Like someone already suggested, CHARMM is licenced. So it would be illegal for anyone to share it here.
If you want the same functionality, you should look at NAMD (http://www.ks.uiuc.edu/Development/Download/download.cgi?PackageName=NAMD). It is a free simulation engine designed to work with CHARMM based force fields. So it will produce the same result as CHARMM does.
Once you register in the link provided in a previous message (which is the official CHARMM site), you should get an email with a link to download the software once you register. The installation is straight forward and the free version gives you a good start on using this software for simulations (it has most of the capabilities form the licensed version, as stated on their page).
Try checking you email spam folder, maybe you got the download link there,
the link provided by Amin should work. This version of CHARMM does not contain the DOMDEC code for parallel simulations on HPC systems. From personal experience I recommend to use CHARMM as modelling tool for setting up the simulation system, do patching etc. CHARMM is extremly powerful and has implemented the most complete set of modern sampling techniques. However, it is by far not the fastest computation software for MD simulations, although things are changing.
For the final production runs on a HPC cluster, the easiest way would be to use NAMD which reads PSF and PDB files as well as the CHARMM force field files. NAMD scales pretty well for larger systems (>100 000 atoms) and has pretty good performance on CPUs as well as on GPUs.
If you are running smaller systems, GROMACS is the better choice but requires several additional steps to convert the CHARMM PDB files in GROMACS readable file formats.