
errors Too many errors (may have bug)
Task
click for details Computer Sent Time reported
or deadline
explain Status Run time
(sec) CPU time
(sec) Credit Application
4669 9 8 Apr 2015, 17:11:21 UTC 9 Apr 2015, 5:10:34 UTC Not started by deadline - canceled 0.00 0.00 --- Carro-Rodriguez-Laguna-Pueyo Epicardial Model (Carro et al. 2011) for human ventricular cells v1.00
4670 9 8 Apr 2015, 17:11:21 UTC 9 Apr 2015, 5:10:34 UTC Not started by deadline - canceled 0.00 0.00 --- Carro-Rodriguez-Laguna-Pueyo Epicardial Model (Carro et al. 2011) for human ventricular cells v1.00
14557 16 9 Apr 2015, 5:11:01 UTC 9 Apr 2015, 17:11:50 UTC Completed, can't validate 170.38 162.51 0.00 Carro-Rodriguez-Laguna-Pueyo Epicardial Model (Carro et al. 2011) for human ventricular cells v1.00
14558 128 9 Apr 2015, 5:11:31 UTC 9 Apr 2015, 17:11:31 UTC Timed out - no response 0.00 0.00 --- Carro-Rodriguez-Laguna-Pueyo Epicardial Model (Carro et al. 2011) for human ventricular cells v1.00
32435 49 9 Apr 2015, 19:46:51 UTC 9 Apr 2015, 19:52:31 UTC Error while downloading 0.00 0.00 --- Carro-Rodriguez-Laguna-Pueyo Epicardial Model (Carro et al. 2011) for human ventricular cells v1.00
32441 260 9 Apr 2015, 19:45:48 UTC 9 Apr 2015, 19:47:04 UTC Error while downloading 0.00 0.00 --- Carro-Rodriguez-Laguna-Pueyo Epicardial Model (Carro et al. 2011) for human ventricular cells v1.00
7) Message boards : Cafe : The inevitable ATA thread (Message 68)
Posted 9 Apr 2015 by NATE1
Post:
ATA?


The first rule of ATA is: You do not talk about ATA.
The second rule of ATA is: You do not talk about ATA.

:P


The third rule of ATA is: See the first and second rule! :O
8) Message boards : Number crunching : max number of task in progress (Message 59)
Posted 9 Apr 2015 by NATE1
Post:
ok, look like it was one of those one off hic ups things.
has the right number of task per core.

(I'm having a priest come over and do an exorcism on the computers because they must be possessed) :)

thanks.....
9) Message boards : Number crunching : max number of task in progress (Message 53)
Posted 9 Apr 2015 by NATE1
Post:
reb, I know
what I am trying to figure out is
did I just run into a case of the boinc client counting the asic not only as coproc but also as cores
so I need to know what this projects set as x for the max number of denis task * ncpu in progress

thanks


Hello,

Now we setup the max_wus_in_progress to 3.

I hope this information can be helpful for you. If you have any other question please ask to us!

best regards, Joel.


thanks, it does

I'm going to let it run for now the way it is, any not started by deadline will get auto aborted.
at 3 per ncpu I should have max of 9 on one host and 6 on the other,
if I end up with (3 cores + 14 miners) * 3 task max = 51 and
(2 cores + 14 miners) * 3 task max = 48 tasks then yes there is a bug.
stay tuned, same bat-time, same bat-channel
I'll let you know......
10) Message boards : Number crunching : max number of task in progress (Message 49)
Posted 9 Apr 2015 by NATE1
Post:
reb, I know
what I am trying to figure out is
did I just run into a case of the boinc client counting the asic not only as coproc but also as cores
so I need to know what this projects set as x for the max number of denis task * ncpu in progress

thanks
11) Message boards : Number crunching : max number of task in progress (Message 44)
Posted 9 Apr 2015 by NATE1
Post:
Hello,

We are working to solve it. And yes, the aborted task are resend to complete them, thank you for your attention.


in config.xml add

<max_wus_in_progress>x</max_wus_in_progress>

x is the value for WUs * NCPUs


Thank you! we have change it one hour ago! we hope you will see that change. We are a small team and these first days are plenty of work!


thanks,
what did you set x to?
I need to know, there maybe a problem with the count if you are running BU with ASIC attached to the same host got 2 host each with 14 ASIC on them running BU and this project, one has over 900 DENIS tasks on it and the other has over 700 DENIS tasks. may have to go back to the boinc dev for a fix. if it is counting the ASIC as cpus (cores)

anybody else seeing this?

thanks..
12) Message boards : Number crunching : max number of task in progress (Message 26)
Posted 8 Apr 2015 by NATE1
Post:
just aborted over 900 off 3 host.
hope they got the server set up correctly to resend them to new homes to get processed...

your welcome...
13) Message boards : Number crunching : max number of task in progress (Message 24)
Posted 8 Apr 2015 by NATE1
Post:
need to set a max number of tasks in progress per host.
just found one host with well over 600 task on it, 4 cores each task takes 30 minutes to run 12 hour deadline and...
looks like you will be getting a lot of them back, starting at about 08:00 utc
on 9 April 2015....

going on that you can not set the flop bound(?) on these tasks yet.
then max in progress would be a big help..
thanks..
14) Message boards : Number crunching : Request: Increase deadlines (Message 11)
Posted 8 Apr 2015 by NATE1
Post:
Absolutely, I'm glad to be here, and glad to be able to give feedback. :)
So far, my feedback is: Deadlines are too short, causing my GPUs and bitcoin miner coprocessors to go idle, because BOINC schedules your tasks as priority to run them in "deadline-risk mode / earliest-deadline-first mode".

If you don't really need them in 12 hours time, then I recommend something like 5 days or 2 weeks.

Side note: For me, they complete in 2 minutes. That's potentially a lot of overhead in transferring data back and forth, and in looking at the task results lists. If possible, it might be better to have the tasks take around 4-48 hours each, with checkpoints.


looks like you got that one, the run times have gone from 2 minutes to about 30
fyi...
15) Message boards : Cafe : boincstats challenge anyone (Message 8)
Posted 8 Apr 2015 by NATE1
Post:
any team want to do a challenge on this project?

just thinking nice way to run in the servers and get a lot of work done

what, too soon?