| ??? 07/17/09 15:48 Read: times |
#167504 - It's not what they're taught to consider Responding to: ???'s previous message |
Per Westermark said:
Richard Erlacher said:
You can avoid the big float penalty without giving up 'C' ... but you have to be willing to do what's necessary.
How many 'C'-coders would be willing? If you were interested, you would be able to find huge amounts of forum posts about the use of fixed-point arithmetic in C. But you are not interested, for the simple reason that you have already decided that a developer using C is stupid. It wasn't more than a couple of hours since I posted to the Keil forum about using fixed point for some evaluations. I'm not uninterested in the issue, but I'm sufficiently so to allow me to avoid the KEIL forum. I've gotten one lie too many from the KEIL tech-droids, and at least one bit of misinformation too many as well. I do follow the SDCC user forum though not rabidly. What concerns me is not that there are 'C' coders who know that one can do some arithmetic in fixed-point rather than float with satisfactory results, but that see evidence that far too few actually do it because they aren't taught about conserving code space, hence, they don't even consider it. After all, they're taught to program on machines with 8GB of SDRAM and floating point hardware. That's fine on a PC, but in the embedded controller world, it's not so safe an assumption that they have that available. RE |
