??? 03/16/07 12:08 Read: times |
#135101 - optimization = lost control Responding to: ???'s previous message |
I HATE optimizers. When you see most of the "wonderful" stuff optimizers do, the first question is "why does the compiler not do so?".
Debugging optimized code is a true hassle, you set a breakpoint and because the optimizer decided that "this can be shared" it stop for some unrelated reason. OK, on a very tight memory budget there are a few things like common exit routines that only the linker (and thus an optimizer) can do, but there is a tendency throughout the industry (Keil do you hear me) to put improvements that the compiler could do even easier and maybe better into the optimizer, just to make the optimizer "look good". Who in his right mind would be more impressed if a better result is because of an optimizer than. If it was a better compiler, I would be impressed. Erik |
Topic | Author | Date |
akkkk! stupid compilers, something to watch | 01/01/70 00:00 | |
ehm, isn't it the user? | 01/01/70 00:00 | |
Well it might be phooey;- but its pretty common | 01/01/70 00:00 | |
have you considered using a timer, perhaps... :-) | 01/01/70 00:00 | |
yep but... | 01/01/70 00:00 | |
Sorry - not this time, Jez | 01/01/70 00:00 | |
if it is THAT short... | 01/01/70 00:00 | |
Maybe not... | 01/01/70 00:00 | |
Beware. | 01/01/70 00:00 | |
Well, any compiler with half a brain does that. | 01/01/70 00:00 | |
volatile | 01/01/70 00:00 | |
use util/delay.h and read the lib documents | 01/01/70 00:00 | |
optimization = lost control | 01/01/70 00:00 | |
well that is the first mistake | 01/01/70 00:00 | |
First mistake is using a HLL | 01/01/70 00:00 | |
Optimise & kill code | 01/01/70 00:00 | |
optimizers and NASA | 01/01/70 00:00 | |
There are many levels in Keil | 01/01/70 00:00 | |
not always a good idea | 01/01/70 00:00 |