DocScrutinizer05 has quit [Disconnected by services]
DocScrutinizer05 has joined ##openfpga
DocScrutinizer05 has quit [Disconnected by services]
DocScrutinizer05 has joined ##openfpga
DocScrutinizer05 has quit [Disconnected by services]
DocScrutinizer05 has joined ##openfpga
X-Scale has quit [Ping timeout: 264 seconds]
X-Scale has joined ##openfpga
<bofh_> pie_: are there even still any 80s telco sats in use? pretty sure in 30-35 years you'd run out of hydrazine for stationkeeping in GEO (so you'd wind up drifting into a geosynch but not stationary orbit, significantly decreasing your utility from a telco POV), and that's *well* past the lifetime of LEO sats.
<pie_> bofh_, well now im that much smarter
X-Scale has quit [Ping timeout: 260 seconds]
<awygle> ideally they've moved up into a graveyard orbit
<awygle> you're not supposed to leave junk in GEO
<awygle> although apparently through 2005 only 1/3 of satellites succeeded in an orbit raising maneuver at EOL
<awygle> which is kind of pathetic
<pie_> thats pretty bad
<zkms> relatable tbh
<awygle> although 2005-2011 has apparently done better
* awygle is just quoting wikipedia into irc
<pie_> but, space is hard i guess
X-Scale has joined ##openfpga
<Bike> sometimes i think satellite engineering must be really good because it lasts for years with no maintenance and such
<awygle> i mean yeah space is hard but if you can do stationkeeping you can do a graveyard raise
<Bike> and then i hear something like that
<awygle> suonds to me like they just got greedy and pushed it too far lol
<kc8apf> isn't the international agreement to deorbit in 20 years?
<awygle> i think that's only LEO sats
<kc8apf> that makes sense
<kc8apf> I never worked on anything other than LEO
<awygle> (also google says it's 25)
<awygle> that's nonsense-long btw. it should be like... 5 years after end of service life
<kc8apf> it allows for passive deorbiting
<awygle> yeah, i'm arguing that's not good enough
<kc8apf> which is dumb
<awygle> i wonder where the break point is at which 25 is no longer the rule
<shapr> awygle: recently, NASA got approval to use an FPGA for the first time
<shapr> sufficiently space hardened and expensive
<shapr> I forget the brand name, I'll ask my NASA friend
<kc8apf> there are a few different rad-hard and space-rated FPGAs
<kc8apf> moderately common in commercial sats
<awygle> microsemi and older xilinx, typically
<kc8apf> atmel had a few
<awygle> you mean actel?
<awygle> oh huh, no, you mean atmel
<awygle> TIL
<awygle> "FreeRAM", excellent
<kc8apf> yeah, looks like Microchip is dropping those parts though
azonenberg_work has quit [Ping timeout: 240 seconds]
<shapr> awygle: my friend told me it's actel
<kc8apf> interestingly, they have a rad-tol ATmega128
<awygle> kc8apf: yeah that's really intriguing
<awygle> shapr: ah ok. actel is microsemi now :)
<awygle> is microchip now
<kc8apf> Arduino in SPAAAAAAACE
<shapr> yay!
<awygle> i guess technically that deal hasn't gone through yet. so they're still microsemi for like another week
<shapr> My friend was in town a few months back to go over a bunch of RFPs
<shapr> NASA is funding ideas for space instruments, vaguely DARPA style
<awygle> kc8apf: iirc GOMSpace uses the atmega? or it might be an AVR32 or whatever that weird arch is
<kc8apf> Aeroflex has a few available still as well https://ams.aeroflex.com/pagesproduct/prods-hirel-fpga.cfm
<shapr> so she was checking the proposals for sanity
<kc8apf> just don't take Kinetis parts anywhere near radiation
<awygle> lol really?
<awygle> that's fun
<kc8apf> we put K20s in a beam. hard latch up
<awygle> how beam are we talking?
<kc8apf> can't remember which facility we used. It was targeting a LEO application
<awygle> to quote a former boss, "my maiden aunt can tolerate 5 krad"
<kc8apf> probably UC Davis
<shapr> now that I think of it, my friend's expertise is in heliophysics, she built a faraday cup and now she's working on ... something way more complicated
<shapr> getting close to the sun may be a more 'energetic' problem
<awygle> look at kc8apf with his fancy calibrated university facilities :p
<awygle> we used medical linacs for the most part
<kc8apf> that team loved to spend money
<shapr> did they get results?
<kc8apf> built 3 full RF chambers for both near-field and far-field
<awygle> protip - the FRAM MSP430s are absolutely killer for space
* shapr follows kc8apf on twitter
* awygle salivates
<awygle> i would have killed for a far-field chamber
<shapr> digital magnetic logic is the way to go
<kc8apf> shapr: I'm not _that_ interesting. Those projects are long dead
<shapr> kc8apf: I make my own decisions :-P
<kc8apf> unless you count the trail of projects created as the team looked for funding
tinyfpga has quit [Ping timeout: 268 seconds]
<awygle> sounds like the google startup ecosystem imported a lot of the bad parts of the real startup ecosystem :p
<shapr> awygle: you ever read up on the purely magnetic computers that existed between vacuum tubes and semiconductors?
<awygle> shapr: not really, although i'm familiar with rope core memory
<awygle> and MRAM for that matter i guess
<kc8apf> awygle: Greg Wyler was a decent salesman
<awygle> kek
<awygle> former bosses used to joke that Greg Wyler had huge.... spectrum allocations.
<awygle> (read in monty python voice)
<kc8apf> yup
<kc8apf> he used that as his main bartering chip
<kc8apf> he lacked the leadership to actually deliver anything to use that spectrum though
<awygle> i was about to say "yeah i've heard that" and then i remembered i heard that _from you_
<kc8apf> heh
<awygle> speaking of, realized i missed you last weekend - whoops
<kc8apf> turns out I wasn't there
<sorear> digital magnetic logic is interesting because it kind of came too late
<awygle> ah! alls well that ends not-my-fault then
<kc8apf> I got my schedule wrong.
<awygle> well, do or don't hit me up next time you're in town, as you like
<kc8apf> gah. VCs will fund anything with blockchain in the description.
<kc8apf> I could probably get funded for a blockchain built on blockchain
<shapr> half the Haskell jobs floating around in blockchain
<shapr> much as I love writing Haskell, I also prefer sanity
<sorear> there was a guy at the local rust meetup last month who said his work was haskell on the blockchain. i quietly excused myself
<shapr> it does pay money
<shapr> very much money even
<shapr> I'm just not convinced the blockchain is more than a fad
<awygle> yeah i always argue with myself about how much i enjoy eating my principles when these things come through
<kc8apf> "AI Blockchain OnDemand SaaS Commerical Enablement Platform"
<kc8apf> I can't make this up
<Bike> aibodsscep isn't very pronounceable
<awygle> invent blockchain that only FPGAs can do well. embed secret hack that halves compute time but requires FPGA reverse engineering. complete project x-ray.
* awygle finally leaves the office
<sorear> or I could just announce a blockchain where the proofs of work are NFS relations for the DNSSEC root zone signing key
<qu1j0t3> shapr | very much money even // lots of things pay money that i lack either self respect or motivation to go near
<qu1j0t3> shapr: like you i have a low boredom threshold, too :)
digshadow has quit [Ping timeout: 255 seconds]
DocScrutinizer05 has quit [Quit: EEEEEEK]
DocScrutinizer05 has joined ##openfpga
digshadow has joined ##openfpga
tinyfpga has joined ##openfpga
azonenberg_work has joined ##openfpga
Bike has quit [Quit: Lost terminal]
digshadow has quit [Quit: Leaving.]
digshadow has joined ##openfpga
digshadow has quit [Read error: Connection reset by peer]
digshadow has joined ##openfpga
azonenberg_work has quit [Ping timeout: 248 seconds]
rohitksingh_work has joined ##openfpga
<azonenberg> sorear: me and rqou joked about making a cryptocurrency where proof of work involved precomputing 2048-bit discrete logs
<azonenberg> neither of us gave enough to actually do it
genii has quit [Remote host closed the connection]
m_w has quit [Quit: leaving]
m_w has joined ##openfpga
steakpizza has joined ##openfpga
steakpizza has quit [Remote host closed the connection]
<tnt> implr: to interface with surplus GSM equipement being decommisionned.
<tnt> atm the e1 card is more expensive than the BTS :p
<gruetzkopf> i might be able to get a few hundred siemens BS241, though for funny frequencies
futarisIRCcloud has joined ##openfpga
steakpizza has joined ##openfpga
steakpizza has quit [Ping timeout: 260 seconds]
azonenberg_work has joined ##openfpga
StCipher has joined ##openfpga
StCypher has quit [Ping timeout: 246 seconds]
cr1901_modern has quit [Ping timeout: 248 seconds]
cr1901_modern has joined ##openfpga
<awygle> sorta wanna spend five grand on an sla printer...
<azonenberg> awygle: lol
* azonenberg would be more interested in SLS
<gruetzkopf> ;)
<gruetzkopf> okay, this is enough multi-square-meter schematic reading and kilo-relay-debugging for today
<awygle> too expensive to feed
<awygle> plastic is cheaper
<azonenberg> awygle: yeah but i just dont have many uses for weird 3d geometry plastic
<azonenberg> most of the time i either need panel enclosures etc that are essentially 2D (laser cutting is way faster and cheaper)
<awygle> I think it would be fun to rapid prototype things like rack mount switches
<azonenberg> or significant strength
<awygle> or cat toys
<azonenberg> honestly, i'd make it out of sheet metal
<azonenberg> laser cut or waterjet the body then bend
<awygle> if I needed strength I'd get Real Manufacturing (TM)
<awygle> note that I don't have a laser cutter, or a water jet, or sheet metal bending tools. lol
<gruetzkopf> i have a much bigger pressure washer than ben krasnow, should try building a water jet cutter
<azonenberg> awygle: me neither
<azonenberg> but i know a bunch of machinists :p
<gruetzkopf> (the kind that needs a 32A 400V threephase socket PLUS (optionally) fuel for the heater)
<azonenberg> lolr
<azonenberg> oh, THAT kind of pressure washer
<azonenberg> what's it for, cleaning aircraft runways? :p
<gruetzkopf> no idea, looted it :D
<gruetzkopf> (cleaning food processing rooms)
<gruetzkopf> near-boiling water at that flow rate and pressure kills many many things.
<azonenberg> Lol yes
steakpizza has joined ##openfpga
steakpizza has quit [Ping timeout: 240 seconds]
bitd has joined ##openfpga
futarisIRCcloud has quit [Quit: Connection closed for inactivity]
pie_ has quit [Ping timeout: 255 seconds]
steakpizza has joined ##openfpga
steakpizza has quit [Ping timeout: 248 seconds]
indy has quit [Quit: ZNC - http://znc.sourceforge.net]
indy has joined ##openfpga
pie_ has joined ##openfpga
SpaceCoaster has quit [Ping timeout: 265 seconds]
rohitksingh_work has quit [Read error: Connection reset by peer]
mnr has joined ##openfpga
<implr> https://ripe76.ripe.net/presentations/139-139-Turris-20180517-OF-MOX-RIPE.pdf pages 7 and 10: interesting approach to modular network hardware with the pci connector stack
Bike has joined ##openfpga
scrts has quit [Ping timeout: 260 seconds]
rohitksingh has joined ##openfpga
rohitksingh1 has joined ##openfpga
rohitksingh has quit [Ping timeout: 264 seconds]
genii has joined ##openfpga
scrts has joined ##openfpga
scrts has quit [Ping timeout: 255 seconds]
<shapr> gruetzkopf: I *really* want to see pix of that pressure washer
indy has quit [Quit: ZNC - http://znc.sourceforge.net]
steakpizza has joined ##openfpga
indy has joined ##openfpga
steakpizza has quit [Ping timeout: 240 seconds]
SpaceCoaster has joined ##openfpga
<pie_> interesting
<pie_> @ implr
SpaceCoaster has quit [Ping timeout: 256 seconds]
SpaceCoaster has joined ##openfpga
SpaceCoaster has quit [Ping timeout: 255 seconds]
steakpizza has joined ##openfpga
steakpizza has quit [Remote host closed the connection]
steakpizza has joined ##openfpga
scrts has joined ##openfpga
<awygle> i just got an ad for a men's health magazine on twitter, and before the _ad_ ran there was _another ad_ for the Little Witch Academia video game.
<Bike> ads are lucrative. it makes perfect sense for anyone airing video to sell ads. ads are a video, therefore
ironsteel has joined ##openfpga
<rqou> very recently I've been getting a _ton_ of ads on birbsite
<rqou> all for relatively mundane "everyday" items
<rqou> i suspect somebody scraped the fact that i graduated
ironsteel has quit [Ping timeout: 265 seconds]
<azonenberg> rqou: lol
<azonenberg> i just block all ads indscriminately
<azonenberg> they can data mine whatever they want but it wont do them any good
<rqou> yeah me too
<rqou> still not as creepy as iirc Target and their pregnancy detection ML
<azonenberg> lol
<azonenberg> yes i remember that story
<zkms> i only block uninteresting/gross/boring ads
<zkms> frankly, the bird site should be paying *me*, given how i've managed to refine their ad targeting
<azonenberg> zkms: I consider any attempt to influence my behavior, other than by providing useful, unbiased, factual information
<azonenberg> to be highly offensive
<Ultrasauce> if anyone's getting paid for my attention it's gonna be me
<qu1j0t3> anyone know of MCUs with UARTS configurable as wide as 15 bits?
<azonenberg> qu1j0t3: a) why would you need that? b) sounds like a job for... FPGA-Man!
<qu1j0t3> azonenberg: :)
<qu1j0t3> azonenberg: short answer, HP-HIL
<azonenberg> I've thankfully never heard of it
<azonenberg> :p
<qu1j0t3> and yes it does
<kc8apf> HP-HIL sounds terrible
<kc8apf> https://github.com/drom/bitfield is fantastic for documentation
<kc8apf> input format is a bit limited
<kc8apf> but the output is excellent
<rqou> oh btw azonenberg
<rqou> would it make sense for me to pay you a visit over memorial day weekend?
<rqou> (since I'm burned out on Fanime and am not going)
<rqou> or is everything on your end way too much of a mess right now?
SpaceCoaster has joined ##openfpga
steakpizza has quit [Remote host closed the connection]
<shapr> azonenberg: is cash useful information for biasing your behavior?
<pie_> <azonenberg> zkms: I consider any attempt to influence my behavior, other than by providing useful, unbiased, factual information
<pie_> <azonenberg> to be highly offensive
<whitequark> is there even such a thing as "unbiased information"
<zkms> i've blocked enough promoted accounts on twittre such that the ads i now get are for stuff like ice40 devboards, industrial gear oils, weird PCB materials, and R&S oscilloscopes and really i don't mind it that much
m_w has quit [Quit: leaving]
<lain> I don't spend enough time on twitter to train its adbots, so I just block the ads outright. When I don't, it's stuff like sportball which I have zero interest in, and that should be apparent from who I follow
<lain> >exclusively follow electronics and security people
<lain> >get ads for popular things that I show no obvious interest in
<lain> it's clear the ad system doesn't even *try* until you start blocking things :/
<lain> if they bothered to make it even a little bit intelligent it might actually make them money :P
<lain> </rant>
<Ultrasauce> oh is it time to waffle about epistemology? i'm in
rohitksingh1 has quit [Quit: Leaving.]
<Bike> no way, go lower level. if i poke you on the shoulder in order to draw your attention am i influencing your behavior? does the poke constitute information? what if i poke you on the right shoulder then appear to your left to trick you
<Ultrasauce> what about constructing that hypothetical situation
<Bike> what if someone pokes you on the shoulder and you're aware that you're a character in a hypothetical situation
<Bike> hm... If you were informed you were in a hypothetical situation, what would you do
<Ultrasauce> I guess it depends on whose hypothesis
<Ultrasauce> the nature of its construction and its purpose
<Ultrasauce> I guess a thought complex enough to contain our entire universe might be a bit beyond our comprehension though
<Bike> what if you were falsely informed you were part of a hypothetical situation
<Ultrasauce> trying to determine the above would be even more of a losing battle in that case
<Ultrasauce> hofstadter has a pretty decent argument for the universe being a mathematical construct
<Ultrasauce> not that I'm necessarily on board with that line of thinking....the whole point of math is to be capable of representing the universe after all
<Ultrasauce> no not hofstadter, schmidhuber. weird pair of people to mix up
<Bike> weird germanic names
<Ultrasauce> actually I guess a genesis-thought could potentially be quite simple from a reductionist viewpoint
<Ultrasauce> perhaps all complexity is a consequence of our limited perception
<Ultrasauce> what if the hypothetical situation in which we reside was created for the purpose of deception?
* shapr checks the channel
<shapr> man, I thought this was ##dependent for a sec
<shapr> oh HEY! did you see bunnie's NeTV2 ?
<rqou> yeah, imo not that interesting
<rqou> well, the political aspects are interesting
<rqou> the hardware less so
<pie_> shapr, heh
<pie_> CONVERGENCE
<pie_> Ultrasauce, i was like wait are you sure its not hofstadter? then i realized i was thinking of tegmark
<Ultrasauce> nah this is what i was thinking of https://www.youtube.com/watch?v=T1Ogwa76yQo
<Ultrasauce> trigger warning: tedx
<balrog> someone wrote an hdcp stripper firmware for the original NeTV and posted it on Tor
<balrog> (specifically -- whoever worked on the Fairplay DRM bypass)
<rqou> yeah of course
<rqou> drm is teh dumb
<rqou> offtopic: the accessible xbox controller is interesting
<rqou> I'm a bit surprised the "anti-cheat" people okayed it
<Ultrasauce> a generic 'button' port is a pretty neat addition
<Ultrasauce> will hopefully cultivate some interesting diy efforts
<rqou> historically a number of vidya i used to play would _explicitly_ block a11y solutions
<rqou> so that macros couldn't use them to inject inputs
<Ultrasauce> well and there's the whole peripheral licensing as a revenue stream thing
<rqou> oh yeah, that too
<rqou> those used to both override a11y concerns
<rqou> hmm, i think server-side "machine learning" anti-cheat is what is making the old bullshit unnecessary
SpaceCoaster has quit [Ping timeout: 256 seconds]
shuffle2 has left ##openfpga ["never coming back"]
<awygle> NeTV2 is probably interesting to people who care about media
<awygle> i have at least one friend who's getting it
<florolf> we're looking at that for c3voc purposes
<rqou> meanwhile I'd rather have something that just ingests a ton of hdmi and spits it out over Ethernet
<rqou> so that you can escape the "weird AV world where everything you know is wrong" as fast as possible
<florolf> rqou: actually, having this be a hdmi mux and hdbaset-bridge was the idea
<rqou> no, not hdbaset
<rqou> that's still in the weird "everything you know is wrong" universe
<Ultrasauce> i dont think "encapsulating the weird av world where everything is wrong" in IP makes it any less that
<florolf> rqou: well, that was the idea that someone threw around. i don't particularly care about the encapsulation
<rqou> Ultrasauce: but all your tools like vlans, media converters, etc at least work now
<q3k> you can buy cheap chinese hdmi to 'ethernet' adapters
<rqou> yeah i have one
<q3k> which end up translating hdmi into properietary ip streaming magic
<rqou> it doesn't actually output Ethernet frames
<rqou> oh, that must be a slightly more expensive version
<q3k> no, the ones I know of (implr might know more) actually can plug into your network
<q3k> it still isn't an actual standard, just whatever they came up with that they could implement cheaply
<rqou> yeah, see, Chinese engineers also get it
<rqou> "AV engineers" seem to all be either smoking something or just love weird special NIH
<q3k> you also need to remember the requirements they have (low latency, low jittler, synchronization) are difficult/impossible to implement on an IP network
<rqou> those are all possible if you're good?
<rqou> e.g. synchronous Ethernet+1588 is really cool
<q3k> it's doable on a point-to-point ethernet link, but the moment you connect that to your shitty QoSless ip backhaul you're fucked
<rqou> oh yeah of course
<q3k> also switch and router buffers will fuck things up, too
<G33KatWork> reminds me of EtherCAT
<rqou> not BRCM (if you configure it correctly)
<florolf> at least that's the one i've seen in a couple of similar devices
<G33KatWork> some german PLC company (Beckhoff) does IP in their PLCs and they seem to be able to achieve realtime performance
<q3k> florolf: ask inf (on freenode) or implr (here)
<q3k> G33KatWork: i know the robot arm we have at hswaw also does ethernet but will basically e-stop if it doesn't get an ack within a few hundred us or so
<q3k> G33KatWork: which is all cool when you have a dedicated l1/l2 for it, but falls apart otherwise
<rqou> yeah these all sound like usable solutions that don't suck too much
<rqou> unlike AV's "let's fuck up our scrambler and then force people to design AGC to work around it"
<q3k> i think that's what we use yes
<G33KatWork> q3k: I think Beckhoff does most of this stuff in hardware. There also seems to be software support for EtherCAT for the PRUs on a beaglebone black
<florolf> G33KatWork: yeah, that looks familiar
<q3k> speaking of networks, my laptop now speaks bgp and lives at 185.236.243.2
<Ultrasauce> my thing uses h265
<q3k> (well, I cheated a bit, and I aggregate the entire /24 on the edge, but eh)
<Ultrasauce> turns out realtime interframe codecs are hard
<G33KatWork> > 22/tcp open ssh
<G33KatWork> OHAI
<q3k> G33KatWork: hack me bro
<G33KatWork> > Received disconnect from 185.236.243.2 port 22:2: Too many authentication failures
<G33KatWork> hunter2 didn't work
SpaceCoaster has joined ##openfpga
<q3k> G33KatWork: i actually thought about this and have a whitelist-based pf ruleset
SpaceCoaster has quit [Ping timeout: 265 seconds]
<implr> that robot is devicenet, not ethercat iirc
SpaceCoaster has joined ##openfpga
<gruetzkopf> i have beckhoff stuff around
<q3k> beckhoff sounds like what I would say drunkenly to someone who pissed me off
<gruetzkopf> wanted to sniff the 100M ethernet between the io cards (apparently LVDS)
<awygle> huh, microsemi wants to use ReRAM apparently
<q3k> the talos bmc apparently is an ICE40 https://git.raptorcs.com/git/talos-system-fpga/tree/main.v
<q3k> also // Thanks to Clifford Wolf for the idea and basic code!
<q3k> heh
<awygle> "nasty ring oscillator"
<q3k> what the fuck is going on in there
<q3k> oh it's running arachne with a different -m
<q3k> then comparing the timing between them
<q3k> and choosing the best one
<q3k> jesus christ how horrifying
<q3k> oh no, with a different -s (seed)
<daveshah> I've thought about something like that for a while
<daveshah> As a quick way to add multiprocessing to arachne
<awygle> but it's only using seeds 1 through 64?
<q3k> and it's hand-roller, hard-coded
<q3k> *rolled
<q3k> daveshah: i mena, thanks to the power of make, this is basically parallelized
<q3k> daveshah: if you -j $(nproc)
<awygle> like i get running it with 64 _random_ seeds i guess
<awygle> but you could run 1 through 64 offline and just pick the best one
<daveshah> By the best one will change for each design change
<daveshah> Significantly
<G33KatWork> > $4,925.00
<G33KatWork> a bargain
<daveshah> There is effectively no difference between 64 random seeds and 1-64
<q3k> G33KatWork: there's a new cheapo preorder talos
<q3k> G33KatWork: for $1k5 or so, apparently
<awygle> well, yes, i guess if you're still working on the thing
<q3k> G33KatWork: single socket
<q3k> G33KatWork: still BYOCPU
<G33KatWork> 0/10 doesn't run crysis
<daveshah> The best seed will even change with yosys and arachne version
<G33KatWork> how close is POWER9 to PowerPC?
<q3k> G33KatWork: it still has 'eieio', and that's the only thing that matters
<q3k> also 'darn'
<G33KatWork> ah, ppc mnemonics <3
<G33KatWork> rldicr
<balrog> where does GP Ultralite fit in that?
<G33KatWork> > Rotate Left Double Word Immediate then Clear Right
<jn__> balrog: what's that?
<balrog> jn__: G5
<oeuf> meow?
<G33KatWork> rofl
<oeuf> ah, dwords not double-precision floating point, nevermind
<kc8apf> q3k: Talos FPGA is just doing reset and power sequencing
<kc8apf> there's still a BMC
<awygle> what's a bmc? guessing not bounded model checker
<jn__> baseboard management controller
<jn__> the thing that may or may not implement IPMI
<kc8apf> balrog: G5 was PPC970 which was POWER4+Altivec
<kc8apf> POWER is pretty close but not quite PowerPC
<q3k> kc8apf: yeah, didn't know how to call it if not 'bmc'
<qu1j0t3> British Motor Company
<kc8apf> Talos II is based on an IBM reference design. Uses Aspeed 2500 BMC.
<kc8apf> q3k: miscellaneous logic bits
<kc8apf> Intel tends to put down CPLDs for similar uses
<q3k> kc8apf: if it's glue logic that looks at buses, let's call it "sniffing glue"
<kc8apf> hahaha
<kc8apf> Zaius uses PMICs instead
<kc8apf> I'd rather have an ARM64 workstation than POWER9.
<kc8apf> I probably have a better chance of getting a RISC-V server chip than an ARM64 server chip though
<rqou> heh issues with BRCM and LEDs lolol
<q3k> i'd rather have a risc-v workstation than {arm64,power9}
<awygle> i just want a billion-core system
<awygle> like certain people claerly already have
<kc8apf> awygle: no comment
<q3k> awygle: join google, run things on borg, ????, profit
<rqou> idk if i said this (or if it's "supposed to be" public or not) but BRCM switches (the fancy ones) have a tiny microcontroller just for the LEDs
* awygle was actually sub-irc-ing azonenberg
<kc8apf> some of the BRCM switches have a ARM core in them for the OEM's software stack
<awygle> i'll settle for six cores for now i guess
<jn__> ^ ARM64 desktop/workstation stuff
<balrog> kc8apf: the Intel X557 PHY has an MCU core
<balrog> and external firmware
<rqou> yeah all the new fancy switches have a cortex-a
<awygle> only 9500 pounds!
<rqou> since using freescale ppcs apparently sucked
<kc8apf> ThunderX for 1350 pounds is kinda reasonable
<rqou> bonus effective use of silicon if you strap the cortex-a off :P
<awygle> that desktop is actually pretty interesting
<rqou> (this is intended for multi-asic-in-a-switch-box use cases)
<awygle> 32 core, and lots of networking
<awygle> ECC DDR4
<kc8apf> hmm, Avantek's workstation model is Gigabyte's ODM design
* oeuf pokes q3k with a configuration language
<q3k> oeuf: small brain: bcl; normal brain: gcl; cosmic brain: https://jsonnet.org/
<whitequark> oh god that thing
<awygle> "jsonnet" sounds like a java port of a 2.5D EM solver
<whitequark> rqou: lol @ led microcontroller
<whitequark> they couldn't do it in the ASIC?
<oeuf> q3k: what about n
<oeuf> >_>
<q3k> oeuf: multiverse brain: generating gcl from jsonnet from sis*
<q3k> oeuf: never used it, actually
<rqou> whitequark: afaik too many customers annoyed them with too many different requests for how LEDs should work
<q3k> oeuf: oh no, n* was used in ra*
<q3k> oeuf: that was horrible and weird and not fun and I remember it now
<rqou> so now it becomes "program whatever LED pattern you want"
<oeuf> q3k: okay I'm having trouble expanding that glob now
<oeuf> (tbh my day-to-day insanity is looking at ICU so I'm not familiar with the configuration languages anyway)
<q3k> oeuf: the build system
<oeuf> ah.
* oeuf pets
* implr is triggered by n*
<implr> I wrote approximately 15 lines in it, once. Took 2 weeks.
<pie_> n?
<rqou> lol
<rqou> i told you, LEDs are hard :P
<florolf> rqou: i recently realized the 53128 stupid roboswitch has a 8051 in it, which the the ROM uses to run a repl, but no interface seems to be bonded out
<rqou> i believe it
<florolf> also, green ethernet mumble mumble
<rqou> i don't know much about robo though
digshadow has quit [Ping timeout: 264 seconds]
<florolf> the fun part is that the 8051 can also TX packets
<florolf> the sad part is that it's an 8051
<florolf> with some weird-ass banking scheme
<awygle> lulz
digshadow has joined ##openfpga
<awygle> Fun With LEDs
<awygle> but no // duty cycle 4% to not blind customers
Bike has quit [Ping timeout: 260 seconds]
digshadow has left ##openfpga [##openfpga]
digshadow has joined ##openfpga
digshadow has quit [Client Quit]
digshadow has joined ##openfpga
digshadow has quit [Client Quit]
digshadow has joined ##openfpga
Bike has joined ##openfpga
<whitequark> florolf: is it MPAGE?
<whitequark> MPAGE/XPAGE/etc
<whitequark> that's actually not too bad, it's more or less the same as what 8086 does
<whitequark> compared to the insane scheme 8-bit PICs use
<lain> "Compressing a 15-minute 4K video takes ~7.5 hours" on what hardware lol
<florolf> whitequark: no, bank switching happens via P2 and is configurable via MMIO which is mapped into XRAM
<lain> oh ok that was single-threaded vp9 enc or some such
<lain> awygle: that's pretty neat
<whitequark> florolf: yes that's XPAGE
<awygle> lain: yeah i thought so. wish it was cloud-agnostic tho.
<whitequark> different name, same mechanism
<whitequark> lmk if you have any issues with it and e.g. sdcc, i had to figure out that shit recently
<awygle> luv 2 build amazon into critical development loops
<florolf> the weirdness is not in the switching itself, but in the configuration. i don't recall the details though, has been a couple of months since i last looked into it
<florolf> but thanks
<lain> awygle: lol yeahhhh
<kc8apf> awygle: I've shifted focus from argo to airflow for running batch jobs
<kc8apf> seems much more broadly supported
<awygle> mk
SpaceCoaster has quit [Ping timeout: 256 seconds]
<awygle> serverless (the framework), while terribly named, seems like a nice abstraction over FaaS platforms
<awygle> and openfaas seems like a nice on-prem option
<awygle> said the ignoramus
bitd has quit [Remote host closed the connection]
<q3k> serverless is basically going back to CGI
<q3k> except the interface is a bit less terrible
futarisIRCcloud has joined ##openfpga
benreynw1r has quit [Remote host closed the connection]
benreynwar has joined ##openfpga
<pie_> peanut butter and sandwitch sandwitch
<pie_> sandwich
m_w has joined ##openfpga