TI has a native double float support and low cost emulators. So it's very attractive.
So far our firm used ADSP-218x (about 1000-2000 piece in year), ADSP-2191 (about 50-100 piece in year). For Russia it's a big purchase amount.
ADSP-21992 (about 25-50 piece in year) processors. Now we want to develop a new high-performance control system. And we saw that AD has a poor support and invalid promotion of processor production that TI:
1. Lack of support of 40-bit float point feature in C/С++ VDSP++.
I udnerstand that it's to support I must to declare 40-bit width of the internal memory and turn off RND32=0. So I get 40-bit float point support. That to use 40-bit float point feature in C/С++ VDSP++ I learn Sharc's manual and search in forums. The application from AD is absent. It's very bad.
Besides that it's a problem of saving in external memory 40-bit float point.
TI has a native support of the double float point.
2. Lack the low cost emulators with BTC for customer's target boards
The low cost emulators with BTC can become the standalone debug agent. It's very good idea to use in the low cost emulators the Blackfin's processors.
TI has low cost emulators which the price is lower than AD's.
Do AD plan to a new sale programm "piggy-back promotion ": VDSP++ and ADZS-HPUSB-ICE ? As it was December 2008 - January 2009?