If you use system time by checking it in every bloc, its more efficient to count blocs since it uses less CPU instructions.
You can use sample rate and bloc size to work out how many blocs make up the time interval, then just set a counter to the value and subtract 1 every bloc until the value gets to zero, then do the action and reset the bloc counter.
An example of counting blocs with interval approximately 0.1 seconds:
void ModuleClass::onInitModule (MasterInfo* pMasterInfo, ModuleInfo* pModuleInfo)
{
double interval_secs = 0.1; // Use double data type since sdkGetSampleRate returns double
int numblocs = (sdkGetSampleRate() / (double)sdkGetBlocSize()) * interval_secs; // "int" numblocs to truncate result
counter = numblocs;
}
void ModuleClass::onProcess(void)
{
if (--counter == 0) {
doAction();
counter = numblocs;
}
}
This is not always accurate due to rounding off to get an exact number of blocs. If you need it to be accurate on average over time, you can keep track of the truncated fraction and use it to adjust the bloc count.
If exact timing is required (perhaps part way through a bloc), a timer might be required (thread timer / windows timer, etc). If you use a callback timer the callback might be in a different thread so you might need thread protection (mutexes, etc.)
Hope that helps.Statistics: Posted by sm_jamieson — 17 Dec 2017, 21:40
]]>