kv83 wrote:To be honest, I think the best thing is to copy existing systems.
Only if those systems have at least remotely similar capabilities. Since the flash ROM and a HDD are completely different devices, similar file systems are out of the question. There has been a lengthy talk about it already, why don't you check it out?
I still can't find anything in that discussion where it says that it ain't possible. To be honest, I think it could work out nice. Even if it isn't a HDD, so what? You can use a FAT also for formating SD-cards, Floppy's and memory keys and it works quite fine afaik. Of course you may need some other techniques for writing or reading, but the general idea would be possible imo
Well, for one there are 64k big quanta we have to consider any time we need to set a single bit in flash memory, i. e. clusters couldn't be written independently. A direct consequence is that to modify a file you need to write it back to a separate place while marking the previous version for deletion, unless you want to garbage collect after virtually every file operation. Also, having every file waste a cluster would be totally unacceptable: just imagine all TIOS variables having their size rounded up to the nearest multiple of 256 bytes—just the 26 reals would take up 6.5k! No way.
I agree with cobb, this isn't an sd card. Hell this is a lot crappier than the cheapest sdcard. 64k Sectors is the biggest limiting factor. Really even though what TI came up with for its flash can be really annoying, for all practicallity sake, it is pretty good. I don't see a reason why a 3rd party OS would really want to escape from it so far.
[quote="CoBB"]Well, for one there are 64k big quanta we have to consider any time we need to set a single bit in flash memory, i. e. clusters couldn't be written independently. A direct consequence is that to modify a file you need to write it back to a separate place while marking the previous version for deletion, unless you want to garbage collect after virtually every file operation. Also, having every file waste a cluster would be totally unacceptable: just imagine all TIOS variables having their size rounded up to the nearest multiple of 256 bytesâ€â€Â
would it be possible to utilize the extra RAM pages on the higher end calcs? make it so that of the two currently loaded RAM pages, on is always the page you execute out of, and the other one can be switched around to retrieve more data.
