OK. Using a 128K for a r/w buffer still allows an additional 380+K. Or make it 256K buffer which still allows for an 256K of additional storage. Still better than what's in there now.
I'm an embedded engineer (for 40 years) and it particularly troubles me that so much flash memory is not utilized. It's always treated as a precious resource. If anything, it could be used as a virtual flash file system if not used for the uPython OS.
Moving forward, embedded MCU's are only going to have increased memory and performance. The new PyBoard in beta has an M7 core with a large memory footprint, eventually moving to 1M SRAM in place of relatively slow Flash. It would be a shame if fast SRAM were not fully utilized.
SD card vs flash drive, speed?
Re: SD card vs flash drive, speed?
The STM32F405 (which the pyboard is based on) has 64K of RAM in the CCM and 128K of regular RAM. Currently, the 64K CCM RAM is used for the flash buffer and the 128K is used for stack and heap.
It's probably possible to reduce the heap/stack down to 64K in order to support 128K flash buffer, but there currently isn't any code to do that.
Contributions for improvements are welcome.
There are always tradeoffs to be made, and what makes one person happy, makes somebody else unhappy.bradstew wrote: ↑Wed Jan 09, 2019 2:50 pmMoving forward, embedded MCU's are only going to have increased memory and performance. The new PyBoard in beta has an M7 core with a large memory footprint, eventually moving to 1M SRAM in place of relatively slow Flash. It would be a shame if fast SRAM were not fully utilized.
Some of the other processors like the STM32L4 series have much smaller flash blocks so more of the flash can be used for filesystem.
I believe that the new M7 based board will have external SPI based flash which will allow for much larger code and filesystems.