[Cbc] Consultant about Coin CBC

[YU Chenmin (RBCD/TEF)]

Dear cbc team:

I am a data scientist in Bosch, now I am working with cbc to solve a MILP problem. I have several questions regarding the running time of cbc need your support.


1.     When I run cbc solver on my computer(CPU 8cores + 32memory), I monitored my cpu performance and found the usage is about 35%. It seems that only 2 cores are working in parallel.

I also tried to set the thread by solver.options['threads'] =?(I am using pyomo to create to model), no matter how many threads I set, the running time has no significant change.



So my first question is does cbc only support 2 cores computation in parallel?



2.     I am using both cbc2.7.5 and cbc2.10.3, I found in some scenarios, cbc2.10.3 is even slower than the old version.

e.g. for the attached model, the running time for 2.7.5 is 220s, for 2.10.3 is more than 10000s. Could you help to check why and how to avoid that?


Thanks and looking forward to receiving your reply.


Best regards,

Chenmin YU
RBCD/MFE

Tel. +86 510 8110-7322 | Mobil +86 018862117764

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/cbc/attachments/20190826/4323eb2c/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Modle.lp
Type: application/octet-stream
Size: 2114239 bytes
Desc: Modle.lp
URL: <http://list.coin-or.org/pipermail/cbc/attachments/20190826/4323eb2c/attachment-0001.obj>