<rqou>
lesson: udp send() is much faster than printf (cc azonenberg :P :P )
<rqou>
but sendto() is slow as shit (at least in the snek language)
<rqou>
on my <redacted> device, print in the snek language ~ 30 ms; sendto ~ 400 ms; send < 30 ms
<azonenberg>
o_O
<azonenberg>
Sounds like sendto() needs to get its snoot booped
<pie__>
boop
jhol has quit [Quit: Coyote finally caught me]
oeuf has joined ##openfpga
Ultrasauce has quit [Quit: No Ping reply in 180 seconds.]
Ultrasauce has joined ##openfpga
Bike has quit [Quit: Lost terminal]
Ultrasauce has quit [Ping timeout: 248 seconds]
GenTooMan has quit [Quit: Leaving]
Ultrasauce has joined ##openfpga
nrossi has joined ##openfpga
day__ has joined ##openfpga
daey has quit [Ping timeout: 248 seconds]
day__ is now known as daey
pie___ has joined ##openfpga
pie__ has quit [Ping timeout: 268 seconds]
pie___ is now known as pie_
Hootch has quit [Read error: Connection reset by peer]
Hootch has joined ##openfpga
massi has joined ##openfpga
promach has quit [Ping timeout: 258 seconds]
Dolu has joined ##openfpga
balrog has quit [Ping timeout: 240 seconds]
balrog has joined ##openfpga
cr1901_modern1 has joined ##openfpga
cr1901_modern has quit [*.net *.split]
azonenberg_work has quit [*.net *.split]
bubble_buster has quit [*.net *.split]
Morn_ has quit [*.net *.split]
mtp has quit [Ping timeout: 265 seconds]
mtp has joined ##openfpga
mtp has quit [Changing host]
mtp has joined ##openfpga
bubble_buster has joined ##openfpga
Morn_ has joined ##openfpga
azonenberg_work has joined ##openfpga
m_t has joined ##openfpga
azonenberg_work has quit [Ping timeout: 250 seconds]
Dolu has quit [Read error: Connection reset by peer]
Dolu has joined ##openfpga
azonenberg_work has joined ##openfpga
Dolu has quit [Ping timeout: 250 seconds]
scrts has quit [Ping timeout: 240 seconds]
jarhab[m] has quit [Ping timeout: 255 seconds]
pointfree1 has quit [Ping timeout: 264 seconds]
azonenberg has quit [Ping timeout: 246 seconds]
azonenberg has joined ##openfpga
scrts has joined ##openfpga
promach has joined ##openfpga
pointfree1 has joined ##openfpga
teepee has quit [Ping timeout: 248 seconds]
teepee has joined ##openfpga
jarhab[m] has joined ##openfpga
Dolu has joined ##openfpga
teepee has quit [Ping timeout: 260 seconds]
teepee has joined ##openfpga
pie_ has quit [Ping timeout: 268 seconds]
pie_ has joined ##openfpga
pie__ has joined ##openfpga
pie_ has quit [Remote host closed the connection]
teepee has quit [Ping timeout: 240 seconds]
teepee has joined ##openfpga
jhol has joined ##openfpga
mifune has quit [Ping timeout: 240 seconds]
mifune has joined ##openfpga
pie__ has quit [Read error: Connection reset by peer]
pie_ has joined ##openfpga
teepee has quit [Ping timeout: 272 seconds]
teepee has joined ##openfpga
shellrider has quit [Quit: Lost terminal]
Dolu has quit [Ping timeout: 240 seconds]
Dolu has joined ##openfpga
pie_ has quit [Read error: Connection reset by peer]
pie_ has joined ##openfpga
m_t has quit [Quit: Leaving]
user10032 has joined ##openfpga
pie_ has quit [Ping timeout: 258 seconds]
massi has quit [Remote host closed the connection]
pie_ has joined ##openfpga
mifune has quit [Ping timeout: 248 seconds]
mifune has joined ##openfpga
digshadow has quit [Ping timeout: 276 seconds]
user10032 has quit [Ping timeout: 264 seconds]
<awygle>
azonenberg: gut check - how much ram would you expect an fpga workstation that might be running our parallel placer to have?
user10032 has joined ##openfpga
<azonenberg>
Depends on how big the chip design is
<azonenberg>
i'd say 16-32 GB quad core is a reasonable minimum for working on moderately large designs (say kintex ultrascale class)
<azonenberg>
But i really want to be able to scale to much larger stuff
<azonenberg>
My new workstation will be 2 sockets, probably 12 cores per socket so 24 cores total, and 128 GB total ram
<azonenberg>
I dont want to require that much, but if somebody is parallelizing to 24 cores they probably will have it
<awygle>
Okay, cool.
<Dolu>
Running multiple synthesis in // could be funny too ^^
<azonenberg>
I think the static usage will be pretty minimal though
<azonenberg>
i.e. the O(1)
<azonenberg>
most of the ram will be dependent on the design size and/or number of cores
<awygle>
16gb should fit even the biggest VU+ in ram as long as I'm remotely careful
<awygle>
Based on some back of the envelope
<azonenberg>
Yes, but thats for one copy
<azonenberg>
Presumably each core will be working on its own 2D tile of the chip
<azonenberg>
but maintain replicas of some state in the adjacent tiles
<awygle>
Yeah, obviously there's a lot more to it. I just wanted to see if I was completely out of bed.
<azonenberg>
So as your tile count grows, and so does the overlap %age, you could potentially have ~8 copies of the chip (every Manhattan-adjacent tile)
digshadow has joined ##openfpga
<awygle>
That's still ~512 bytes/node, even assuming somebody's doing VU+ on 16gb (id assume/hope workstation ram would expand as chip size does) so doesn't seem terrible
<azonenberg>
awygle: by way of reference i think vivado uses 9-12 GB for a relatively full xc7v2000t
<azonenberg>
So you can calculate their efficiency and use it as a rough yardstick
<awygle>
Sounds good
<azonenberg>
Also keep in mind that the actual bitstream stores things in bits but for efficiency of access our pips etc will probably be a full byte boolean and possibly even a 64-bit pointer
<awygle>
Yup
<azonenberg>
awygle: What's your availability for a get-together the 17th of this month? For an in-person brainstorming session / planning fest
user10032 has quit [Quit: Leaving]
user10032 has joined ##openfpga
scrts has quit [Ping timeout: 255 seconds]
<awygle>
azonenberg: that seems doable. Should just squeak in before relatives arrive
theMagnumOrange has quit [Ping timeout: 255 seconds]
<azonenberg>
awygle: yeah, i'll be traveling for the next 2 weeks then home for a couple of days then off for xmas stuff again
<azonenberg>
Let's discuss in more detail later on but tentatively plan for that
<azonenberg>
Easiest is probably to meet at my lab, i can grab you from the ferry terminal since the buses kitsap-side dont have good service on weekends
<rqou>
wait wait the 17th?
<rqou>
i think i can make that
nrossi has quit [Quit: Connection closed for inactivity]
theMagnumOrange has joined ##openfpga
<azonenberg>
rqou: you're gonna be in town for that?
<rqou>
afaict i'll be done with finals but not in germany yet
<azonenberg>
meanwhile i'll have just come home from germany :p
scrts has joined ##openfpga
Hootch has quit [Quit: Leaving]
scrts has quit [Ping timeout: 260 seconds]
user10032 has quit [Quit: Leaving]
pie_ has quit [Ping timeout: 240 seconds]
m_w has quit [Ping timeout: 276 seconds]
m_w has joined ##openfpga
* awygle
has never been to Germany :-(
* jn__
has rarely been outside of germany
scrts has joined ##openfpga
<shapr>
jn__: come visit me in Atlanta!
<qu1j0t3>
i've been in a couple of times, but i got out again
<shapr>
It'll be fun!
<awygle>
y'all'd have fun
<shapr>
very southern, bonus points to you
<awygle>
I used to live in LA, I can make fun
<awygle>
Louisiana that is, not Los Angeles
<shapr>
also not Lower Alabama? :-)
<awygle>
No, but I've been to Huntsville many times
Dolu has quit [Ping timeout: 240 seconds]
<shapr>
I used to live in HSV
<rqou>
the colorspace? :P
<shapr>
yeah, it's just next to CMYK
<rqou>
where's La*b*?
<shapr>
We'd have to do an A* search to find it
<shapr>
rqou: ok ok, jokes aside, HSV is the airport code for Huntsville
scrts has quit [Ping timeout: 240 seconds]
ZipCPU|Laptop has joined ##openfpga
<awygle>
I was only there because there's a Nasa center there that my dad got sent to semiregularly
m_w has quit [Quit: leaving]
<awygle>
I am an absolute ESD disaster today. The RH in this video must be 0%
<awygle>
.. Building, wtf phone "video"?
<shapr>
awygle: My family is from Alabama, so I ended up in the only tech city in Alabama for coding
scrts has joined ##openfpga
<rqou>
awygle: move to HK, the RH will be 99% all day every day
<rqou>
:P
ZipCPU|Laptop has quit [Ping timeout: 240 seconds]
<awygle>
rqou: "thiiis humidity is toooooo high!"
scrts has quit [Ping timeout: 255 seconds]
<awygle>
Actuality it's literally raining outside, it's just the building heat
scrts has joined ##openfpga
Dolu has joined ##openfpga
mifune has quit [Remote host closed the connection]
<felix_>
hmm, i'll be in hong kong for 12h in the beginning of january; do you now if it's easily possible and worth it to leave the airport and have a look at the city?
<rqou>
you can go visit m-labs :P
ZipCPU|Laptop has joined ##openfpga
<rqou>
do note that getting to/from the airport takes a while
scrts has quit [Ping timeout: 248 seconds]
<rqou>
the HK airport is on a reclaimed island very far from everything (because they didn't want a repeat of what happened to the previous one)
<felix_>
oh, right, m-labs is in hong kong
<rqou>
as an EU citizen you don't need a visa
<felix_>
yep, it's schengen area
<azonenberg>
jn__: whereabouts in germany?
<azonenberg>
I'll be flying to Dusseldorf in a couple weeks and staying in Bochum
scrts has joined ##openfpga
<awygle>
Hong Kong used to be next on my list for travel, but I might end up moving Europe up ahead of Asia
<rqou>
hmm, i don't think HK is particularly amazing for traveling to
<rqou>
although you can always cross over to SZ instead
<rqou>
(make sure to get your visa sorted out first if you want to do that)
<awygle>
Yeah I'd probably make it part of a larger tour. Like I did Singapore and Malaysia in the same trip.
<rqou>
idk if you can directly apply for a multi-entry visa nowadays
<rqou>
china keeps changing the policy
cr1901_modern1 is now known as cr1901_modern
ZipCPU|Laptop has quit [Ping timeout: 240 seconds]
ZipCPU|Laptop has joined ##openfpga
scrts has quit [Ping timeout: 260 seconds]
Dolu has quit [Ping timeout: 248 seconds]
Dolu has joined ##openfpga
Dolu has quit [Read error: Connection reset by peer]
ZipCPU|Laptop has quit [Ping timeout: 248 seconds]
scrts has joined ##openfpga
Dolu has joined ##openfpga
Dolu has quit [Read error: Connection reset by peer]