When the correlator host computer boots, the three low level programs (in Figure 26.3 are started. The correlator itself is usually started before this computer is booted, but in principle the correlator can be brought up at any time before higher level programs are executed. Next, `acq' is started, and it connects to two low level programs, the data collector, and the interrupt server. As long as `acq' is alive it takes data from the data collector, time stamps it using the information from the interrupt server (for more details on time-stamping see below) and then places the time-stamped data in a shared memory. `acq' does not care whether any such programs exist to use this shared memory.
At this point, the correlator configurator program (corr-config), is run to setup the correlator in the required mode. Ideally, this should be done when `acq30' is instantiated, but as of now this is not recommended. `acq30' and the peripheral programs (`collect', `dassrv', `record') can then be started. At this point the entire data acquisition chain is up. The `record' program can be started anytime after `collect' has been started. Any number of `record' programs could run at a time. Similarly, any number of monitoring and online display programs can run simultaneously to follow the progress of the observation.
A large number of small tools are developed to look at the raw data as placed in the shared memory of `acq' on the host computer. These programs provide invaluable tools to debug the correlator problems, and also to make consistency checks at run time. These programs do not interfere with the normal observations, therefore, any number of them could be run at a time.