<cr1901_modern>
Hope it is useful to some people in here
<awygle>
cr1901_modern: cool, will definitely check this out
<cr1901_modern>
awygle: Tyvm :)
<azonenberg>
cr1901_modern: re-replying here just so the whole channel can see
<azonenberg>
In general, posting any material related to open source FPGA tools, at any level (from HDLs/synthesizers down to bitstream twiddling) is considered on topic
[X-Scale] has joined ##openfpga
mtp_ has joined ##openfpga
waylon531 has joined ##openfpga
X-Scale has quit [*.net *.split]
sunxi_fan has quit [*.net *.split]
mtp has quit [*.net *.split]
specing has quit [*.net *.split]
uelen has quit [*.net *.split]
mtp_ is now known as mtp
mtp has quit [Changing host]
mtp has joined ##openfpga
[X-Scale] is now known as X-Scale
specing has joined ##openfpga
digshadow has quit [Quit: Leaving.]
fpgacraft1 has quit [Quit: ZNC 1.7.x-git-709-1bb0199 - http://znc.in]
<balrog>
(well for the value one could get out of it)
<sn00n>
ah ok, a mounting kit
<sn00n>
i should by a cool rack too
<azonenberg>
sn00n: Yeah i have a 16U rack on my desk now
<azonenberg>
right now it's just a PDU, then my lecroy dso is just kinda sitting on the base plate
<azonenberg>
once all the stuff gets in there will be two 2U half-width PSUs side by side at the bottom
<azonenberg>
then a 1U BNC patch panel
<azonenberg>
then the lecroy, then the rigol, then the PDU
<azonenberg>
and i think i'll have 1U free
<azonenberg>
the lecroy is 7U racked and has three BNCs on the back (trigger in, trigger out, func gen) that i wanted to bring out to the front so a 1U patch panel brings it to 8U total
<azonenberg>
then the PSUs side by side are 2 so that's 10U
<azonenberg>
the rigol is 4U so 14U, then the PDU is 1U so 15
<azonenberg>
yeah i have one free
<azonenberg>
if i add a patch panel for the rigol that will fill it up
azonenberg_work has joined ##openfpga
<whitequark>
balrog: no comment
<balrog>
whitequark: hah
<balrog>
I'm wondering if they're any good :p
<sn00n>
azonenberg: yeah, a patch panel would be quite handy for me too
sgstair_ has joined ##openfpga
sgstair has quit [Disconnected by services]
sgstair_ is now known as sgstair
sgstair_ has joined ##openfpga
sgstair has quit [Disconnected by services]
sgstair_ is now known as sgstair
kristianpaul has quit [Quit: Lost terminal]
egg|zzz|egg has joined ##openfpga
eduardo__ has joined ##openfpga
eduardo_ has quit [Ping timeout: 240 seconds]
gnufan has joined ##openfpga
gnufan1 has quit [Ping timeout: 264 seconds]
kristianpaul has joined ##openfpga
sgstair_ has joined ##openfpga
sgstair has quit [Disconnected by services]
sgstair_ is now known as sgstair
carl0s has joined ##openfpga
<carl0s>
Hi, what is the status of CoolRunner II support on yosys? Saw on twitter that 32A devices are supported, but most of the boards on ebay have a 62 device in it.
<rqou>
whee, new and exciting failure mode in my apartment: the sink has hot water, but the shower does not have hot water
<rqou>
how is this even possible?
<shapr>
rqou: insufficient water pressure? blockage in the pipe?
<rqou>
no, water is flowing just fine, it's just cold
<rqou>
but the sink isn't
<rqou>
carl0s: coolrunner-ii support is "almost working"
<shapr>
Do you have two separate shower valves? turn the flow entirely hot?
<rqou>
yeah, i did
<rqou>
something is f*cked in our apartment
<carl0s>
rqou, the pnr is done with Xilinx tools? Just searched a bit and the ISE Webpack is a little big
<rqou>
place and route is not being done with xilinx tools
<rqou>
we are working on our own place and route for cpld/crossbars
<rqou>
i was going to work on it more this week, but stuff keeps coming up (such as midterms)
<carl0s>
oh nice, thanks for that work
<azonenberg>
carl0s: so a bit of background
<azonenberg>
CoolRunner-2 has two main sub-families
<azonenberg>
the 32 and 64 are low-end and are missing a bunch of features like SSTL I/Os (for DDR memory) and clock dividers
<azonenberg>
I've focused mostly on the low end, although rqou has done some RE on the high end parts too
<azonenberg>
The bitstream for the 32a is fully understood, although the tools aren't 100% done yet
<azonenberg>
We have yosys based synthesis working, though not super efficient, and the PAR is in progress
<azonenberg>
The 64a is mostly there, we have to recover the interconnect structure for the ZIA (routing fabric) still but that should be relatively easy
<azonenberg>
just hasnt been a priority
<azonenberg>
all of the PLAs are the same for all devices, though coded slightly differently in the bitstream
<azonenberg>
The macrocells for the high-end vs low-end devices add a few more features
<azonenberg>
I have no idea what the routing for the high-end parts looks like, rqou has touched it a bit but not in extreme detail
<azonenberg>
carl0s: in general coolrunner is probably the 3rd most feature complete chip with an open toolchain right now
<azonenberg>
ice40 is top (not our team but we collaborate with clifford etc a lot), greenpak4 is second
<azonenberg>
then 4th is probably psoc5lp, although that is far from usable atm
<rqou>
azonenberg: the high end part routing is understood enough to program it
<rqou>
but i have no idea what the physical structure is
<rqou>
probably a tree of muxes
<azonenberg>
ah, ok
<shapr>
I lurk here for the extemporaneous lectures
<azonenberg>
yes a tree of one-hot muxes is the most plausible explanation
<azonenberg>
We still need the ZIA via muxes for all >= 64 chips though
<rqou>
yeah, i've been testing with the illegal source
<rqou>
which isn't in our repo, so it should be ok?
<carl0s>
thanks azonenberg, i just wanted to use lower end PLs, iCE40 is big for my toy projects, there is a board on ebay with a CR 64A and is really cheap, enough logic cells for me
<rqou>
yeah, about those boards...
<azonenberg>
rqou: lol
<rqou>
btw this reminds me
<rqou>
azonenberg you still need to image the chips i sent you
<azonenberg>
Yes, i know
<azonenberg>
see PM
<carl0s>
those boards are that bad?
<rqou>
they're not made for the most legitimate of reasons
<azonenberg>
carl0s: they're intended to be used as xbox modchips (i forget which generation)
<azonenberg>
not that you can't use them for legitimate purposes
<rqou>
360
<carl0s>
i noticed that, another cheap board is one on Seed but i think they are out of stock
<rqou>
they're somewhat limited on broken-out pins though
<azonenberg>
Yeah
<azonenberg>
carl0s: you can get coolrunners on digikey for $2ish, a 44-TQFP isn't that hard to hand solder
<azonenberg>
just pick up a random tqfp breakout from adafruit or something
<azonenberg>
then you can use all the pins, you just have to add a 1.8V regulator
<azonenberg>
which is easy enough
<carl0s>
yeah, that might end up cheaper and easier to get started with
sgstair_ has joined ##openfpga
sgstair has quit [Disconnected by services]
sgstair_ is now known as sgstair
<azonenberg>
the modchips arent a bad choice if you only need a few pins
<azonenberg>
But, do you? :p
<carl0s>
the more the best haha
<carl0s>
i will use it for toy projects, nothing serious
carl0s has quit [Quit: Leaving]
<azonenberg>
Hmm apparently the round machine pins i got are the wrong diameter
<azonenberg>
right pitch though, which is a start...
santitox has joined ##openfpga
Zorix has quit [Ping timeout: 264 seconds]
<balrog>
azonenberg: just drill it out (just kidding lol)
<azonenberg>
drill what out?
<azonenberg>
my pins are too small for the mating socket
<azonenberg>
they dont mate firmly
<azonenberg>
Drilling the BNC was easy, i used a cutoff wheel and it came off nicely
Zorix has joined ##openfpga
<azonenberg>
Also, good thing that i wore a face shield when doing that
<azonenberg>
when the BNC shell broke completely it pinched a bit and shattered my dremel cutoff wheel
<azonenberg>
Goggles would have stopped the worst of it but i probably still would have caught some shrapnel
teepee has quit [Ping timeout: 258 seconds]
teepee has joined ##openfpga
<balrog>
ohh, too small :(
<azonenberg>
yeah i have 0.47mm diameter pins
<azonenberg>
and an 0.75mm socket
<qu1j0t3>
azonenberg: +1 for safety equipment, wow
<azonenberg>
Yeah i was pretty close to it too, had my face down level since the workpiece was small
<azonenberg>
I always wear eye pro when working with any power tools, but i've started using the face shield whenever i'm working with grinding wheels just b/c i've had so many shatter when something shifted and pinched them or something
<azonenberg>
Especially the tiny dremel ones
<azonenberg>
Not as energetic as a big angle grinder, but they break a heck of a lot more frequently
<rqou>
azonenberg the PPE guy :P
<azonenberg>
rqou: i dont have any H2SO4 burns, unlike dig
<azonenberg>
so i must be doing something right
<rqou>
are nitric burns better or worse? :P
<azonenberg>
It's probably not *quite* as nasty to the skin
<azonenberg>
but NO2 is way worse for your lungs
<rqou>
it reacts with proteins and stains your skin
<azonenberg>
yeah but it doesnt burn you as quickly/severely
xdeller has joined ##openfpga
<azonenberg>
in terms of long term scarring etc
<azonenberg>
It just vaporizes, gets in your lungs, and kills you that way :p
<rqou>
but in that case (cold) sulfuric doesn't either
<azonenberg>
eh
<azonenberg>
cold sulfuric burns too, just not quite as fast as when it's at 200C
<rqou>
it does, but not instantly
<azonenberg>
and it's syrupy which makes it a pain to decon
<rqou>
my high school chem teacher spilled some concentrated sulfuric acid on her hand years ago with no lasting effects
<rqou>
but this was right next to the sink
<rqou>
so she instantly washed it off
<rqou>
nilered on youtube has tried it too
<azonenberg>
Doesnt make it a good idea :p
<azonenberg>
and trust me, when its hot from a decap
<azonenberg>
you wont get it off fast enough
<rqou>
oh yeah definitely
<rqou>
i think my chemistry teacher was just doing a carbon pillar demo
<rqou>
not a hot decap
<rqou>
but then again i'm the one researching how to ghetto-distill WFNA, so... :P
<azonenberg>
bad idea
<rqou>
i know someone who's done it before :P
<azonenberg>
me too
<rqou>
on the topic of permanent scars, what is your opinion of one of the most common causes of permanent scars, graphite pencils? :P
teepee has quit [Ping timeout: 248 seconds]
teepee has joined ##openfpga
<azonenberg>
I have a graphite tattoo on my right wrist
<azonenberg>
Lol
<azonenberg>
i wouldnt call it a scar, the surface tissue is flat
<azonenberg>
no raised scarring
<azonenberg>
But there's embedded graphite that will probably never go away
<azonenberg>
i've had it for like 15 years
<rqou>
so azonenberg when do we try ghetto HF? :P
<azonenberg>
Pretty sure dig or whitequark has done it
<azonenberg>
but count me out
<rqou>
last time this was brought up whitequark says he can "just acquire" HF in RU
* qu1j0t3
proposes renaming the channel to #openflasksofacid
<rqou>
no, i don't keep open flasks of acid lying around
<azonenberg>
lol
<rqou>
i only keep closed flasks lying around, usually with labels :P
<awygle>
I have gotten burned by acetic acid
<awygle>
Obviously not badly, but it turns out "glacial" on an acid label is important
<rqou>
lol yeah
sgstair_ has joined ##openfpga
sgstair has quit [Disconnected by services]
sgstair_ is now known as sgstair
<awygle>
Who the hell decided that "Unicode" without qualification meant UTF-16 Little Endian?
<rqou>
microsoft
<awygle>
Apparently JEDEC as well
<rqou>
wait really?
<rqou>
why does JEDEC care?
<awygle>
This standard describes "a Unicode string" without expanding further, and it turns out to be as above
<mtp>
store it as UTF-9
<rqou>
but JEDEC does "hardware"
<mtp>
in case we have to interoperate with a PDP-10
<rqou>
i didn't expect "hardware" people to even know what this whole "unicode" business is about :P
<rqou>
so i've actually been thinking about creating a character encoding called UTF-⑨ as a "f*ck you" universal filename storing encoding
<awygle>
rqou: lol "the strongest encoding"
<rqou>
somehow as a hybrid of surrogateescape+WTF-8
<rqou>
because linux filenames can be arbitrary bytes, and windows filenames can be arbitrary 16-bit units
<rqou>
and no encoding i'm aware of can store both
<rqou>
and yes, definitely the strongest encoding :P
<awygle>
Dammit now I have one of the songs stuck in my head, thanks a lot rqou
<rqou>
lolol
<rqou>
anyways, can you mention what jedec spec is amazingly aware that unicode exists?
<rqou>
or is this super-NDA?
<awygle>
Just regular NDA :-P
<rqou>
still better than VHDL stuck with latin-1
<awygle>
Seriously? Upsetting.
<rqou>
yeah, changing it would be a breaking change
<rqou>
because vhdl is case-insensitive
<rqou>
and accented letters are allowed in names
<awygle>
I am upset about the way breaking changes are currently handled in the world
<rqou>
oh, correction
<azonenberg>
Re UTF-16
<rqou>
arbitrary bytes are allowed in comments
<rqou>
:P
<azonenberg>
UTF-16 used to be the only unicode there was
<azonenberg>
the original intent of the committee was that a wchar_t was 16 bits, and there would be at most 2^16 glyphs in all the world's languages
<azonenberg>
and the only encoding would be big or little endian
<rqou>
er, all the "actually used by countries with money" languages :P
<awygle>
Trivially falsifiable, but sure
<azonenberg>
Then people started adding all sorts of accented characters and compositional characters (rather than just storing the individual strokes for kanji, or a U and an umlaut as two consecutive code points, etc)
<azonenberg>
and it exploded
<awygle>
I believe you, but that shouldn't mean they get to bogart unqualified Unicode as a descriptor
<azonenberg>
What i mean is, the standard may have been written when unicode meant an array of wchar_t's
<awygle>
But I guess that's be a ~~breaking change~~
<awygle>
Ah I see
<rqou>
azonenberg: you forgot about the requirement for legacy round-trip compatibility
<rqou>
also messing up and encoding hangul three times
<awygle>
Well this standard is newer than that but now I get what you're saying
<azonenberg>
How long ago did unicode grow >2^16 code points?
<rqou>
more than a decade ago
<azonenberg>
yes but how much longer?
<azonenberg>
because when i first learned about unicode it was in a textbook written for Visual Studio '98
<azonenberg>
Which came out nearly two decades ago :p
<azonenberg>
It was maybe 2001-2002 when i was doing this but the book and compiler were from '98
<azonenberg>
vs.net didn't come out until 2003
<rqou>
so apparently 2.0 already had planes
<rqou>
and that came out in 1996
<azonenberg>
interesting
<rqou>
3.1 definitely has astral plane characters, and it came out in 2001
<azonenberg>
Yeah but that was long after this book
<azonenberg>
:p
<azonenberg>
i'm also legitimately curious whether unicode actually needs this many code points
<azonenberg>
emoji, for example, belong in a dingbat font
<azonenberg>
Not as code points IMO
<rqou>
well, now we have a tertiary ideographic plane because some people got (more) serious about adding Chữ Nôm and Zhuang to unicode
<azonenberg>
In general are there that many truly unique ideographs?
<rqou>
yes
<rqou>
because of Chữ Nôm and Zhuang :P
<azonenberg>
From what little i know of chinese, most chinese characters are compositions of simple strokes
<awygle>
The history of emoji is pretty darn interesting
<azonenberg>
That could each be code points in their own right
<azonenberg>
and then you'd just have four or five code points for a complex "word" ideograph
<rqou>
ah, so "ideographic description characters" and ZWJs
<azonenberg>
which is on par with how many it takes for a typical word in a phonetic language
<azonenberg>
obviously its way too late for that now
<rqou>
that might work in most cases, but there's almost certainly corner cases
<rqou>
e.g. does Zhuyin get its own codepoints?
<azonenberg>
but i'm curious, if minimizing addition of code points was actually a goal from the start
<azonenberg>
would it be possible to accurately describe all of the languages unicode now supports
<azonenberg>
with <2^16 CPs?
<rqou>
maybe, but we'd probably end up with more hilarious bugs
<pointfree>
I'd much rather use dingbats/different fonts/a character set in which the the meaning of the smallest character codes are not lost to time. I can't remember the last time I typed a character not available on my keyboard.
<pointfree>
Wouldn't it be nice of machine code instructions were printable and typable characters?
<pointfree>
*if
<azonenberg>
pointfree: yes, the majority of people write in one language most of the time
<azonenberg>
or at least one character set / two very similar ones
<rqou>
hey, i have a troll-y idea
<azonenberg>
like english and german, for example
<azonenberg>
or even english and a Cyrillic language
<azonenberg>
can coexist well
<rqou>
256 codepoints representing integers (extends like ASN.1 DER)
<rqou>
and then you ZWJ them up into images :P
<rqou>
now you can have unicode bugs, ZWJ-related bugs, and ASN.1-related bugs all at the same time
<rqou>
or a different idea:
<rqou>
U+xxxx00 EMBEDDED PNG IMAGE BYTE 00
<rqou>
U+xxxx01 EMBEDDED PNG IMAGE BYTE 01
<rqou>
etc.
<rqou>
with ZWJs in between :P
<mtp>
lole
<mtp>
same
<azonenberg>
...
<mtp>
or how about
<mtp>
U+xxxx00 BEGIN ENCODED PNG
<mtp>
U+xxxxFF END ENCODED PNG
<mtp>
and just a bytestream of image data between those two markers
<rqou>
but that causes the parser to have to change state
<rqou>
my idea doesn't
<rqou>
wait what
<rqou>
wikipedia says that they're working on asn.1 JER
<rqou>
json encoding rules
<azonenberg>
wait what
<azonenberg>
now you can have json parsing bugs in your asn.1 parsing bugs?
<rqou>
yup
<rqou>
your JER will now contain an octet stream that is DER-encoded data :P
<rqou>
waiting for DJER
<rqou>
which of course won't be JWS compatible
<rqou>
azonenberg: i should create a piece of software that uses XER and then ask you/IOA to audit it :P
<azonenberg>
XER?
<azonenberg>
is that... XML?
<rqou>
xml encoding rules
<azonenberg>
asn.1 and xml in one place
<rqou>
oh and according to wikipedia CXER exists
<azonenberg>
that can't end well
<rqou>
canonical XML
<rqou>
so why no CJER?
<rqou>
azonenberg: "we added encryption support to our data by using elliptic-curve diffie-hellman written in XSLT" :P :P