Zarutian_HTC has quit [Read error: Connection reset by peer]
Zarutian_HTC has joined #forth
<tp>
crest, wow, thats a big improvement
<tp>
20x +
<tp>
crest, stm32 -> host direction is going to be the fastest anyway
cp- has quit [Quit: Disappeared in a puff of smoke]
cp- has joined #forth
<tabemann>
back
<tabemann>
ugh
<tabemann>
working with the nRF52840-DONGLE board is such a fucking pain
<tp>
why ?
<tabemann>
they had to make everything so complex, and they seem to omit key points like where in hell is the application vectors
<tabemann>
is it just at the start of the application binary?
<tabemann>
if not, where?
<tp>
I have a svd2forth for it
<tp>
not that it helps there
<tabemann>
they also had to make their installation process so complex
<tp>
flashing you mean ?
<tabemann>
yeah
<tabemann>
and you have to make sure you don't overwrite the MBR at the start of flash, or else you've effectively bricked the device, unless you manage to solder a connector to the SWD pads on the bottom of the board
<tabemann>
but at the same time, nrfutil only flashes the pages where code is being installed, so you have to make sure to flash the space after it (up to the bootloader code) if you want to use it for Forth
<tp>
eww with other NRF boards you just connect the usual SWDIO, SWCLK and GND pins and use openocd as usual.
<tabemann>
I think this has these, but instead of pins they are surface mounts
<tabemann>
and I'm not good at soldering, and forget about surface mounts
cp- has quit [Quit: Disappeared in a puff of smoke]
<tp>
yes, I recall the pic, those test pads are just unusable
* tabemann
has been reorganizing his documentation for zeptoforth
<tp>
cool
<tabemann>
the only issue I have is that the level-5 headers with code strings appear correctly in github markdown, but in recommonmark markdown the headers just seem too big
<tabemann>
I could make them just code strings, but then the user wouldn't be able to link to the strings themselves
<tabemann>
that's my only gripe w.r.t. sphinx + recommonmark
<tabemann>
with github markdown the headers are just bolded some relative to normal code strings
<tp>
could it be your browser settings ?
<tabemann>
nah - I'm viewing the two on the same browser
<tp>
well github is probably wrong
<tp>
I know that most github pages dont render properly for me
<tp>
most are unreadable
<tabemann>
okay, sphinx+recommonmark are converting my level-5 headers to level-2 headers
<tabemann>
whereas github is keeping them as level-5 headers
WickedShell has quit [Remote host closed the connection]
<tp>
honestly many github sites are unreadable for me
<tp>
they have been for a long tim
<tp>
the fonts are widely spaced or too big or whatever
<tp>
I hardly ever browse github sites
<tabemann>
I haven't had any problems with github sites myself
<tabemann>
are you sure this isn't a font problem?
<tabemann>
okay, I should hit the sack
<tabemann>
g'night
<tp>
i dont know, but i never have that problem anywhere else
<tp>
gnight tabemann
jsoft has joined #forth
cp- has joined #forth
cp- has quit [Client Quit]
cp- has joined #forth
cp- has quit [Client Quit]
cp- has joined #forth
dddddd has quit [Ping timeout: 260 seconds]
dave0 has joined #forth
antaoiseach has joined #forth
dys has quit [Ping timeout: 256 seconds]
reepca` is now known as reepca
TCZ has joined #forth
TCZ has quit [Remote host closed the connection]
TCZ has joined #forth
TCZ has quit [Remote host closed the connection]
TCZ has joined #forth
mtsd has joined #forth
TCZ has quit [Client Quit]
TCZ has joined #forth
andrei-n has joined #forth
proteus-guy has quit [Ping timeout: 256 seconds]
reepca has quit [Read error: Connection reset by peer]
reepca has joined #forth
dys has joined #forth
[1]MrMobius has joined #forth
MrMobius has quit [Ping timeout: 256 seconds]
[1]MrMobius is now known as MrMobius
TCZ has quit [Quit: Leaving]
gravicappa has joined #forth
proteus-guy has joined #forth
xek has joined #forth
proteus-guy has quit [Ping timeout: 272 seconds]
reepca has quit [Remote host closed the connection]
reepca has joined #forth
<crest>
tp: i spent last night on the second prototype
<crest>
the first one just used four bytes in a single cell
<crest>
that worked but was very inefficient
<crest>
because the host and stm32 are always waiting for each other
<crest>
and the host just polled the whole 32bit word again and again until something changed
<crest>
if it wanted to change something it wrote (ack a byte, send a byte and mark the buffer used) it used one or two 8 bit writes
<crest>
it tranfered very little data per command
<tp>
interesting!
<crest>
and every single byte transmission that direction was blocked until the receiver polled
<crest>
now i used a large 256 (- 1) byte ring buffers per direction
<tp>
so you won the battle of the buffers ?
<crest>
all four indicies (read and write per direction) still fit in a single 32bit word
<crest>
the host still polls a single word for changes
<tp>
wow, serious design skills you have there
antaoiseach has quit [Quit: leaving]
<crest>
but this time host drains the whole buffer with one or two commands
<tp>
so 4x faster ?
<crest>
for the data and on 8 bit write to move the data
<crest>
much more
<tp>
very nice
<crest>
because the stlink does a block copy of 32bit words
reepca has quit [Remote host closed the connection]
reepca has joined #forth
<crest>
the host only has pad the read sizes to natural alignment and ignore the padding
<crest>
i wanted to avoid using aligned loads because they result in twice as many sram accesses
<crest>
stolen by the debug hardware
<crest>
now i have to find time to implement the host -> stm32 direction
<crest>
and add things to make it useful like file upload
<crest>
and the ability to reset the stm32 via swd
<tp>
is your new technique responsible for the under one second wordlisting ?
<tp>
reset the stm32 via swd, hmm thats handy if it can reset under any conditions
<crest>
it should be able to reset under all conditions except if you disabled swd at runtime
<crest>
yes adding a ring buffer to decouple the producer and consumer helped a lot
<tp>
I'm going to fit a touch pad on the development box for resetting the MCU as I do it a million times a day anyway
<tp>
unlike a PB switch I cant wear a touchpad out
<crest>
but doing the data transfers in large batched made a real difference
<tp>
the device is always right next to me anyway, so resetting it is second nature now
<tp>
crest, seems like youyre making real headway there
<tp>
tis will be the first 'new' mecrisp-stellaris thing Ive seen in a while
<tp>
we received a donated python 'ok-wait' program the other day