[Tinyos-help] more questions about powerTOSSIM

Giovanni Vanini giovanni.vanini at gmail.com
Tue Mar 29 20:49:05 PST 2005


Hi!
I have some questions related to the cpu cycles counting on
powerTOSSIM, which is surely the most tricky part of the simulator to
understand:

First of all, as noticed by Samitinjoy Pal (there is a post  in the
middle of March), it seems that there is no actual simulation of the
changing of the cpu state, while no component is calling the command
async command void PowerState.cpuState(uint8_t sm) in PowerStateM.  By
default the CPU consumes as it was always in IDLE. Even when it is
computed and added the power in active mode, the IDLE consumption
seems to be the same (overestimation of consumption). Can you confirm?
I cannot test it by myself because I am not able to make the the
cilly.asm.exe program working under cygwin (the error "program too big
to fit the memory" affects more that one cygwin machines  I've tried
on. Any idea?). I'm setting up a linux box right now to check...
Even after reading the paper and presentation of powerTOSSIM at SENSYS04,
I cannot see how can happen (it is claimed to be one of the main
source of powerTOSSIM inaccuracy) that a line of the intermediate C
code is belonging ("may be compiled") to more than one basic block, as
soon as the identification of the BB is done directly on the C code
itself. Can someone try to explain it more in detail?
Thank you for any help

Giovanni


More information about the Tinyos-help mailing list