And before that HandelC but not sure about C++ aspect as I don't recall any class support.
You have some learning curve in front of you which ever direction you go in as you will be touching real HW. Unless you are in a .edu environment you will be paying a pretty penny for any C/C++ to HW generation.
You could just as easily take a look at any Verilog/VHDL text on DSP design and see if it wouldn't make more sense to go direct to HW that way. You only need to learn a very small subset of either language to get the results you might want, more if you use the language fully for verification, less if you use C. More if you want best performance possible.
Upside is your tools will be almost free for X/A for either V/V language (upto modest size designs that is) and your results will be ultimately much better than any C/C++ path.
Downside is you may be looking at multiple language representations of the same code. Thats not necessarily hard but it takes time to learn how to think HW and code appropriately while also wearing your C SW hat.
One aproach I use is to use a cycle C form that essentially looks like Verilog line for line until you look close enough. Cycle C is obviously free for me and runs on any C like env, but doesn't take much time either to fix the syntax back into proper Verilog. It is limited in many ways (flattish hierarchies) but works well for DSP/cpu projetcs for me and test vectors used in the C model are easy to convert to Verilog.
Another poster might suggest Confluence, or SystemC to look at too..
Once you choose the V/V HDL route you also get the benefit of a far larger NG base to call upon, here, verilog/vhdl and of course dsp NGs etc. And in particular a few experts around here and quite a bit of free source code from opencores and also X/A websites. If you use a proprietary C++HDL you are confined to a much smaller source of examples, free goodies and advice!
If that tickles your fancy then Palnitkar, Doug Smith and a few others are very good HDL authors.
regards
johnjakson at usa dot com