reepca has quit [Read error: Connection reset by peer]
reepca has joined #forth
gravicappa has joined #forth
proteus-guy has quit [Ping timeout: 260 seconds]
<tabemann>
hey guys
<tp>
hey tabemann
<tabemann>
I wonder what would be a better use of $100... the higher end nRF board... or the numworks calculator (which I know has a mecrisp-stellaris impl. for but which I'd like to see if I could make a zeptoforth impl. for)
<tp>
here are a 200mhz Forth compiling stats : 706 lines and 7,234 bytes compiled, with 0 errors in 2,323ms ok
<tp>
tabemann, hard question :)
<tp>
and my setup on a 72Mhz M0: 11 seconds to upload and compile 42,606 bytes of source with no comments
<tp>
hmm, theyre about the same ?
<tp>
the first Forth uses interrupts and buffers and Im using default Mecrisp-Stellaris
<tabemann>
and an advantage that zeptoforth has over mecrisp-stellaris is that zeptoforth has sleeping right out of the box, whereas you need to connect the numworks calculator to a power supply or reimage it at he end of the day or else it'll drain its battery
<tp>
tabemann, yes, but your sleeping isnt the lowest power option is it ?
<tabemann>
that is true
<tp>
low power is only a 'on/off' choice on the MSP430, for everything else it's very complex
<tabemann>
and I have it being frequently woken up by the systick, so I'd need to turn systick off
<tabemann>
but systick is only really essential if I want to do something timing-related
<tp>
yeah, basically one can't slap 'low power' on afterwards, it has to be designed in from the very start
<tp>
I've been a dismal failure with cortex-m0 and low, low power
<tp>
(so far)
<tabemann>
ugh numworks isn't actually free software, as it is non-commercial... but of course I'd be replacing the numworks image with my own, so that wouldn't matter
<tp>
yeah, but the numworks design is open
<tp>
tabemann, a programmer is going to be attracted to the numworks, a embedded person to the higher end nRF board
<tabemann>
I'm obviously not an embedded person as all I do with my boards is hook up serial and usb and blink leds
<tabemann>
and play around with software
<tp>
maybe the numworks is a better choice for you then ?
<tp>
it's a excellent bit of hardware also
<tabemann>
it has a ton of flash
<tabemann>
64K internal, which is small, but 8M external, which is humungous
<tabemann>
the calculator has been ordered
<tabemann>
the only thing is that I'd like to figure out how to control the screen, rather than just serial
<tabemann>
so ideally the user could enter forth code with the keyboard
reepca has quit [Read error: Connection reset by peer]
reepca has joined #forth
<tp>
the keyboard is just a scanned martrix, I'm sure you could make it do anything
<tabemann>
well yes
<tabemann>
at least the source code for epsilon is available, if not completely free
<tp>
does it matter as Mecrisp-Stellaris is available for it under the gpl ?
<tabemann>
oh wait a minute, now I've found driver code for it
<tabemann>
so I'm not going to have to reverse engineer epsilon
<tabemann>
that is good, because I was worried about the license used by numworks infecting my code
<tabemann>
I'm okay with the GPL3 infecting my code, OTOH ha ha
<tp>
very nice calculator, excellent purchase!
<tabemann>
now I'm gonna shameless steal the driver code from mecrisp-stellaris so I can make zeptoforth run on the board :D
<tabemann>
about this:
<tabemann>
Flashing of Mecrisp-Stellaris to the numworks calculator:
<tabemann>
No need to disassemble your calculator, terminal is on USART6: PA11 = D- = TX. PA12 = D+ = RX.
<tabemann>
You need to type "eraseflash" once to clear the free dictionary space after you flash the Forth binary,
<tabemann>
dfu-util -i 0 -a 0 -s 0x08000000 -D mecrisp-stellaris-numworks-with-sources.bin
<tabemann>
Just cut an old USB cable and connect it to a USB-serial-brige with 115200 Baud 8N1.
<tabemann>
as dfu-util leaves old contents in flash blocks which are not used.
<tabemann>
how are we to know which wires in the cut USB cable to connect?
<tp>
youll need a usb cable wiring diagram I guess
<tp>
plus usb wires are very thin
<tp>
I think there are only 3 wires
<tp>
+v, d+, d- and 0v is the shield or may be a wire
<tp>
but this is confusing
<tp>
the calculator has a usb port with a virtual modem as the terminal ?
<tabemann>
if it did it'd be far more helpful than chopping up USB cables
<tp>
i hate usb cables, very fragile
<tp>
and to get super high speeds they must have very low internal capacitance, which means very thin wires
<tabemann>
from an example I saw what they did was connect a USB-A connector to connections on a breadboard, then connected each of these to serial RX and TX
<tabemann>
so no actual cutting took place
<tp>
1689 bytes/sec uploaded and compiled with tachyon Forth at 200Mhz ... thats a bit slower than my system at 3873 byte/sec
<tp>
tabemann, I've no idea what that numworks connection text means
<tabemann>
I think I've figured it out
<tabemann>
and I've put together everything I've gotta do to make it work
<tp>
tabemann, ok the numworks has a USB Micro-B connector on the case
<tabemann>
it's simply taking a Micro B to A cable and plugging it into a breakout board, then plugging the breakout board into a breadboard, then plugging male to female dupont cables into the breadboard and hooking them up to a serial-to-USB dongle
<tp>
thats connected to the MCU serial port
<tp>
what theyre doing is using the USB Micro-B connector on the case as a 3.3 SERIAL connector
<tp>
and why theyre cutting the cable, tho Id use a usb socket myself and solder the wired to that, then just use a full usb cable instead of cutting it
<tp>
the calculator is only usb when it's running firmware that uses the MCU usb peripheral, which Mecrisp-Stellaris doesnt
<tabemann>
yes
<tabemann>
I'm using a USB socket and a breadboard, because I can't see how I'd get cutting the cable and then reconnecting the wires right, and I can't solder
<tp>
good idea
<tp>
besides, the usb wires are so thin they can easily break when used outside the cable
<tabemann>
exactly
<tp>
very handy having usb and serial connections on the same pins on the stm32
<tp>
sensible design strategy
<tabemann>
brilliant design indeed
<tabemann>
I should get to bed though
<tabemann>
have done enough shopping for today
<tp>
shopping!
<tp>
buybuybuy
<tp>
night tabemann
<tabemann>
well today I bought a calculator, another usb-serial-converter (so I don't have to disconnect my current one from my F407 board, as reconnecting the two is a royal pain in the ass), usb a to micro b cable, a usb a breakout board, a breadboard, and a fuckload of male to female dupont cables
<tabemann>
I don't know why you can't buy dupont cables in small numbers
<tabemann>
as I only need two of them
<tp>
they dont last long
<tp>
better to buy in bulk as theyre crap
<tabemann>
might as well then
<tp>
it's not like theyre expensive
<tabemann>
they're really cheap
<tp>
so true
<tp>
+ nasty
<tabemann>
anyways, time to hit the sack - g'night
<tp>
theyre the worst of the worst
<tp>
cya
jsoft has joined #forth
dddddd has quit [Ping timeout: 260 seconds]
WickedShell has quit [Remote host closed the connection]
andrei-n has joined #forth
antaoiseach has joined #forth
mtsd has joined #forth
dys has joined #forth
reepca has quit [Read error: Connection reset by peer]
reepca has joined #forth
jsoft has quit [Ping timeout: 246 seconds]
xek__ has joined #forth
reepca` has joined #forth
reepca has quit [Read error: Connection reset by peer]
<crest>
so far i only implemented transfers from the stm32f103 to the host
<crest>
not the other way around
<crest>
and iirc changeing the clock speed didn't help much for "words"
<tp>
you also use a STM32F103 which is pretty fast, must have been very impressive in it's day
<tp>
in my case I can see that uploads at 72MHz are faster than 48MHz, but I'll add a proper timer soon
<crest>
i suspect that the core speed is pretty irrelevant because the debugging hardware is limited by the swd clock speed and the the stlink firmware
<crest>
sure compiling code is heavly core frequency dependent
<crest>
until you run into flash write speed limits for trivial things like just lots of constants via ,
<tp>
sure
<tp>
I think I ran into M0 core limitations at 115200 but I try everything I can for extra speed
<crest>
i suspect that for compilation my approach won't help much because it its not serial i/o bound
<crest>
if it helps it's probably just the ring buffer and you could have the same gain by putting a ring buffer between key?/key and the usart
<crest>
after the usart has a very fast sync mode with dma that should at least reach 10Mb/s
<crest>
*after all
<tp>
crest, only if dma was used with the usart
<crest>
i remeber abusing that once upon a time in university. i had to add crc based framing and a parity *byte* because i got to close to the theoretical bandwidth
<tp>
crest, I suspect Im near optimum upload speeds for a M0 atm, I was suprised to see my IDE is actually faster uploading than Tachyon Forth with a 200mhz cpu and all the frills
<crest>
and they refused to add a rs485 transceiver on each end
<tp>
damn their scaly hides!
<crest>
the crc32 hardware was fast enough
<crest>
but the poor sob who had to add dma from spi to crc and usart wanted to kill them
<crest>
but he did implement my crazy atm inspired idea and it worked
<crest>
to push 9,x Mb/s video data through the usart over too damn long unshielded cables
<tp>
I have a certain 'slowness' trigger point for all my development stuff and as long as I'm above that I'm happy. I cant stand repetitive delays
<crest>
just because "it worked single ended before"
<tp>
thats pretty cool
<crest>
sadly i wasn't allowed to keep a copy of the code
<crest>
but i remeber that he had do something strange to the dma controller to duplicate the stream from the spi to the crc and usart at the same time
<tp>
I've only ever used dma once I think, and that was with a GCC adc demo years ago
<crest>
because there wasn't enough sram to keep two frames of video
<tp>
the only speed I need is uploading, everything else is usually orders of magnitude more than I need
<crest>
if you want to play with my slow working prototype i can send you the host side c code as well
<crest>
(the one that uses as single 32bit word for everything)
<tp>
no thanks, I don't have enough experience to do anything with it, plus I'm busy on my own projects right now
<tp>
I do look forward to your updates as you develop it tho
forthen has quit [Quit: Connection closed]
Zarutian_HTC has quit [Read error: Connection reset by peer]
Zarutian_HTC has joined #forth
Zarutian_HTC| has joined #forth
Zarutian_HTC has quit [Read error: Connection reset by peer]
Zarutian_HTC| is now known as Zarutian_HTC
TCZ has joined #forth
TCZ is now known as DoomSlayer2020
gravicappa has quit [Ping timeout: 256 seconds]
dave0 has joined #forth
andrei-n has quit [Quit: Leaving]
Keshl has quit [Read error: Connection reset by peer]
Keshl_ has joined #forth
Keshl_ is now known as Keshl
DoomSlayer2020 has quit [Quit: Leaving]
reepca` has quit [Remote host closed the connection]