| ??? 07/24/09 16:10 Read: times |
#167828 - Hardly! Responding to: ???'s previous message |
Andy Neil said:
Richard Erlacher said:
'C' isn't quicker for us, here, because it's got to work with hardware we already have in hand. The fact that the hardware is already in hand is, surely, irrelevant? Hardly ... If I have ASM software that has been proven to work correctly, I'm certainly not going to rewrite it in 'C' just to please someone else. 'C' doesn't make for smaller and faster code, after all. The point is, presumably, that you already have the software "drivers" for that hardware? If you had no pre-existing software, then it's highly likely 'C' would be quicker. Timing, which is frequently ultra-critical in the domain within which I work, is much easier to predict when code is written in ASM. There have been frequent comments alluding to the unpredictability of timing of 'C'. Generally, I don't have time to experiment with 'C' compilers to see what they do. I know what the assembler does. I know what the code does. I know what the MCU does. How's 'C' going to help? Nearly any function I need can be encapsulated as a callable or a paste-able macro, and the assembler will do the pasting. If I need n copies of an instruction sequence, the macro assembler will allow me to specify that with a single line, and the result will be exactly what I want. Why would I want to use 'C'? RE |
