| ??? 02/03/06 13:06 Read: times |
#109058 - The reason... Responding to: ???'s previous message |
Vignesh Prasad said:
..generic '51 has less RAM or something else technically? How about a port to higher and denser derivatives of '51? The specific Show-Stopper is the lack of Stack "higher and denser" derivates may add loads of XDATA, but they do not change the hardware Stack limitation. This is the reason why Keil et al go to all that trouble of messing about with Overlaying - that effectively does at compile-time what a Stack does at run-time (it doesn't quite to it all - which is why it gives particular problems with things like function pointers). Thus getting GCC onto an 8051 is not just another "port" to a new instruction set - it would require a fundamental re-think of the entire function-calling process! As Jez said, you can implement a "software" stack in XDATA for parameter passing, etc - but that incurs a horrendous performance & code-size penalty. [personal opinion] GCC appears to me to be very much targetted at "large" processors; I think making it appropriate to "small" processors like the 8051 would require so many tweaks that it would become unrecognisable as GCC I think SDCC addresses these specific issues by focussing closely on the requirements of "small" processors - and not trying to be a universal, one-compiler-fits-all, solution. [/personal opinion] |
| Topic | Author | Date |
| GNU gcc compiler for 8052 | 01/01/70 00:00 | |
| No! | 01/01/70 00:00 | |
| sdcc | 01/01/70 00:00 | |
| And never will be | 01/01/70 00:00 | |
| Hmm well | 01/01/70 00:00 | |
| Stack Size | 01/01/70 00:00 | |
| Is it so because.. | 01/01/70 00:00 | |
| processors supported by gcc | 01/01/70 00:00 | |
| The reason... | 01/01/70 00:00 | |
| bigstack | 01/01/70 00:00 | |
| Probably not | 01/01/70 00:00 | |
| Bigger Stack | 01/01/70 00:00 | |
| how much is enough? | 01/01/70 00:00 | |
| Who? | 01/01/70 00:00 | |
| but who??? | 01/01/70 00:00 | |
| SDCC | 01/01/70 00:00 | |
| No. | 01/01/70 00:00 | |
| only preprocessor | 01/01/70 00:00 | |
| Successful? | 01/01/70 00:00 | |
| routinely | 01/01/70 00:00 | |
| The AVR tinys have no ram | 01/01/70 00:00 | |
| Not me | 01/01/70 00:00 | |
| Wheel re-invention | 01/01/70 00:00 | |
| I suppose | 01/01/70 00:00 | |
| sdcc - where to start | 01/01/70 00:00 | |
| Thanks | 01/01/70 00:00 | |
| Sorry | 01/01/70 00:00 | |
| No stack required | 01/01/70 00:00 | |
| reentrancy | 01/01/70 00:00 | |
| The right tool for the job | 01/01/70 00:00 | |
| RE: The right tool for the job | 01/01/70 00:00 | |
| Good point | 01/01/70 00:00 | |
| ... but who cares, anyway? | 01/01/70 00:00 | |
| Sad, but true! | 01/01/70 00:00 | |
| not convincing | 01/01/70 00:00 | |
| have been designed to fit C | 01/01/70 00:00 | |
| it IS relevant | 01/01/70 00:00 | |
| not rteally | 01/01/70 00:00 | |
| choices | 01/01/70 00:00 | |
| ...and Pascal | 01/01/70 00:00 | |
| If you are planning to make 1.000.000 un | 01/01/70 00:00 | |
| Choices | 01/01/70 00:00 | |
| naah... | 01/01/70 00:00 | |
| Yup | 01/01/70 00:00 | |
| oh $8 | 01/01/70 00:00 | |
| I don't buy that one | 01/01/70 00:00 | |
you have to work with what you get | 01/01/70 00:00 | |
| Even assembler needs justification | 01/01/70 00:00 | |
| you DID use the %! | 01/01/70 00:00 | |
| as to above | 01/01/70 00:00 | |
| tradeoffs | 01/01/70 00:00 | |
| I'd buy Keil if I had to make 2 units | 01/01/70 00:00 | |
| my original point was... | 01/01/70 00:00 | |
| Dan, you missed a detail :) | 01/01/70 00:00 | |
| In good company | 01/01/70 00:00 | |
| deadlines | 01/01/70 00:00 | |
| assumptions | 01/01/70 00:00 | |
| could not say it better | 01/01/70 00:00 | |
| Nothing new there, then! | 01/01/70 00:00 | |
| oh well, more words | 01/01/70 00:00 | |
| Normal | 01/01/70 00:00 | |
| have a look | 01/01/70 00:00 | |
| yes, but most want their tool | 01/01/70 00:00 | |
| I dont know where to reply! | 01/01/70 00:00 | |
| as so often before | 01/01/70 00:00 | |
| Seen it so many times | 01/01/70 00:00 |