You could also try putting the DB into RCSI.
As there is the potential for write-skew anomalies you would need to review all SPs and do a server side trace of any critical parts of the system. There would also be a greater load on tempdb.
Good point, thanks. I did try that some months back (in anticipation of NOLOCK being removed), the slow-down was dreadful. Not sure why that might be but my instinct is the amount of RBAR, rather than set-based SQL statements, in the 3rd party code ... worth having another go, maybe some things have been improved since then, but it would most likely help with the BLOCKING side.
If RCSI is causing a slow-down, it is worth double checking that the number of files in tempdb equals the number of cores available to the instance. I am sure you already have that one covered.