<tp>
remexre, heheh, you'll probably be the one writing the Forth GPU driver
<remexre>
yuuuuup
* tabemann
likes platforms where everything is in the datasheet, and the datasheet is less than a thousand pages long
<remexre>
been trying to convince a samsung guy I know to do it (not their gpu)
<remexre>
he's proving resistant to writing any code that's not C/C++, sadly
<tp>
tabemann, so you dont like STM32 ?
<tabemann>
tp: well, it's spread amongst three different datasheets
<rdrop-exit>
check out the iPhone SoC, 6-core + "machine learning accelerators"
<tp>
remexre, youre not having much luck getting a C/C++ Samsung guy tor write a GPU driver in Forth for the Apple_A13 ? ... colour me not suprised ;-)
<tabemann>
one for the specific MCU, one for the processor architecture, and one for the board-specific info
<rdrop-exit>
+ a 4-core GPU
<remexre>
er, it's not the apple A13; it's an RK3399 (on the rockpro64)
<remexre>
the apple gpu does seem laptoppy, though
<remexre>
er, soc*
<tp>
remexre, and youd need the Samsung guy to do it probably as the GPU is probably password protected or some crap
<remexre>
tp: linux drivers do exist, but I've been trying not to read any linux src so I don't get infected by gpl
<tp>
remexre, I have a Asus 'transformer' tablet, it's a great unit, made in 2012 and still perfect, including both internal batteries .... BUT the GPU is locked, it can never run unix
<remexre>
might try to clean-room it w/ the samsung guy, though
<remexre>
oof, tragic; the rk3399 is pretty open, as far as I know the GPU is the only thing for which the Linux driver is less than perfect
<tp>
yeah, they were warned in New York about that but they had a big Linux GPU party and now everyone is infected!
<tabemann>
lol
<tabemann>
zeptoforth got infected because I wanted to borrow code from mecrisp-stellaris
<tp>
remexre, well OSS has always had problems with GPU's because like FPGA's the manufacturers get all IP mad about their 'baby'
<tp>
tabemann, that will teach you to wear a N4 mask next time while coding!
boru` has joined #forth
boru has quit [Disconnected by services]
boru` is now known as boru
<tabemann>
lol
<remexre>
tp: yeah, it's unfortunate... tho the samsung guy says there are non-trivial tricks in the driver that do actually provide a perf advantage on android, so maybe there's something to it
<remexre>
whereas the same prolly isn't true for an ethernet adapter or smth
<tp>
remexre, I guess everyone has their own 'special tricks', lol, which are probably the same as the other guys 'special tricks' but as theyre all sooper-seeekret, no one knows
<tp>
plus they probably all steal the methods from each other and use "IP" to hide it anyway
<remexre>
can't say personally, but my impression was that the mobile gpu space is kinda a shitshow, to the point where some guys' tricks are actively harmful
<tp>
remexre, i bet it is because it's so sooper-secret
<tp>
where does super secret actually help in OSS ?
<remexre>
oh, I'm not saying it helps driver code quality or anything; I'm saying it makes whatever popular android game of the week run at the same quality with 0.1% less battery use, and that's enough to make their SoC get a $10M bulk sale
<tp>
remexre, you may be interested to know that the GPU's used in certain ATI boards that are fully open source are said to work far better than the official ATI drivers
<tp>
remexre, lol
<tp>
remexre, I have a $650 nvidia GTX660 in this box and frankly a $35 ATI card in my other box running FreeBSD looks identical on the same monitor, I cant notice the difference in X mode
<remexre>
oof, tbf I've heard nothing good about the amd driver dev team going back years now...
<tp>
remexre, yeah, same here
<tp>
bit the FreeBSD driver for that card is perfect, and no hassle, it runs perfectly after the base install
<tp>
when I saw how good it was, I felt a bit ripped off paying $650 for the nvidia. Sure nvidia has the CUDA cores and if you run Blender youll want them, and Kinux
<tp>
Linux
<tabemann>
back
<tp>
but for a x server and the occasional game, $35 ATI has been flawless for me
<tp>
and could have saved me $615
<tabemann>
okay, did an official release including the numbered local labels
<rdrop-exit>
congrats
<tp>
tabemann, Ive dl it, havent run it yet as Im still finishing my bit@ page
<tp>
tabemann, ime excited to try it because then I can try the RTS assembly mod! very exciting!
<tabemann>
cool
<tp>
tabemann, if it works ok, it will save you years of work everytime someone want to port to their STM32 chip
<tabemann>
unfortunately I can't try out RTS because one of my boards uses USB-serial built in and the other, well, uses serial, but my USB-serial dongle doesn't have RTS/CTS pins
<tp>
in that you can write a page about how to do it with your dissasembler, heck, I'll do one
<tp>
yeah I know, but Im perfectly placed to do it for the F407
<tp>
'porting' usually involves changing the USART with Mecrisp-Stellaris
<tp>
then handling the flash controller
<tp>
thats about it afaik
<tp>
tabemann, I see youve been forked now as well!
<tabemann>
yeah
<tp>
tabemann, someone forked you!
<rdrop-exit>
that's gotta hurt
<tp>
Elliot Williams
<tp>
hes the main Crapaday Forth writer
<tp>
and his hobby seems to be forking anything hes interested in on github
<tabemann>
I should check whether he's forked STM8EF
<tp>
I've noticed a LOT of coders who actually never release anything original original forking other peoples work
<tp>
he has
<rdrop-exit>
sporkers
<tp>
so github has tons of forked and utterly unmaintained programs everywhere
* tabemann
got in contact with the STM8EF guy and now he wants to try out zeptoforth
<tp>
these people dont even say why they forked
<tp>
hes a cool guy!
<tp>
and STM8EF is a mod of E Tings "eforth" for the STM8, it looks damn good
<tabemann>
I envy how small STM8EF manages to bbe
<tp>
well it is a 8 bit mcu
<tp>
the smaller the cell size the smaller the Forth I think
<tp>
I thought a tethered STM8 would be awesome, but matthias doesnt like the ISA
<tp>
shame as the MCU can be bought brand new for $0.20 or so
<tabemann>
apparently there's lots of ubercheap devices from China which use either STM8S or clones thereof
<tp>
yes there is
<tp>
one can get a 8 digit temperature controller with the STM8 in it from china for $2
<tp>
Im not aware of any STM8 clones, but given the Chinese drive to counterfeit everything, it wouldnt be surprising
<tp>
just writing a binary to find the counterfeit STM32F103's was enough for me in this lifetime
<tp>
which incidentally has been dl 273 times now
* tabemann
would prefer not to deal with counterfeit chips, hence why he spent $25 on a DISCOVERY board from STMicroelectronics rather than a $3 blue pill from who knows where with who knows what
<tp>
MrMobius, good advice, the MSP430 is definitely special in many ways.
<tp>
MrMobius, I think one of those ways is the tight peripheral integration making actions such as a one word low power mode apply to the usart
<tabemann>
low power modes with the STM32L476 and STM32F407, well, you have to remember to keep your USART on while sleeping or stopped, or otherwise your USART interrupts won't fire
<tp>
tabemann, the stm32X low power modes are very complex compared to the MSP430 which leads the field in this area
<tp>
being 16 bits, the MSP430 is probably the lowest power MCU around
<tp>
tabemann, I know the STM32Lxxxx has a 'low power usart' but I havent looked at it yet
<tp>
tabemann, I have three variants of the STM32L0xxxx here now
<tabemann>
yeah, I saw the "low power usart", but I don't know how to use it
<tp>
tabemann, it seems to have a bunch of special low power facilities and I'll get to it one day!
<tp>
tabemann, Ive a STM32L073 Nucleo with 192kB flash Ive added RTS handshaking to
<tp>
tabemann, so next in it's agenda is speeding up the clock to 35Mhz, then finding out how low I can get the power usage, then the low power usart
<tabemann>
good luck!
<tp>
it has to wait until my current projs are dome
<tp>
done
<tp>
the STM32L073 will all be grunt work really
<tp>
because I have my svd2forth stuff, this kind of thing is a lot easier for me
<tabemann>
now, done with my bugfix release
deesix has quit [Ping timeout: 258 seconds]
<tabemann>
this's been two releases in the same day, but I kind of have adopted the policy of always making a bugfix release, no matter how minor the issue, when I find a bug somewhere
dddddd has quit [Ping timeout: 264 seconds]
deesix has joined #forth
<tabemann>
okay, I'm gonna hit the sack - g'night guys
WickedShell has quit [Remote host closed the connection]
rdrop-exit has quit [Quit: Lost terminal]
antaoiseach has joined #forth
gravicappa has joined #forth
_whitelogger has joined #forth
Labu has joined #forth
proteusguy has quit [Ping timeout: 272 seconds]
dys has joined #forth
proteusguy has joined #forth
dys has quit [Ping timeout: 256 seconds]
antaoiseach has quit [Quit: leaving]
TCZ has joined #forth
guan has quit [Ping timeout: 252 seconds]
guan has joined #forth
xek has joined #forth
X-Scale` has joined #forth
X-Scale has quit [Ping timeout: 256 seconds]
X-Scale` is now known as X-Scale
TCZ has quit [Quit: Leaving]
rdrop-exit has joined #forth
X-Scale has quit [Ping timeout: 258 seconds]
X-Scale` has joined #forth
X-Scale` is now known as X-Scale
dys has joined #forth
dddddd has joined #forth
iyzsong has joined #forth
dave0 has quit [Quit: dave's not here]
rdrop-exit has quit [Quit: Lost terminal]
TCZ has joined #forth
_whitelogger has joined #forth
jedb_ has joined #forth
jedb has quit [Read error: Connection reset by peer]
<tp>
lol, look at this STM8 Hackaday review sentence, see anything wrong with it ? "The good thing is, this isn’t a blinking LED example. The code he shows uses interrupts and reads analog values. He doesn’t get into a lot of details"
<tp>
somehow it's now popular for people who probably cannot blink a led if their life depended on it, to cast disdain on any led blinking article while unknowingly showcasing their utter ignorance of all things embedded
<tp>
I guess that Hackaday journo cant imagine a blinking led example using interrupts ?
<tp>
MrMobius, you are right!
<tp>
Following the market launch of the award winning STM32TM microcontroller,
<tp>
STMicroelectronics completes the renewal of its microcontroller product line with the announcement of the STM8S family.
reepca has quit [Read error: Connection reset by peer]
reepca has joined #forth
<tp>
MrMobius, hmm, the stm32F103 was released in 2004
TCZ has quit [Quit: Leaving]
Labu has quit [Quit: Leaving.]
TCZ has joined #forth
TCZ has quit [Quit: Leaving]
<tp>
MrMobius, and the STM8S announced in July 2009
reepca has quit [Read error: Connection reset by peer]
reepca has joined #forth
<tp>
so the STM8S appears to be a more recent product than the STM32F103, but it's 2 years before the chip I use, the STM32F051
<MrMobius>
hmm, some of the docs say it has something that is good for romless designs
<MrMobius>
and accessing external code memory serially or in parallel
<MrMobius>
cant find any examples of this though
<tp>
MrMobius, I think the STM8S is a underrated chip in the west, seems very popular in China, and Ive seen them for sale at $0.18 USD each, tho not for a while
<MrMobius>
interesting that they went with the x and y index register setup
<MrMobius>
as opposed to a bunch of byte sized registers
TCZ has joined #forth
<tp>
true .. I asked the author of Mecrisp-Stellaris if he would consider a tethered forth for it on the same basis as mecrisp-across, but he said he dislikes the STM8 ISA
<tp>
where MSP430 chips are $5 compared to $0.20 for a STM8
<tp>
during my admittedly brief STM8S investigation, ISA looked good to me