Free RAM in 0.9.1.b1

[bmarkus]

Updated LoPy to 0.9.1.b1 firmware released today. Now free RAM is only 71,152 bytes (it was 191,088 in previous release). Also, every execution of gc.mem_free() is decreasing it by 64 bytes :(

Is it the expected value?

[LoneTech]

@jeremy I did; there was a different problem (two of the lopys got corrupted versions in flash, one was readonly), which I've worked around as described in the how to solve the firmware update failures thread. I did once upload without naming it properly which explains the fourth mystery copy.

[daniel]

BTW, does the reduction of free RAM means, that the code implementing the new features actually are running from RAM instead of flash?

No, that's not the case. Most code is running from flash, except interrupt handlers which are placed in IRAM.

[bmarkus]

BTW, does the reduction of free RAM means, that the code implementing the new features actually are running from RAM instead of flash?

[jeremy]

@LoneTech did you rename the file to appimg.bin ?

[bmarkus]

@daniel

Thanks, looking forward the new version.

@LoneTech

I used the standalone Windows updater. os.uname() reports release='0.9.1.b1'

[LoneTech]

Would you mind telling which firmware image you managed to load? I've downloaded all three updater packages, and tried updating three LoPys, but all of them seem to ignore the update.

de79a8db26bb6274d18fb7bb5cfbe6ad  /home/yann/Downloads/pycom_firmware_update_0.9.1.b1.exe
bd7b59899c57a919c4e7959fdc844330  /home/yann/Downloads/pycom_firmware_updater.0.9.1.b1.tar.gz
3baad66ae55f84123df223032336b0b6  /home/yann/Downloads/pycom_update_0.9.1.b1.dmg

The contained firmwares in all three are:

ce48656f828f652b04d608d763d3a08e  lopy_0.9.1.b1_868.bin
b6e45294663f0053c01c60bd4d533aee  lopy_0.9.1.b1_915.bin
3186027a26a6423a7e6949ba60a59809  wipy_0.9.1.b1.bin

I've FTPd over both lopy images, from windows, linux and osx, on three lopy boards (two earlier updated, one with only factory firmware), with no effect each time.

[daniel]

Hello bmarkus,

71,152 is the expected value for now. We will increase it in future releases, and we should end-up with something around 150K. The old value of 191,088 was wrong and that was causing the gc crashes.

Calling gc.mem_free() on the REPL will decrease the value as it creates new objects. As soon as you call gc.collect() the free memory amount will increase again.

Cheers,
Daniel