[codes-ross-users] codes simple net problems

Mubarak, Misbah mmubarak at anl.gov
Fri Sep 30 08:43:15 CDT 2016


Hi Lei,

Here is some feedback regarding your questions:

- Yes, for transferring large data through simple-net you will need to use
the message_size parameter.

In terms of simulation performance, I had a question for John i.e. do we
break down messages into packets at the model-net layer for simple-net
model as well? If so, the simulation may be getting slower due to a use of
very small packet size (by default it is 512 bytes in the configuration
file). 
 
- The bandwidth is in Megabytes per seconds. Since we are using all lower
cases, it gets confused by Mbps. To make this clear, I have added the
bandwidth units in the example config file.

Thanks,
Misbah 

On 9/30/16, 5:23 AM, "codes-ross-users-bounces at lists.mcs.anl.gov on behalf
of 于雷(Yu Lei)" <codes-ross-users-bounces at lists.mcs.anl.gov on behalf of
yulei at buaa.edu.cn> wrote:

>Hi John,
>
>we have two questions about simple-net model.
>
>1/If we want to transfer "big data" between two servers, like 100GB, we
>only need to set the parameter message_size to 100*1024*1024*1024 in the
>method model_net_event_return model_net_event_annotated(
>         int net_id,
>         char const * annotation,
>         char const * category,
>         tw_lpid final_dest_lp,
>         uint64_t message_size,
>         tw_stime offset,
>         int remote_event_size,
>         void const * remote_event,
>         int self_event_size,
>         void const * self_event,
>         tw_lp *sender);
>
>Am I right? Because when we change the parameter message_size to
>100*1024*1024*102, the simulation becomes very slow.
>
>2/In the configuration file of simple-net, net_bw_mbps="4000" means the
>bandwidth is 4000*1024*1024 bit/s and   4000*1024*1024/8 byte/s?
>
>Thanks,
>
>
>Lei
>
>
>
>_______________________________________________
>codes-ross-users mailing list
>codes-ross-users at lists.mcs.anl.gov
>https://lists.mcs.anl.gov/mailman/listinfo/codes-ross-users



More information about the codes-ross-users mailing list