Can you give to PMP a try? The next version will be awesome! Vasi
On May 9, 12:55 pm, Sebastien Lelong <[email protected]> wrote: > Hi guys, > > Minix is a big lib, minix can't compile without -no-variable-reuse. Since > compiler can't handle this amount of code (it takes hours to compile, maybe > days, actually I didn't wait for complete compilation, it's just too > long...), I tried to reuse variable myself, within the code. Following is a > blinded example with fat32.jal library, another big one. > > First, results before optimization: > > jallib compile -no-variable-reuse sample/18f4550_fat32_sd_card.jal > jal 2.4o (compiled May 8 2011) > generating p-code > 6309 tokens, 357042 chars; 8703 lines; 16 files > generating PIC code pass 1 > generating PIC code pass 2 > writing result > Code area: 19364 of 32768 used (bytes) > Data area: 1542 of 1952 used > Software stack available: 410 bytes > Hardware stack depth 10 of 31 > 0 errors, 0 warnings > > Now, let's dig one procedure, "randomly" picked: fat32_file_entry_data. It > declares, within its body, the following variables (I just read it, > extracting "var ..."). > > var byte sector_after_entry_location > var word sector_entry_number > var byte char_pos = 0 > var byte number_of_entries > var byte step3 > var word sector_entry_number2 = sector_entry_number * 32 > var byte long_name_step = FAT32_FILE_NAME_SIZE - 13 > var byte sector_step = 0 > var byte first_char > var byte step3 > var byte step > var byte _cluster_address[4] at fat32_file_cluster_address > var byte _file_size[4] at fat32_file_size > var byte _cluster_address[4] at fat32_file_cluster_address > > These variables are declared either at body's root, or within "for", "if" > block. Some can't be reuse, but, *maybe*, some can actually be reused, > because their content is only used within a limited, predefined timeline. > Let's first move all these declaration at the top of body. Compiling reveals > the following duplicates: step3, _cluster_address, and _file_size. Removing > duplicates gives the following output: > > Code area: 19424 of 32768 used (bytes) > Data area: 1540 of 1952 used > > Why program memory grew that way ?? I can't explain... Let's continue. step > and step3 are using within for loops, not at the same time. We can remove > step3 and use step instead: > > Code area: 19530 of 32768 used (bytes) > Data area: 1539 of 1952 used > > 19530 bytes !!! Why ? One byte RAM saved... > (is that about banking ?) > > Oh well... I rewrote that mail 10 times as I actually discovered the results > :) I'm stopping here, and starting to understand why my Minix lib is so big. > I indeed used a lot of "alias" in order to try to reuse variables > "manually", and save RAM, and it does but the amazingly growing number of > program bytes is quite scary... > > Cheers, > Seb -- You received this message because you are subscribed to the Google Groups "jallib" group. To post to this group, send email to [email protected]. To unsubscribe from this group, send email to [email protected]. For more options, visit this group at http://groups.google.com/group/jallib?hl=en.
