Bike has joined ##openfpga
scrts has quit [Ping timeout: 250 seconds]
scrts has joined ##openfpga
Bob_Dole has quit [Ping timeout: 240 seconds]
<sorear> o/ duck2 mwk
scrts has quit [Ping timeout: 245 seconds]
Bob_Dole has joined ##openfpga
Zylellrenfar has quit [Quit: Bye bye~]
Zylellrenfar has joined ##openfpga
scrts has joined ##openfpga
emeb_mac has joined ##openfpga
ZipCPU has quit [Ping timeout: 258 seconds]
ZipCPU has joined ##openfpga
gsi__ has joined ##openfpga
gsi_ has quit [Ping timeout: 250 seconds]
<TD-Linux> cool getting there. there are still memtest failures, but less. looks like a bad address line
<sorear> what are you making ?
<TD-Linux> ulx3s
cr1901_modern1 has joined ##openfpga
cr1901_modern has quit [Ping timeout: 244 seconds]
<TD-Linux> Memtest OK
<TD-Linux> so 2 problems together: bad original reflow, + bad memtest
<TD-Linux> I think the reason the original reflow failed was because I had it too close to the front of the oven for a good video, and it didn't quite completely reflow
<TD-Linux> I put it in for a 2nd cycle, but the 1st cycle already burned off the no clean flux so it wasn't a good fix
cr1901_modern1 has quit [Quit: Leaving.]
cr1901_modern has joined ##openfpga
unixb0y has quit [Ping timeout: 245 seconds]
unixb0y has joined ##openfpga
rohitksingh has joined ##openfpga
Flea86 has quit [Quit: Goodbye and thanks for all the dirty sand ;-)]
<hl> ah
rohitksingh has quit [Remote host closed the connection]
rohitksingh_work has joined ##openfpga
mepholic has joined ##openfpga
<mepholic> hello
<hl> 'lo
pie___ has joined ##openfpga
pie__ has quit [Ping timeout: 245 seconds]
Bike has quit [Quit: Lost terminal]
genii has quit [Remote host closed the connection]
emeb has left ##openfpga [##openfpga]
ayjay_t has quit [Read error: Connection reset by peer]
ayjay_t has joined ##openfpga
futarisIRCcloud has quit [Quit: Connection closed for inactivity]
<RaYmAn> daveshah: I think should be ready for merge now tbh - unless you have more review things I should clean up :)
wren6991_ has quit [Ping timeout: 256 seconds]
emeb_mac has quit [Ping timeout: 252 seconds]
m4ssi has joined ##openfpga
Flea86 has joined ##openfpga
<Flea86> Turns out I was having some external DRAM problems of my own... which turned out to be bad soldering (and very hard to observe even under magnification!)
<Flea86> Either that, or my eyesight is failing :p
wren6991 has joined ##openfpga
<whitequark> 22:20 < cyrozap> TD-Linux: I guess my point is we can have all the ideological purity in the world and still never produce a standard that
<whitequark> people will actually use. The standard that exists and is in widespread use today is infinitely superior to the one that is
<whitequark> ideologically pure but either doesn't exist or is not in use.
<whitequark> this is of course not true
<whitequark> thunderbolt 3 did not exist a few years ago at all, it is a completely new thing
<whitequark> so is usb-c
<whitequark> the modus operandi of modern tech is introduce a new shitty standard every two years and have everyone furiously upgrade to it, then deprecate it back
<whitequark> so i don't buy this argument at all.
<whitequark> again
<whitequark> before engaging in usb/tbt wankery go read the specs. then design some devices. then you gain the right to speak about them.
<whitequark> thunderbolt 2 effectively doesn't exist anymore. you can't really use it. like you theoretically have adapters, but they just... don't work
<whitequark> that's a few more years back.
<whitequark> can we get rid of thunderbolt 3 just like that? of course we can
<TD-Linux> kinda want to do a 2nd board now just so I can see if I can make one boot rework-free
<sorear> I still don’t understand your proposal for plug and play devices over Ethernet framing
wren6991 has quit [Ping timeout: 256 seconds]
<whitequark> sorear: which part?
<sorear> The last time this came up on Twitter you advocated *BASE-CX with PCIe compatible levels and termination
<sorear> This is obviously fine for networking but less so for e.g. a mouse or camera
* sorear ponders DNS-SD enabled disk drives
<whitequark> sorear: mouse: yes, you need a lower power mode, of course
<whitequark> why not a camera?
<whitequark> or a disk drive, yes
<whitequark> you're having a camera on a dedicated "ethernet" controller with predictable latency
<whitequark> people definitely put cameras in this configuration in settings where cost isn't an issue. people put cnc mills in this configuration.
<sorear> mostly because I remember upnp being awful
<whitequark> oh, i wasn't clear enough then
<whitequark> i'm not suggesting we reuse anything above ipv6/ethernet
<whitequark> it's all shit
<whitequark> and it's not really generic enough to replace usb
<whitequark> you'd need a new layer comparable to usb control pipe, for sure
<sorear> also if you plug an Ethernet keyboard into a computer with no input devices, how do you configure the link
<sorear> (not to say the status quo of “immediately trust anything exposing a HID endpoint” is entirely satisfactory)
<whitequark> it only uses link local addresses, which require no configuration
<whitequark> like making the devices on this bus routable is not in scope of my suggestion
<whitequark> you could add that, but i don't suggest to.
<sorear> existing network based peripherals all require a pairing dance. PCIe and USB don’t, which is why I see the latter two as substitutable but not the first.
<whitequark> i never suggested making existing network based peripherals exposed via this scheme
<whitequark> take another look at the thread, it went like this
<whitequark> "i think we need a completely new bus, everything from scratch, to replace usb"
<whitequark> "in fact you could reuse ipv6 and ethernet on top of parts of pcie physical layer to save yourself a lot of work"
<whitequark> whatever specification this ends up as, will have specific strict requirements for configuration mechanism (one that is mandatory and legal) and for the plugnplay discovery layer (one that is mandatory and legal)
<whitequark> this is a vast improvement over the other suggestion there, which starts with "let's throw out everything we know about SERDES and use manchester encoding"
<sorear> I read that in a context of “wait, you don’t actually like TLPs and BARs?”
<whitequark> my suggestion would fit very nicely and easily into every existing OS stack
<whitequark> worst case you just shoehorn it into your ordinary network stuff
<whitequark> TLPs and BARs, not so much, they are not really plug and play
<whitequark> try making a device with a 4 GB BAR and plug it into a PC, possibly with Windows if you want a hard mode
<whitequark> worse yet, that still requires binary drivers for every OS
<whitequark> my suggestion trivially allows userspace drivers using existing APIs on every OS
<whitequark> you only need one driver to set up the link layer
<whitequark> of course you *could* do in-kernel drivers as well, submit the skbs or whatever
<sorear> windows lacks any equivalent to vfio?
<whitequark> if you have a vendor specific PCIe device you need a kernel driver signed by Microsoft, yes
<whitequark> there are some exceptions, i think graphics drivers were somehow moved to userland, mostly
<whitequark> WDDM is mostly userspace
<whitequark> ... because GPU vendors can't make their code not crash and Windows ends up having to restart it a lot
<sorear> does retail Windows do VT-x ootb yet
<whitequark> VT-x is subject to market segmentation
<whitequark> you can't rely on it
<whitequark> there are laptops with VT-x and VT-d arbitrarily fused off or missing in firmware
<whitequark> or just broken
<whitequark> though less so these days
<sorear> We’re already assuming new hardware though
<whitequark> not really
<whitequark> you could take an ivy bridge laptop and add one mux
<whitequark> and change cpu straps
<sorear> I’d like to avoid inventing yet another autoconfiguration protocol and rewriting every non-printer driver
<whitequark> usb storage is scsi over usb, so wqsb storage would be scsi over ethernet
<whitequark> which i think we already have
<whitequark> that being said
<sorear> yes. idk how zeroconf it is
<sorear> the only device class I’ve ever seen successfully zeroconf over IP is printers
<whitequark> i don't actually know if it makes more sense to reuse or reinvent zeroconf
<sorear> (personally)
<sorear> I did not realize zeroconf was a proper noun and was not using it as such.
<whitequark> oh
<whitequark> well yes, zeroconf is an existing thing
<whitequark> apple is heavily invested in it
<whitequark> the linux version is called avahi
<whitequark> i have never used it successfully but it might or might not be the fault of the protocol
duck2 has quit [Ping timeout: 256 seconds]
<whitequark> as for rewriting every driver
<whitequark> having to redo UAC and UVC from scratch is a *feature*.
<whitequark> others not so much, i mean there's nothing wrong afaict with reusing HID descriptors and sending the reports over IP instead
<whitequark> but yes, if you want to remain 100% compatible with usb you have to use usb
<sorear> my mininum-sw-effort proposal for purging the world of SE0 is xhci-over-sata express
<whitequark> or you could just say that USB 1/2 is no longer required
<whitequark> i don't think that really achieves much
<whitequark> like yes, 1/2 transceivers are a shit design, but they don't result in that much problems in real life
<sorear> but it sounds like you have something much more ambitious in mind than “kill the USB electrical interface” or “manually configured IP with bespoke protobuf drivers for everything is fine actually”
<whitequark> correct
futarisIRCcloud has joined ##openfpga
utzig has left ##openfpga [##openfpga]
_whitelogger has joined ##openfpga
zem has quit [Ping timeout: 250 seconds]
<tnt> Huh, does usb have some restriction on maxpacket size being power of 2 or something ?
zem has joined ##openfpga
rohitksingh_work has quit [Read error: Connection reset by peer]
<_whitenotifier> [Boneless-CPU] zignig synchronize pull request #3: Text Assembler for boneless (WIP) -
<whitequark> it's worse than that
<whitequark> there's a fairly arbitrary set of values you can set maxpacketsize to
<whitequark> sometimes it's exactly one value
rohitksingh has joined ##openfpga
tlwoerner has joined ##openfpga
duck2 has joined ##openfpga
<tnt> whitequark: :/ I don't really see where those are in the specs. Unless you're meant to take table 5-4 values as the only possible values rather than examples of what bandwidht can be achieved.
wbraun has quit [Quit: Connection closed for inactivity]
Flea86 has quit [Quit: Goodbye and thanks for all the dirty sand ;-)]
rohitksingh has quit [Ping timeout: 245 seconds]
futarisIRCcloud has quit [Quit: Connection closed for inactivity]
rohitksingh has joined ##openfpga
emeb has joined ##openfpga
genii has joined ##openfpga
rohitksingh has quit [Ping timeout: 245 seconds]
rohitksingh has joined ##openfpga
rohitksingh has quit [Ping timeout: 252 seconds]
mossmann has quit [Read error: Connection reset by peer]
mossmann has joined ##openfpga
ZipCPU|Alt has joined ##openfpga
ZipCPU|Alt has quit [Remote host closed the connection]
rohitksingh has joined ##openfpga
ayjay_t has quit [Read error: Connection reset by peer]
ayjay_t has joined ##openfpga
wbraun has joined ##openfpga
m4ssi has quit [Remote host closed the connection]
Asu has joined ##openfpga
mearon has joined ##openfpga
rohitksingh has quit [Remote host closed the connection]
<tnt> Anyone knows what a device is supposed to do for an isochronous IN if it doesn't have any data ? Zero-length packet ?
<tnt> apparently yes.
<whitequark> yes
q3k has quit [Ping timeout: 240 seconds]
implr has quit [Ping timeout: 245 seconds]
<balrog> > Key characteristics of the USB4 solution include:
<balrog> > [...] Backward compatibility with USB 3.2, USB 2.0 and Thunderbolt 3
<balrog> whitequark: sounds like that will be a shitshow then
q3k has joined ##openfpga
implr has joined ##openfpga
<fseidel> oh jeez, that seems like a complete mess
<fseidel> nice for the end-user, but horrible for the implementer
<cr1901_modern> ^At least that means that any microcontroller project I have using USB will still keep working :P. B/c that's clearly the most important thing.
<fseidel> also, congrats to rqou on becoming twitter famous, I just saw him mentioned in a thing about password reuse
<cr1901_modern> I've never even gotten a tweet past 40 RTs T_T
<cr1901_modern> Also thank you rqou for this, considering the amount of shitheads it brought into your mentions:
<cr1901_modern> ... lots and lots of shitheads ._.
<fseidel> anyway, now I know I should change my password :-P
* cr1901_modern grabs another vaudeville hook for fseidel
<cr1901_modern> how many of these do I have to keep around?
<fseidel> more than you have space for
<cr1901_modern> mmmm
<fseidel> on a more on-topic note, any recommendations for a BLIF parser in Python?
<fseidel> I'm trying to consume stuff that VPR spits out and have found a few options, not sure if anyone had one they think is better than the others or something
<ylamarre> Can't yosys be used as a translator?
<ylamarre> BLIF to json?
<cr1901_modern> read_blif should be available.
<fseidel> oh interesting, I've just been using Yosys to spit out BLIFs but I didn't think about trying the other way around
<cr1901_modern> (read/write command formats in yosys are _not_ symmetric)
<fseidel> thanks!
<ylamarre> I give no garantee, I was asking if the option had been considered...
<fseidel> ah okay
<fseidel> I'll check the docs
<fseidel> looks like it can do it :-D
asante has quit [Ping timeout: 250 seconds]
asante has joined ##openfpga
asante has joined ##openfpga
asante has quit [Changing host]
<azonenberg> fseidel: i was cc'd on that thread from rqou
<azonenberg> i'm STILL getting spammed from it
<cr1901_modern> So why haven't you muted it?
<azonenberg> i havent tweeted in the last few days
<azonenberg> so i have no legit traffic to get lost in the noise :p
<whitequark> azonenberg: imagine what my notifications are like
<whitequark> i usually wake up to 300-1000 notifications
<whitequark> on, like, normal slow days
<azonenberg> o_O
<azonenberg> i usually have low tens on a normal/slow day
<TD-Linux> I always assumed twitter had some way to managed large numbers of notifications
<whitequark> if i tweet something really popular it can cross 2-3k
<whitequark> lmao no
<whitequark> they just dump them all into your browser
<whitequark> which becomes very slow and eventually crashes
<TD-Linux> though now that I understand the """design""" of twitter I expect less
<whitequark> they also randomly drop notifications
<azonenberg> This is why i have a dedicated VM for fb and twitter
<whitequark> usally the ones i actually wanted to see
<azonenberg> that i don't use for anything else
<azonenberg> if they get pwned, or it runs out of ram
<azonenberg> idgaf
<whitequark> you don't understanad the problem
<whitequark> i do care if twitter crashes, it means i miss some conversation
<whitequark> but there's literally no amount of ram i can give it that it would be okay with
<whitequark> because it just has shit algorithmic complexity in general
<cr1901_modern> you're so popular, whitequark :3
<whitequark> fun fact: i spent like, years, studying how to put so much pedantry into a single tweet that no man will ever reply to me with an explanation of my joke
<whitequark> it used to be really bad and i decided to take the, ahem, technical approach to a social problem
<azonenberg> whitequark: well if it makes you feel any better
<azonenberg> facebook isnt any better wrt making browsers oomkill
<whitequark> twitter is just full of nazis, facebook is full of nazis *and* they do everything to exploit me semi-competently
<whitequark> i refuse to touch facebook
<whitequark> if you're living in a country, make sure it's a decently corrupt one. if fyou're using a social network, make sure it's the one that's so incompetent they can't even exploit you
Asu has quit [Remote host closed the connection]
<whitequark> also don't take advice from me probably
<cr1901_modern> I want to delete facebook, but I lose access to some meatspace friends (... all 8 of them) I can't currently contact easily
<whitequark> solution: don't have meatspace friends
<whitequark> see above re: taking advice from me
<TD-Linux> you could move to silicon valley and then all your meatspace friends would be debating signal vs wire
<TD-Linux> but then you live in silicon valley so
<whitequark> i hve seriously considered moving to silicon valley for that reason
<whitequark> then brain broke badly enough i'm probably not moving anywhere without like years of therapy
<whitequark> oops
<TD-Linux> on an unrelated note: litex is pretty cool
<TD-Linux> I originally just wanted to steal the sdram controller but the rest is nice too. might try putting tg68k in as a soc core
<cr1901_modern> >but then you live in silicon valley so <-- you could not pay me to live there
<cr1901_modern> whitequark: And the truth is, the 8 is a global maximum. Number is prob closer to 2 now because I just don't reach out anymore
<bubble_buster> you can switch your fb account to messenger only
<cr1901_modern> bubble_buster: Really?!
<cr1901_modern> Hmmm
<bubble_buster> I've been tempted to switch because 90% of my fb usage is mindlessly scrolling through the feed for 10-15 minutes before I roll out of bed
<bubble_buster> but I feel like deactivating your fb page has strange social consequences , at least among my social network
<cr1901_modern> (cw: suicide) My feed is mostly people getting married, having children, or losing friends at young ages.
<cr1901_modern> Can't relate to the first two, and well the third one is "good" to know about, but not something I really want to see either.
<bubble_buster> like you're some fringe conspiracy theorist worried about data privacy, or you think you're better than people who use fb
<cr1901_modern> it's not a conspiracy theory tho. They are a demonstrably dangerous company that's incapable of responsibly wielding the power they have
<bubble_buster> yeah sure
Flea86 has joined ##openfpga
<bubble_buster> but I'm just saying that's how I think mainstream people might view me or others deactivating their fb page
<bubble_buster> by deactivating, you're implicitly challenging their decision to not deactivate or something
<cr1901_modern> maybe... truth is, I haven't made a post since The Last Jedi came out. Which I partially did to piss the ppl off who think Star Wars is ruined.
<cr1901_modern> No takers :P
<Flea86> Pre-CGI star wars was still the best. Okay I'll let myself out now lol
<bubble_buster> wow I haven't posted in almost 2 years
<cr1901_modern> Flea86: I mean, I like them all lol
<bubble_buster> ep 3 was the best
<cr1901_modern> Ep 8 took a boatload of risks and I think in general paid off. And Force Projection has precedent damnit!
<cr1901_modern> shut up!
<Flea86> Speaking of star wars, anyone see that early CGI footage (~1977) of trio of x-wing fighters doing some formation flying? That actually looked pretty darn good, but can only imagine what it cost to actually make...
<cr1901_modern> Flea86: I have not, but I am absolutely a sucker for early CGI. Including "A Computer Animated Hand", Golgo 13 movie, and Lensman movie
<Flea86> Same
<cr1901_modern> (Like I can't believe "A Computer Animated Hand" was 19*72*!!)
<Flea86> Seems this is all I can find right now about that early x-wing footage
<Flea86> cr1901_modern: Yeah, pretty mindblowing stuff when you consider the hardware they had to work with at the time
<unixb0y> hi
<cr1901_modern> That footage is pretty damn good for 1978...
<Flea86> Indeed. Reminds me of "The Last Starfighter" level quality
<cr1901_modern> Flea86: Definitely not in 1972, but I recall there was some custom gfx hardware (bit slice) in the late 70s early 80s
<cr1901_modern> I thought Pixar used it for Wally B., but it appears not
<cr1901_modern> hmmm that's gonna bug me that I don't remember
emeb has quit [Quit: Leaving.]
<Flea86> cr1901_modern: Evans and Sutherland made the custom hardware IIRC
Bike has joined ##openfpga