| ??? 09/16/05 20:30 Read: times |
#101185 - why the need for a pardon Responding to: ???'s previous message |
1) why the need for a pardon, I hope your airways are in excellent shape.
2) but it is possible to measure the speed of a system. It is not always that easy to determine worst case execution times I can not count the cases where the "measured speed" happened to be just not enough about once a day or less frequent. This was much more rampant when 12 clocks 12MHz was the norm. 3) but if you don't know how to do that, then optimising the code is not a substitute. No one argrued that "optimising the code is a substitute", I would never suggest that (see post above). It is amusing to see you argue that optimization is not the way when you in your previous post just stated: "I think you should always make it work right, and then make it work fast if neccessary" which translates to "whack the code together in the large model, then optimize it". 4) A properly engineered system will work under all circumstances, even when the worst case execution times hit you. totally agree; howevet I can think of no case where using the LARGE model represent "A properly engineered system". The point is that "development shortcuts" alway end up biting you in a large muscle. Erik |
| Topic | Author | Date |
| compiling in SDCC on 89c52 | 01/01/70 00:00 | |
| Move some variables | 01/01/70 00:00 | |
| all in xdata | 01/01/70 00:00 | |
| go have a cup of coffee | 01/01/70 00:00 | |
| If that isn't enough.. | 01/01/70 00:00 | |
| a disastrous approach | 01/01/70 00:00 | |
| Indeed | 01/01/70 00:00 | |
| we agree, I hope, on both points | 01/01/70 00:00 | |
| Pardon me for breathing | 01/01/70 00:00 | |
| why the need for a pardon | 01/01/70 00:00 | |
| Really? | 01/01/70 00:00 | |
| With Craig on this one | 01/01/70 00:00 | |
the full story | 01/01/70 00:00 | |
| how much | 01/01/70 00:00 |