[Csnd] CSound vs C++
| Date | 2013-08-03 04:52 |
| From | lppier |
| Subject | [Csnd] CSound vs C++ |
Hi, Not sure if this is an appropriate place to ask, but I'll ask anyway. For an iOS/Android synth app, I was wondering if it would be more efficient CPU-wise for a synth app to be coded in pure c++ rather than implementing it in CSound? Or would the CPU utilisation be roughly the same for the same kind of synth/effects? I know CSound runs from a command-line, and the opcodes are coded internally in C. Has anyone compared these two cases side-by-side? (similar app, CSound vs C++) Thank you. -- View this message in context: http://csound.1045644.n5.nabble.com/CSound-vs-C-tp5726262.html Sent from the Csound - General mailing list archive at Nabble.com. |
| Date | 2013-08-03 12:24 |
| From | Michael Gogins |
| Subject | Re: [Csnd] CSound vs C++ |
Some years ago I did such a test. Pure c++ can be faster than csound but not by much. The speedup from computing samples in blocks rather than one at a time is much greater, so if you choose c++ be sure to work with blocks of samples, similar to ksmps = 10 or 100. LuaJIT is capable of running faster than orc code and can be used in Csound. In fact my paper on the Lua opcodes compared speeds also. If you do use c++ write your piece itself in c++ and include your opcode code in your piece's code and just call the Csound API to register the opcodes, I think that would work, if not I will change the API to enable it. On Aug 2, 2013 11:52 PM, "lppier" <madstrum@gmail.com> wrote:
Hi, |
| Date | 2013-08-04 06:43 |
| From | Pier |
| Subject | Re: [Csnd] CSound vs C++ |
| Thanks Michael for the information. So the cpu limitations I see are more a function of the tablet's processing limitations, rather than the overhead of running CSound I guess. On 3 Aug, 2013, at 7:24 PM, Michael Gogins <michael.gogins@gmail.com> wrote:
|