Not really a Unix question as such :o, but what advantages or disadvantages are there between using 32bit or 64bit applications on a T5220 running Solaris 10? What about mixing them e.g. 64 bit app using 32 bit libraries or vice versa?
The main advantage of 64 bit is obviously the larger addressing space which allows a process to access more than 4 GB (eg. memory mapped file). Other advantages include faster operations when dealing with 64 bit data (eg: cryptography) and extra registers for FP.
Disadvantages includes larger binaries (typically 10 to 30%) and memory usage that made general purpose applications sometimes running slightly slower when 64 bit, the reason why most userland in Solaris is provided as 32 bit.
You cannot mix 64 bit applications with 32 bit libraries but 32 bit binaries run very well on top of a 64 bit kernel. There is no 32 bit Solaris kernel anyway for a T5220.
Have a look there for more details:
Solaris 64-bit Developer's Guide - Sun Microsystems