In order to run THERMO_PW, you need an input for pw.x, a file called thermo_ control, and an input for ph.x (which must be called ph_control) if required by the task. These files must be in your working directory. The input of pw.x can have any name and is given as input to the THERMO_PW code. It is better not to specify an outdir directory in the ph.x input. Specifying an outdir directory is not forbidden, but for some tasks thermo_pw.x might add a geometry number to outdir and the outdir written in the ph.x input must be consistent.
A typical command for running THERMO_PW is:
mpirun -n np thermo_pw.x -ni ni ... < input_pw > output_thermo_pw
where np is the number of processors and ni is the number of images. The dots indicate the other QUANTUM ESPRESSO parallelization options that you can find in its manual.
Note that it is very easy to waste resources using too many images. Unused images wait for the working images to complete their tasks wasting cpu-time in an endless loop. Some options do not use the image feature, so you have to know how the calculation is divided and the number of images must not be larger than the number of tasks (below I give this number for each option). If you have doubts on this point use one image (ni=1).
The outputs of the THERMO_PW code are one or more postscript or pdf files with plots of the material properties. THERMO_PW produces also files with the data of the plot and scripts for the gnuplot program. Usually, the user does not need to modify these files, but they allow the improvement of the figures when needed. The plot of the Brillouin zone (BZ) is made with the help of the asymptote code. Thermo_pw produces a script for the asymptote code and can also run it to produce the pdf file of the BZ.