| ??? 07/17/09 16:27 Read: times |
#167511 - 2kB of code space is quite a lot for evaluation Responding to: ???'s previous message |
Richard Erlacher said:
It does serve to point out that 2KB is not sufficient to give the KEIL toolset an adequate trial. Something costing as much as the KEIL suite costs should be thoroughly exercised to the satisfaction of the prospective purchaser before purchasing it. Once they have your money, they never give it back! Given the preaches given by the owners of the Keil tools, maybe they don't have to give back any money. How many Keil compiler owners have you seen switching to another compiler because they thought the other compiler was better? 2kB max code size is not a problem. You can fit a huge function into 2kB. The only problem is that you will have to view the generated assembler code in the debugger. Keil don't want you to be able to compile functions one-at-a-time and glue the separately compiled functions into your own assembler framework. I understand that view, since the majority of code isn't time-critical and it would be so simple to use another compiler for 90% of the project and get the Keil evaluation version to compile the most critical functions. But turn it another way. Once upon a time, 2kB of code space was considered a lot. A lot of home-built computers had between 512 byte and 2kB. Why shouldn't 2kB be enough to evaluate if a compiler will produce good code or not? Or is it the 8GB Richard who have suddenly changed side and considers any evaluation version that can't produce code for the maximum possible address range of a processor bad? So if the evaluation version of the compiler can't do 24-bit banked code for a 8051 or supports 4GB of codespace for an ARM it isn't meaningful to investigate? |
