<azonenberg_hk>
but all the mass is on the lateral side of the upper leg
<azonenberg_hk>
the medial is nothing
<whitequark>
azonenberg_hk: did you get that just from biking a lot in sea?
<whitequark>
or was that deliberate exercise?
marex-cloud has quit []
marex-cloud has joined ##openfpga
<azonenberg_hk>
whitequark: Biking 80 miles a week (4 to boat then 4 to office, then home every work day) with ~25 pounds of laptop, spare clothes, water, and assorted other cargo
<azonenberg_hk>
plus the occasional 40/50-mile ride to the monthly meeting of search-and-rescue
<azonenberg_hk>
40 for the day if i went after work (including the commute), 50 if i went there from home
<azonenberg_hk>
With lots of hills
<whitequark>
80km, wow
<whitequark>
thats pretty extreme
<azonenberg_hk>
That 50-mile ride also has something like 2000 feet of elevation gained and lost
<azonenberg_hk>
i forget if that's one way or round trip
<azonenberg_hk>
(25 out then 25 back)
<azonenberg_hk>
Since i got the bike a year and a half ago I burned through two sets of brake pads, a chain, a set of gears
<azonenberg_hk>
surprisingly still on the original tires
<azonenberg_hk>
So yeah, i put some mileage on it :p
<azonenberg_hk>
I never really "exercised" with that as the goal
<azonenberg_hk>
i just used the bike as my primary means of transportation
<azonenberg_hk>
in part because i knew i needed some physical activity after sitting at a desk all day long
<azonenberg_hk>
But apparently i neglected other muscle groups as a result
<azonenberg_hk>
So now i'm being forced to correct that painful mistake lol
<whitequark>
thinkpiece: LOCAL MAN INJURED BY TOO MUCH BIKING
<azonenberg_hk>
lol
<azonenberg_hk>
it was actually kinda funny
<azonenberg_hk>
i knew you could pull a muscle or twist an ankle or whatever
<azonenberg_hk>
or overexert yourself in general
<azonenberg_hk>
i had no idea it was possible for excessive muscle mass to be harmful
<azonenberg_hk>
So it was never something i tried to avoid
<azonenberg_hk>
And in all fairness, having big quads is totally OK
<azonenberg_hk>
as long as the opposing muscle is equally strong
<azonenberg_hk>
and mine are... decidedly not :p
<azonenberg_hk>
So there's lots of sideways leg lifts, stretches, and other fun things in my future lol
qu1j0t3 has quit [Ping timeout: 260 seconds]
qu1j0t3 has joined ##openfpga
pie_ has quit [Ping timeout: 268 seconds]
cosmobird has joined ##openfpga
pie_ has joined ##openfpga
azonenberg_hk has quit [Quit: Leaving.]
azonenberg_hk has joined ##openfpga
<azonenberg_hk>
So what do you guys think of eventually being able to infer GP_VREF blocks?
<azonenberg_hk>
you'll still have to use one if you want a voltage divider, or need to synthesize a fixed reference
<azonenberg_hk>
but if you're driving an input pin or DAC output to the comparator
<azonenberg_hk>
it seems silly to require instantiation of a GP_VREF
<azonenberg_hk>
The same way most FPGA tools will infer a clock buffer between an input and FF, etc
<whitequark>
azonenberg_hk: sure sounds good
* azonenberg_hk
adds to mental TODO
<azonenberg_hk>
Any other inference things you think would be good to have?
<azonenberg_hk>
Already on the list: latches with set/reset (this will require clifford to add support in yosys)
<azonenberg_hk>
COUNT_ADV cells
<azonenberg_hk>
(right now up/direction changeable/clock enableable counters can't be inferred)
<azonenberg_hk>
and then various forms of aggressive optimization we mentioned earlier
<azonenberg_hk>
at some point i want to brainstorm with folks in the channel to figure out all of the differnet isomorphisms between various modes of hard IP and primitives that you might run out of
<azonenberg_hk>
i thought of a few but i'm sure more exist
digshadow has quit [Quit: Leaving.]
<cr1901>
shift reg instantiation needs work
<azonenberg_hk>
cr1901: Not surprised
<azonenberg_hk>
it was added to yosys specifically at my request
<azonenberg_hk>
So i doubt it's seen a lot of use yet
<azonenberg_hk>
especially chained shregs are probably not well tested
<cr1901>
azonenberg_hk: I haven't tested it in a while either, but my gp4-tests repo illustrates the problem
<azonenberg_hk>
Ping me when i get home and i can investigate more thoroughly
<azonenberg_hk>
hard for me to do too much serious dev and concentrate here on this little laptop in a hotel without even a mouse and keyboard
<azonenberg_hk>
i cant see the code, stdout, generated bitstream, test case rtl, etc all at once
<azonenberg_hk>
and jumping back and forth makes it hard to cross-probe and see where things went wrong
<whitequark>
hilariously i have a harder time working on a large display
<azonenberg_hk>
whitequark: o_O
<whitequark>
there's just so much space and it yearns to be filled, and keeps distracting me
<azonenberg_hk>
my setup at home is a 40" 4k flanked by two 22/24" 1080ps
<azonenberg_hk>
i have a standard setup
<cr1901>
IMHO, yosys is doing instantiation correctly/the best it can. It just doesn't have enough information to do instantiation in a GP-friendly way
<whitequark>
vs this tiny cozy display where everything is where i know it is
<azonenberg_hk>
mail client, github issue tracker, and PDF datasheets on left display
<whitequark>
pdf datasheets *almost* convince me to get a proper display but not even them
<azonenberg_hk>
IRC, general web browsing, and calculators/assorted small stuff on right
<whitequark>
dunno why my brain works like that. everyone i know is salivating at massive panels
<azonenberg_hk>
Main display at center has two side by side text editors and a tabbed shell
<whitequark>
i used to have a 19" and even that was too large
<azonenberg_hk>
then a gpak designer window kinda floating in the top left
<azonenberg_hk>
I'm used to turning my head a certain way any time i want to see some specific thing
<azonenberg_hk>
its faster than switching virtual desktops or minimizing windows
<azonenberg_hk>
the only problem is when i first sit down it takes a second or two to find the mouse cursor
<whitequark>
theres an accessibility feature that highlights it
<azonenberg_hk>
because it's hiding somewhere in this 7680x2160 jungle
<whitequark>
if you press ctrl
<azonenberg_hk>
thats windows i think, does X have something similar?
<whitequark>
DEs do
<azonenberg_hk>
also, my displays are big enough that i cant see everything at once
<azonenberg_hk>
If i look at the center display it fills almost my whole FOV
<whitequark>
mhm
<azonenberg_hk>
if i shift left a bit i can see it and one side, but not both
<azonenberg_hk>
I like it this way, it means i can give my full undivided attention to one of them then just tilt my head to context-switch to a different part of the task
<azonenberg_hk>
then go right back to where i was
<azonenberg_hk>
and being able to see the RTL, P&R code, P&R/synth logs, and generated netlist all at once is super handy for stuff like this
<azonenberg_hk>
for EDA i normally have datasheets on the left display, schematic at middle if i'm doing sch capture
<azonenberg_hk>
or sch at right and layout at middle for layout
<azonenberg_hk>
a 40" 4k is absolutely unparallelled for layout
<whitequark>
I do suffer doing EDA on this 13" display, thats true
<azonenberg_hk>
on a large multilayer board
<azonenberg_hk>
you have the resolution to see fine detail and route around obstacles
<azonenberg_hk>
while still maintaining spatial awareness of what's off in the distance
<whitequark>
but EDA and datasheets are probably the only two workloads where I wish I had a larger display
<azonenberg_hk>
to choose a good bga escape route, etc
<azonenberg_hk>
that's most of what i do :p
<whitequark>
ok that makes sense heh
<whitequark>
kde doesn't appear to have this accessibility feature
<whitequark>
... or much of them at all, what a shame
<azonenberg_hk>
i havent looked in xfce (which is what i use)
<azonenberg_hk>
i know it was there on windows
<azonenberg_hk>
i used it for a while
<azonenberg_hk>
But in this case i dont think its all that useful
<azonenberg_hk>
simply because i cannot keep the entire display in my FOV at once at my typical seating distance
<azonenberg_hk>
My normal workflow has the 4k as my primary workspace then i just glance left or right to see supplementary info
<whitequark>
my impression was that no one in xfce cares about accessibility at all
<azonenberg_hk>
its not much different than if i had one 4k monitor and a hardcopy datasheet
<azonenberg_hk>
and i've never looked
<whitequark>
what I do want is one of those A4 ebooks
<whitequark>
they're stellar but...
<whitequark>
even the chinese branded ones at taobao are $700+
<azonenberg_hk>
i went to xfce because i wanted the classic windows 2000 style UI (but with a slightly more modern look on widgets as far as shading etc)
<azonenberg_hk>
gnome/unity/cinnamon/$BUZZWORD were all trying to be osx
<azonenberg_hk>
which i hate
<whitequark>
xfce isnt bad
<azonenberg_hk>
i have nothing against mac hardware in general, and the kernel seems quite nice in fact
<whitequark>
i went for kde because i like its implementation of panels and hw control applets
<azonenberg_hk>
but the ui... the only thing i hate more than that trainwreck is systems that try to clone it
<whitequark>
which are basically the only part of kde that i use
<azonenberg_hk>
and end up being somehow even less usable than the original
<whitequark>
plasma widgets suck
<azonenberg_hk>
menus are part of the app, not the OS
<azonenberg_hk>
they do not belong at the top of the screen
<whitequark>
the idea there was that it's easy to bump the cursor into the top of screen
<whitequark>
far more so than hunting a menubar
<azonenberg_hk>
Yeah but when you have ten apps open on a 40" display
<azonenberg_hk>
trying to move the mouse all the way up there?
<whitequark>
yeah its not really aiming at your use case
<azonenberg_hk>
indeed
<azonenberg_hk>
every time i try to use a mac that is the one thing that bugs me more than anything else
<whitequark>
the really nice thing about osx is the graphics toolkit from programmer's pov
<whitequark>
it's a total pleasure to use, nothing else comes even close
<azonenberg_hk>
They made a lot of good decisions but that is one that i will never support
<whitequark>
unfortunately i really resent most of their ui patterns
<azonenberg_hk>
Yeah same here
<whitequark>
not in the least the tiny unenlargeable fonts
<azonenberg_hk>
not just like "awkward"
<whitequark>
which i simply don't see
<whitequark>
even in glasses
<azonenberg_hk>
but absolutely the polar opposite of what i want/need
<azonenberg_hk>
one button mouse
<whitequark>
oh right
<azonenberg_hk>
window buttons on the wrong side
<azonenberg_hk>
menus attached to the screen instead of the app
<whitequark>
you use the mac hardware :p
<azonenberg_hk>
I dont
<azonenberg_hk>
:p
* whitequark
has only ever used osx in a vm
<azonenberg_hk>
but if we're cmplaining about apple UIs...
<azonenberg_hk>
complain about the ui they invented
<azonenberg_hk>
not the one you used
<whitequark>
its an interesting experience because half of stuff is familiar and the other is alien
<whitequark>
ctrl does a completely different thing
<whitequark>
etc
<whitequark>
i must say they did an impressively good job with their software renderer because it's snappy at compositing in fullhd on cpu
<whitequark>
in a vm
cosmobird has quit [Ping timeout: 252 seconds]
<whitequark>
i mean "snappy" as "snappy" goes when you're stuck in a vm on a cpu
<whitequark>
by some miracle qemu just happens to emulate one of the imacs so it's actually "running" on its native hardware
<azonenberg_hk>
Lol
<whitequark>
its not even a hackintosh, just a regular installation from a bootable cd
<azonenberg_hk>
well i have work in the morning
<azonenberg_hk>
sooo i should probably get to sleep
<whitequark>
good night
<whitequark>
"An optional feature of IPv6, the jumbo payload option in a Hop-By-Hop Options extension header,[8] allows the exchange of packets with payloads of up to one byte less than 4 GB (232 − 1 = 4294967295 bytes), by making use of a 32-bit length field. Packets with such payloads are called jumbograms."
<whitequark>
who thought this is a good idea?
scrts has quit [Ping timeout: 245 seconds]
scrts has joined ##openfpga
Bike has joined ##openfpga
<qu1j0t3>
-_-
cosmobird has joined ##openfpga
amclain has joined ##openfpga
<pie_>
whitequark, maybe they had an extra bit
<whitequark>
they didnt
<whitequark>
its done through a "next header" field
<whitequark>
so its an entire new option header, even fatter than ipv4 options
<nats`>
I hope they removed the 32 bit CRC
<nats`>
:p
<nats`>
because it's fucking pointless int aht case :p
<whitequark>
ip header doesnt have the crc anymore in ipv6
<whitequark>
it relies on ethernet level crc
<whitequark>
and i think jumbograms rely on fragmentation so they use smaller link layer frames
<whitequark>
but like, wow, ipv4 is a mess and ipv6 is also a mess, just a different one
<azonenberg_hk>
whitequark: last time i checked ipv4 had a header checksum but not payload
<azonenberg_hk>
and it was a 16-bit sum, not a crc
<azonenberg_hk>
and yeah jumbograms are stupid
<azonenberg_hk>
theres a lot of ip features i just dont support in my stack
<azonenberg_hk>
because they are rarely used and should never be used :p
<whitequark>
qu1j0t3: ugh
<whitequark>
I knew most of those things separately but
<whitequark>
seeing them put together like that is depressing
digshadow has joined ##openfpga
doomlord has joined ##openfpga
cosmobird has quit [Ping timeout: 250 seconds]
digshadow has quit [Ping timeout: 268 seconds]
digshadow has joined ##openfpga
massi has quit [Remote host closed the connection]
firebird_ has joined ##openfpga
firebird_ has quit [Ping timeout: 248 seconds]
* whitequark
stares at the IPv4 header
<whitequark>
MSB 0 bit numbering?!
<whitequark>
WHY
firebird_ has joined ##openfpga
cr1901 has quit [Ping timeout: 260 seconds]
<rqou>
to cause you bugs? :P
mifune has joined ##openfpga
mifune has joined ##openfpga
mifune has quit [Changing host]
<rqou>
my father had a great story about this too
<rqou>
at one point they were working on a product that layered a MSB-is-0 protocol on top of a LSB-is-0 protocol
<rqou>
but there were a whole bunch of other layers involved so this wasn't obvious
<whitequark>
augh
<rqou>
they needed to do interop testing
<rqou>
only one other vendor had a product at that time
<rqou>
test fails and doesn't pass traffic
<rqou>
after debugging it turns out the other vendor got the bit ordering backwards
<rqou>
but they were first to market and bigger, so their way became the right way
<rqou>
whitequark: I love the name smoltcp
cr1901 has joined ##openfpga
<whitequark>
lol
<rqou>
except for the part where it doesn't do TCP :P
<whitequark>
... yet.
<whitequark>
let me finish ICMP and it will do TCP, like, tomorrow
<rqou>
I assume you got tired of lwIP?
<whitequark>
"tired" is a word not strong enough.
<whitequark>
I feel like the authors of lwIP deserve to be relocated to the United States.
<rqou>
lol
<cr1901>
did the other stack fall through?
* whitequark
makes an indefinite gesture
<cr1901>
well, it's a cute TCP stack
<cr1901>
it's TCP, but smol
cr1901 has quit [Ping timeout: 260 seconds]
mifune has quit [Read error: Connection reset by peer]
mifune has joined ##openfpga
mifune has joined ##openfpga
mifune has quit [Changing host]
digshadow has quit [Quit: Leaving.]
digshadow has joined ##openfpga
mifune has quit [Quit: Leaving]
carl0s has joined ##openfpga
digshadow has quit [Quit: Leaving.]
flaviusb has joined ##openfpga
<rqou>
people who generate unindexed PDFs are evil
SalvorH has joined ##openfpga
digshadow has joined ##openfpga
SalvorH_ has quit [Ping timeout: 264 seconds]
<cr1901_modern>
s/unindexed//
pie_ has quit [Changing host]
pie_ has joined ##openfpga
digshadow has quit [Ping timeout: 248 seconds]
<azonenberg_hk>
rqou: agreed
digshadow has joined ##openfpga
<pie_>
guess thats me
<pie_>
well, idk what unindexed means
<azonenberg_hk>
pie_: without a sidebar table of contents for you to navigate
<azonenberg_hk>
theres a few chip vendors i cant recall off the top of my head who are big offenders
<azonenberg_hk>
i think winbond for one?
<pie_>
oh
<pie_>
yeah i dont do that
<azonenberg_hk>
so you have to go up to the front of the doc, find the printed toc
<azonenberg_hk>
then scroll down X pages
<pie_>
then again my biggest pdf is like 10 pages :P)
<azonenberg_hk>
its horrible, like using a printed book
<azonenberg_hk>
the whole point of electronic docs is to be more efficient
<pie_>
its worse than a printed book
<azonenberg_hk>
and even printed binders can have tabs at starts of chapters etc :p
<pie_>
on a printed book you can at least stick your fingers in places as temporary bookmarks
<azonenberg_hk>
yeah exactly
<cr1901_modern>
I prefer single-page HTML with a TOC and links to next/prev/top/bottom
<cr1901_modern>
Note: My preference is not in contradiction with modern web browsers being a clusterfuck
<azonenberg_hk>
cr1901_modern: ideally, you should have a meta-format you can generate pdf, multi-page html, or unified html
<azonenberg_hk>
as all 3 have their uses
<azonenberg_hk>
multi-page html is good for mobile devices and low bandwidth connections
<lain>
^^^^^^^^
<lain>
this
<azonenberg_hk>
easier to navigate and less to load
<azonenberg_hk>
unified is good for desktop stuff, sometimes
<azonenberg_hk>
pdf is better for printing and other desktop use cases
<cr1901_modern>
azonenberg: it's called pandoc or Sphinx :P
<cr1901_modern>
oh, sorry misread
<azonenberg_hk>
or doxygen or...
<azonenberg_hk>
lots of stuff
<cr1901_modern>
pandoc can't do multi-page html I think. Or if it can, idk how to do it
<azonenberg_hk>
My point is, ideal docs are available in all three
<cr1901_modern>
I acknowledge PDF is important, but I loathe reading documentation in it.
<azonenberg_hk>
why so?
<cr1901_modern>
search is slow, and I have a tendency to jump around a lot. I can at least save my prev position in a browser that can parse HTML
<cr1901_modern>
I would prefer to have a "stack" of previous positions
<azonenberg_hk>
some pdf viewers can do that
<cr1901_modern>
Oh?
<cr1901_modern>
And also, the fact that PDF itself doesn't understand text, but only "how to draw text" kinda skeeves me
<azonenberg_hk>
i forget which ones but i know i've used onesthat did it
<azonenberg_hk>
pdf is a rendering format, not a semantic format
<azonenberg_hk>
i would like a better format for pretty-printed text meant to be displayed on a page
<azonenberg_hk>
maybe latex dvi? no idea how kludgy that one is
<cr1901_modern>
Somebody uses dvi besides as an intermediate format to generate Postscript?
<pie_>
next week we have to give a short 10-15 minute presentation about a topic for one of our classes and im doing semiconductor (IC) manufacturing, and i want to make i reallycool(tm), any recommendations?
<pie_>
for physicis students that dont really know anything about it
<pie_>
*physics
<cr1901_modern>
Show them a silicon ignot?
<cr1901_modern>
tngot*
<cr1901_modern>
ignot** ffs
<azonenberg_hk>
lol
<pie_>
haha
<azonenberg_hk>
Some die photos in general?
<cr1901_modern>
Aaaaaah iNgot*
<azonenberg_hk>
maybe a dash etch shot to show doping
<azonenberg_hk>
and how you can debug it
<pie_>
yeah probably gonna put an assload (metric) of pics in a powerpoint or something
<mtp>
pie_++ for 'metric assload'
<azonenberg_hk>
What distinguishes a metric assload from an imperial one?
<azonenberg_hk>
what's the conversion factor?
<pie_>
idk, sanity points? :P
<mtp>
the imperial assload doesn't actually measure ass
<mtp>
it measures seat capacity
* pie_
ponders how he is going to cram the entire chip process into 15 minutes when even he has no idea what hes talking about
<pie_>
well by then hpefully ill have a clue
<azonenberg_hk>
pie_: read my RE class notes
<pie_>
ALL THE THINGS
<azonenberg_hk>
the lectures on fab and package construction
<mtp>
pie_, there's a children's book called "The New Way Things Work" that describes it pretty well
<azonenberg_hk>
those are a good crash course
<pie_>
azonenberg_hk, yes thats a very good suggestion, ive been meaning to read them anyway
<pie_>
for like...months?
<azonenberg_hk>
i think lectures 2 and... 6ish?
<azonenberg_hk>
Lol
<mtp>
at least i remember it talking about doping and masking and semiconductors
<pie_>
ive got some pdfs of books but those are obviously pretty in-depth
<pie_>
anyway im getting ahead of myself..just need to focus on not failing a bunch of tests this week.... :I
<cr1901_modern>
azonenberg_hk: This is a stupid question, but... mulling over your "do a switch on FPGA" project more. Is there any hardware difference between WAN port and LAN port, or is it simply how the firmware treats the ports that makes them different?