<pointfree>
The HC switching matrix is larger and looks more arbitrary. I think its structure depends more on whatever odds-and-ends are inside the UDB's, so that's the direction routing formulas should come from for that if they are to be concise.
<pointfree>
Also a function to map to HV bytes is needed. Pretty much there with that too.
<pointfree>
cyrozap, it looks like you have more knowledge of those odds-and-ends inside the UDB. I have only synthesized to the PLD's, nothing else.
<pointfree>
It's now straightforward to compute the vseg mask coming out of the top of the HV switching matrix. I should write another tiny word/function for that.
<azonenberg>
rqou: i may be interested once i do the next spin
<azonenberg>
current board design i *know* is too slow
<rqou>
interested in testing or pwning my scope? :P
<rqou>
it runs vxworks btw
<rqou>
updating the firmware was immensely frustrating
<rqou>
it requires four floppy disks
<azonenberg>
in having you test my board :p
<rqou>
oh yeah btw my scope also has the "TDS3ENG" module :P
<azonenberg>
which is?
<rqou>
enables all the extra module features
<rqou>
the ENGineering module :P
<rqou>
i reflashed the TDS3FFT module because the latest firmware has FFT functionality without requiring the module
<azonenberg>
ooh fun
<rqou>
yeah, it has really "useful" features like telecom ds0/ds3 compliance testing
<rqou>
or triggering on composite video lines/fields
<azonenberg>
my rigol can do the latter
<azonenberg>
lol
<rqou>
yeah it doesn't work very well with arcade/retro 240p signals
<azonenberg>
Or DP/HDMI/DVI? :p
<rqou>
yeah, that doesn't work either
<rqou>
iirc it can trigger on SDI if you for some reason are using that
<azonenberg>
But not HD-SDI?
<rqou>
hmm apparently sdi needs a different module
<rqou>
and no hd-sdi
<rqou>
apparently analog 1080i works
<rqou>
if you're weird and want to use interlaced hd video for some strange reason
<rqou>
imho all "video" people are weird/insane
<azonenberg>
interlacing is stupid
<azonenberg>
its a relic from a bygone era
<azonenberg>
we have the bandwidth to do progressive scan 60fps or more
<azonenberg>
just do it
<rqou>
afaik hdmi still allows interlacing
<rqou>
dp might not
<azonenberg>
yes it allows it
<azonenberg>
but what is the point?
<azonenberg>
its a hack to avoid jittering when you dont have the bandwidth to do progressive scan
<rqou>
you need it in order to carry legacy 480i video that you e.g. digitized from a vhs tape :P
<azonenberg>
so 1080i60 looks less jittery than 1080p30 and uses the same bandwidth as 720p60, but is sharper in stills
<azonenberg>
lol
<azonenberg>
you could deinterlace and render progressive scan
<azonenberg>
like it would if you played that same video in a media player on a PC etc
<rqou>
i mean, you can also send h.264/h.265/vp8/vp9 over ethernet :P
<azonenberg>
Indeed
<rqou>
oh iirc h.265/vp9 no longer allow interlacing
<lain>
man
<lain>
every now and then I'll walk past my parents watching some tv on the big screen
<lain>
and I'll notice it looks really... weird. I dunno how to describe it, almost like there's an afterimage
<lain>
I'm 99% sure it's the tv's crappy deinterlacing
<lain>
some stuff is apparently still broadcast interlaced? I dunno
<rqou>
afaik in the US most OTA broadcasts are 1080i60
<rqou>
at least interlacing is not as dumb as overscan
<lain>
this is cable (well, iptv -> hdmi)
<lain>
but yeah
<lain>
most stations look great, but some have this really weird afterimage thing going on and I have no idea what causes it
<lain>
I don't watch tv on the actual tv, and my parents don't notice it, so I haven't investigated further :P
<rqou>
i thought cable/iptv is also famously overcompressed mush?
<rqou>
because of the "gotta have MOAR channels!" thing
<lain>
haha
<lain>
I haven't noticed it looking particularly bad, but I haven't really watched it
<rqou>
yeah i don't watch tv ever
<lain>
they did allocate a lot more bandwidth than our internet plan gives us, with some apparently reserved for phone and iptv
<lain>
yeah, I watch a variety of shows... in vlc.
<rqou>
idk why you need phone
<lain>
we've had the landline since before I was born
<lain>
it's now ip-based, but, meh. parents' friends all know the landline number
<rqou>
my parents keep their landline, but it's been downgraded to the cheapest possible plan that doesn't even have long distance dialing afaik
promach has quit [Ping timeout: 276 seconds]
<rqou>
for emergencies
<lain>
yeah, my parents are mainly on cellular now
<rqou>
we're still on POTS though
<rqou>
meanwhile my apartment afaik doesn't have a landline at all
<rqou>
i've never even seen any jacks
<rqou>
if we have it as part of a bundle it's definitely not hooked up
<lain>
landlines are such fun
<lain>
I've told this story a billion times but the weirdest bug I've ever experienced on a repair call was: doing network troubleshooting at a bank, they have a separate consumer dsl line for the kiosk in the lobby (the actual teller machines and branch servers are all on their own dedicated ATM lines or other magic)
<lain>
we troubleshoot the kiosk, determine it has no interwebs, this is typical. go hunt down the modem in some dark corner or whatever. in this case it was in a cabinet behind the teller area...
<lain>
no link light
<lain>
unplug modem, plug in handset, dialtone. ok, well, that's good. probably just a bad modem, but better double-check the line.
<lain>
dial tom's phone, he's standing right next to me
<lain>
phone simultaneously plays a busy tone and rings
<mtp>
oh man
<lain>
tom answers, I can hear him, we both hear this busy tone
<lain>
hang up, dial operator, same deal: get operator, also get busy tone
<mtp>
this sounds amazing
<lain>
so we play with this a bit longer, but it's always the same story
<mtp>
someone crosswired something down at the exchange
<lain>
so we call the telco
<mtp>
?
<lain>
eventually escalate to someone who isn't reading a script
<lain>
has us run some tests
<lain>
he's totally baffled
<cyrozap>
lain: Did you call them from the line you were troubleshooting? ;)
<lain>
cyrozap: lol well, we did at some point during the testing, but the main call was from my cell :P
<lain>
after quite a while on the phone, the tech figures it out
<lain>
he explains how the exchange is all digital - there's no wiring panels anymore, it's all just entries in a database
<lain>
and somehow
<lain>
against all the safeguards
<lain>
someone managed to digitally wire up two lines to one physical line
<lain>
which of course threw the modem for a loop :P
<mtp>
NICE
<lain>
after tracking down the paperwork and figuring out which number was SUPPOSED to be on that line, all was well :P
<mtp>
i mean
<lain>
:D
<mtp>
i could see how this might happen on, like, a 3ESS
<lain>
I don't know much about how that stuff works behind the scenes tbh
<mtp>
i do :)
<lain>
:o
<azonenberg>
rqou: i have a landline that's VoIP to cable internet
<mtp>
well. not in modern telcos, except from what i've gleaned from other phreaks
<azonenberg>
and its wired point to point from the phone to the voip bridge
<mtp>
but i volunteer at the Seattle Museum of Communications
<azonenberg>
not through the house POTS wiring
<lain>
mtp: usually on those calls it's a dead modem. I LOVE (/s) explaining to first-level tech support that yes, the modem is bad, and no, we don't need to keep reading the script
<lain>
my favorite dead modem story was
<azonenberg>
Because there is something really freaky going on in the house's wiring
<azonenberg>
Nobody has been able to diagnose this one
<lain>
get there, no internet, check modem - it has link light. we unplug the phone line, wait 5 minutes, still have link light.
<lain>
we power cycle modem, phone still disconnected, modem gives link light
<mtp>
nice
<azonenberg>
lol ok wow
<lain>
so we call telco "yeah modem's dead, shows link light with no line attached"
<lain>
"ok I need you to turn on your computer and navigate to our speed test paeg to check the speed"
<lain>
"NO LISTEN"
<lain>
:P
<lain>
took a solid 15 minutes to convince them it was deader than dead
<lain>
at one point I remember they kept asking "but you said the link light is on, so it is working"
<lain>
and we're like "nooooo the light is on BUT ITS UNPLUGGED"
<lain>
"but the light is on?" "yes, but..." "then it is working"
<azonenberg>
...
<lain>
poor, script-reading tier 1 support
<azonenberg>
the 1st level tech support i encounter is dumber than dumb
<lain>
the script says if the link light is on, it's linked.
<azonenberg>
the comcast field service techs usually kidna-sorta know what they're doing
<lain>
THERE CAN BE NO OTHER POSSIBILITY
<lain>
azonenberg: yeah comcast techs I've run into seem to know what's up
<lain>
like the train intefering with the line
<lain>
lol
<azonenberg>
Had a fun chat with one who was here diagnosing a bad splitter a while ago
<azonenberg>
while waiting for the modem to reboot and provision
<lain>
I'll never forget that "HOLY SHIT" from the back corner of the building as the tech saw the event finally happen in real time
<azonenberg>
turned out he was retired USMC IT
<lain>
oh nice
<azonenberg>
was telling me fun stories about running a datacenter in a 40ft container in the middle of afghanistan during a sandstorm
<rqou>
what i've heard is that comcast has really good second (and beyond) tech support
<lain>
woo
<rqou>
but first line is dumb as usual
<lain>
at&t has excellent 2nd and beyond support also
<rqou>
and business practices are, well, comcast
<azonenberg>
Lol yes
<azonenberg>
Still waiting for them to offer native static ipv6 in my area
<rqou>
native dual stack on consumer plans exists here
<azonenberg>
mtp, lain: anyway, i had a really fun bug in my telephone wiring that i never fully debugged
<azonenberg>
We moved here and were cell only for a year and a half, but kept having problems b/c we live in a black hole of cell service
<azonenberg>
tl;dr we're on a ridgeline on an island
<azonenberg>
all of the towers are on the mainland or the far side of the ridge
<azonenberg>
So we decided to get a landline
<azonenberg>
we had comcast internet so i signed up for a voip link from them
<azonenberg>
guy comes down, installs a splitter and a voip gateway (incidentally the splitter that later failed and took out my data service...)
<azonenberg>
Then opens the demarc point
<lain>
at parents house one time (at&t dsl, way back in the day) we were paying for 3mbit/s (state of the art!) and getting about 3kbps. called, escalated to a real tech, he says it's a sync issue but it's not our modem, the dslam is stuck somehow... then he asks "do you like your neighbors?" and I said ".. yeah ... whyyy?" and he says "cause the only way to fix this is to reboot the dslam, and that's gonna take
<azonenberg>
removes the patch cable going out to the street
<lain>
the whole neighborhood offline for a minute. fuck it, let's go" *lines go out* "oh shit wait, you called me on a different line, right? or are you gone..."
<lain>
(I called him on my cell, we had a good laugh)
<lain>
(and it fixed the issue!)
<azonenberg>
lol
<azonenberg>
anyway so he unplugs the cable to the street
<azonenberg>
Runs a new cable from the voip modem in the garage to the demarc point
<azonenberg>
and plugs it into the house wiring
<azonenberg>
Then i pick up the phone in the house, call my cell, it goes through
pie_ has joined ##openfpga
<azonenberg>
so i'm like "ok great it's working"
<azonenberg>
Only after the tech left did I try calling the landline *from* my cell
<azonenberg>
cell showed it ringing nonstop
<azonenberg>
Landline never made a peep
<azonenberg>
poked around a bit, wondered if ringer volume was low or something, but the handset was fine
<azonenberg>
So just for lulz i tried calling the landline from my cell again
<azonenberg>
Picked up the landline handset when i knew it was ringing
<azonenberg>
And the call connected,
<azonenberg>
with good audio quality etc
<azonenberg>
Pulled the handset off the cable, probed with a multimeter
<azonenberg>
-48V just lke there should be
<azonenberg>
then i dialed the landline
<azonenberg>
No ring waveform whatsoever
<lain>
looool
<azonenberg>
So apparently there is something really messed up on my POTS wiring
<azonenberg>
the -48V DC passes
<azonenberg>
The low voltage audio passes
<azonenberg>
low voltage AC*
<azonenberg>
but high voltage AC ring does not
<lain>
I once made the awesome mistake of grabbing a POTS line while it was ringing, and my hands were not dry
<azonenberg>
... ouch
<azonenberg>
I never figured it out, the easier option was to bypass the POTS cabling and run a cat5 line from the voip bridge directly to the handset
<lain>
luckily it was just one hand and I was otherwise floating
<lain>
:P
<azonenberg>
Now it rings fine
<lain>
huh
<lain>
that's wild
<lain>
I have no idea what would cause that
<azonenberg>
yeah me neither
<azonenberg>
AC coupled diode clamp? :p
<lain>
lol
<azonenberg>
thats about the only thing i can think of that would pass DC plus weak AC but not strong AC
<lain>
corrosion can sometimes have diode-like behavior
<azonenberg>
yeah i figured it was some weird parasitic schottky junction or something
<lain>
maybe the NSA landline tap is malfunctioning ;)
<azonenberg>
but i'm only in this house for another year or so anyway, not worth it
<azonenberg>
Doubt it, TLAs almost certainly tap VoIP at the IP layer
<azonenberg>
and not the analog layer :p
<lain>
maybe it's /really old/
<lain>
from previous tenants
<lain>
or maybe the landlord installed an FM bug
<lain>
haha
<azonenberg>
The house was built in 1998
<azonenberg>
so the wiring is coming up on 20yo old
<azonenberg>
thats not THAT old
<rqou>
that's a very modern house
<lain>
indeed
<fpgacraft2>
<nmesisgeek> yeah my parents house was built in the 1930s lol
<rqou>
this Berkeley apartment is similar
<fpgacraft2>
<nmesisgeek> had cast iron drain pipes etc
<fpgacraft2>
<nmesisgeek> kinda surprised no horse hair insulation :p
<rqou>
hence cheater plugs and breaker trips
<azonenberg>
mtp: also you're seattle area?
<mtp>
with high probability
<azonenberg>
and not just a usb protocol in some camera? :p
<lain>
lol
<mtp>
it's short for Mx. The Plague
<azonenberg>
ah, lol
<azonenberg>
we should meet up some time
<mtp>
%work is downtown, i'm booked this week though
<azonenberg>
$work is interbay/queen anne (border area) in my case, and i'm busy this week too
<azonenberg>
but file away for future reference
<rqou>
alright, about to try to install a tpm in my desktop
<rqou>
this will be interesting
<mtp>
azonenberg, i used to work in LQA
<rqou>
hmm TPM -> bootloop
<rqou>
thanks ASRock
<pie_>
wow, even used hardcovers of artin / dummit&foote are expensive as f*** :/
<pie_>
and the booksellers are assholes listing paperbacks as hardcovers to get them in the searches
DocScrutinizer05 has quit [Disconnected by services]
DocScrutinizer05 has joined ##openfpga
pie_ has quit [Quit: Leaving]
amclain has quit [Quit: Leaving]
<rqou>
hmm i figured out why all the intel nics i have were cheap
<rqou>
they're all oem devices with unusually-small option roms
<rqou>
debating if i should switch the chip to a larger one
<rqou>
in case i ever want iscsi or fcoe or whatever
<whitequark>
who even needs option roms?
<rqou>
iscsi/pxe boot?
<azonenberg>
pxe is the main thing i've seen it used for
<whitequark>
right but wasn't there some foss bootloader specifically designed to work on these tiny crappy nics?
<rqou>
also some fcoe thing
<whitequark>
its name elides me, it was like a decade ago I last looked
<rqou>
idk why fcoe needs an option rom
<rqou>
hey, these aren't crappy nics :P
<rqou>
they're legit intel silicon, but oem pcb cards
<whitequark>
ipxe, maaybe?
<whitequark>
ipxe can do fcoe
<rqou>
my cards are "server-grade" nics, not consumer bullshit
azonenberg_work has quit [Remote host closed the connection]
ZipCPU|Laptop has quit [Ping timeout: 240 seconds]
ZipCPU has quit [Ping timeout: 240 seconds]
ZipCPU has joined ##openfpga
ZipCPU|Laptop has joined ##openfpga
pie_ has joined ##openfpga
pie_ has quit [Changing host]
pie_ has joined ##openfpga
scrts has quit [Ping timeout: 248 seconds]
scrts has joined ##openfpga
Bike has quit [Quit: antichrist]
eduardo__ has quit [Quit: Ex-Chat]
<zino>
I PXE-boot all my nodes/servers into ipxe to get them on the same level.
<rqou>
same level of what? (i don't use pxe)
<zino>
Same level of supporting whatever is needed to do the bootup. I just woke up and missed the fcoe thing though. I happyly threw away the last of our FC gear last year.
ZipCPU has quit [Ping timeout: 264 seconds]
ZipCPU|Laptop has quit [Ping timeout: 248 seconds]
<rqou>
i thought pxe was really dumb and was basically "just do a tftp?"
<rqou>
i've never used fcoe either
<rqou>
these things all seem to be really "enterprisey" features that i won't ever need
<zino>
PXE is really dumb, but iPXE dues things like HTTP, which is needed to start up the bittorrent client needed for the rest of the boot on my clusters.
<whitequark>
the what
<rqou>
what
<rqou>
why?
<zino>
The system servers get really overloaded of 2000 clients try to reinstall themselves at the same time. And they reinstall themselves every time they boot. Bittorrent solves that.
<rqou>
are you running a diskless cluster? why?
ZipCPU has joined ##openfpga
<zino>
No, they have disks, but we don't keep permanent state on them.
ZipCPU|Laptop has joined ##openfpga
<rqou>
is this for a school? or a company?
<whitequark>
disturbing
<zino>
University.
* zino
is off to work
<rqou>
ah that makes a lot more sense
<rqou>
$FANCY_SCHOOL here had some diskless sun thin client things until a a few years back
<rqou>
the central server was solaris on sparc just to confuse the freshmen :P
pie_ has quit [Ping timeout: 245 seconds]
marcus_c_ has quit [Ping timeout: 258 seconds]
marcus_c has joined ##openfpga
ZipCPU has quit [Ping timeout: 240 seconds]
ZipCPU|Laptop has quit [Ping timeout: 255 seconds]
ZipCPU has joined ##openfpga
ZipCPU|Laptop has joined ##openfpga
promach has joined ##openfpga
ZipCPU|Laptop has quit [Ping timeout: 252 seconds]
ZipCPU has quit [Ping timeout: 252 seconds]
ZipCPU|Laptop has joined ##openfpga
ZipCPU has joined ##openfpga
pie_ has joined ##openfpga
scrts has quit [Ping timeout: 240 seconds]
scrts has joined ##openfpga
fpgacraft2 has quit [Quit: ZNC 1.7.x-git-709-1bb0199 - http://znc.in]
fpgacraft1 has quit [Quit: ZNC 1.7.x-git-709-1bb0199 - http://znc.in]
fpgacraft2 has joined ##openfpga
fpgacraft1 has joined ##openfpga
rqou has quit [Remote host closed the connection]
rqou has joined ##openfpga
m_t has joined ##openfpga
promach has quit [Quit: Leaving]
pie_ has quit [Ping timeout: 256 seconds]
amclain has joined ##openfpga
ZipCPU has quit [Ping timeout: 255 seconds]
ZipCPU|Laptop has quit [Ping timeout: 240 seconds]
ZipCPU has joined ##openfpga
ZipCPU|Laptop has joined ##openfpga
Bike has joined ##openfpga
_whitelogger has joined ##openfpga
pie_ has joined ##openfpga
<pie_>
<zino> PXE is really dumb, but iPXE dues things like HTTP, which is needed to start up the bittorrent client needed for the rest of the boot on my clusters.
<pie_>
i want this
<pie_>
i mean come on we have intel ME so fuck it
pie__ has joined ##openfpga
pie_ has quit [Ping timeout: 248 seconds]
mifune has joined ##openfpga
mifune has quit [Changing host]
mifune has joined ##openfpga
ZipCPU|Laptop has quit [Ping timeout: 240 seconds]
ZipCPU has quit [Ping timeout: 255 seconds]
ZipCPU|Laptop has joined ##openfpga
ZipCPU has joined ##openfpga
<azonenberg>
pxe over bittorrent? o_O
m_t has quit [Quit: Leaving]
m_w has joined ##openfpga
<nats`>
azonenberg quick question about constraint
<nats`>
when using OFFSET = IN is it possible to define range ?
<azonenberg>
Dont know, i havent done much with io timing constraints
<azonenberg>
most of my stuff has been fast on chip and relatively slow off chip, or used a GTP
<azonenberg>
or i just constrained the FFs to the IOBs and hoped for the best :p
<nats`>
so even in rgmii you don't define it ?
<azonenberg>
No, i just used iob ff's and did the phase shift with a PLL or IODELAY
<azonenberg>
should probably fix that :p
<nats`>
I'm settings it because in my case by default it's the marveel phy which do the delay
<azonenberg>
ah
<nats`>
I guess I should put worst case
<nats`>
the more constrained around the edge
<cr1901_modern>
Why not a DCM? Or did they remove those on series 7?
<azonenberg>
DCMs are long dead
<azonenberg>
they have horrible jitter
<azonenberg>
PLL and MMCM are the options now, both are true analog VCO based clock synthesizers
<zino>
azonenberg: Na. bittorrent loader loaded over ipxe loaded over pxe. :)
<azonenberg>
lolol
<cr1901_modern>
Horrible jitter? I mean, if I synthesize a clk at some integer multiple or divider of an input using a clock DCM, the jitter is low enough that registers can be safely passed between them w/o metastability/cdc
<nats`>
yes if you stay pretty low
<nats`>
but it means you have a less large safe area..
<nats`>
the more you have jitter and skew the more you need tos tay away from edge
<cr1901_modern>
(Btw, most things should of course use clock enable, but sometimes it's just easier to synthesize a divided clock and use that)
<cr1901_modern>
In the particular case I'm thinking of, the core I was provided has no clk enable, and I didn't think manually gating the clock is the way to go.
<cr1901_modern>
azonenberg: How did a DCM synthesize a faster clock anyway? Did it rely on clocking a phase shifted version of the input (up to pi/32 increments)?
<cr1901_modern>
2*pi/32* increments
whitequark has joined ##openfpga
DocScrutinizer05 has quit [Remote host closed the connection]
DocScrutinizer05 has joined ##openfpga
scrts has quit [Ping timeout: 240 seconds]
scrts has joined ##openfpga
<azonenberg>
cr1901_modern: Exact details aren't public but i have some die shots of the spartan3a dcm if anyone wants to look at it and try to figure it out :p
<azonenberg>
only active layer for now though
<azonenberg>
my understanding is that it was DLL based
<azonenberg>
so basically a ring oscillator with pass transistors at various points in the loop to adjust the ring length
<azonenberg>
the simple naive structure is two parallel lines of inverters in a U shape with a single odd inverter at the far end
<azonenberg>
then a line of pass transistors bridging the two halves of the U
<azonenberg>
when all pass transistors are open, the wavefront comes in one leg of the U and out the other end, even inverter count through the U then one extra at the end makes it unstable so it oscillates
<azonenberg>
all but the last one*
<azonenberg>
To speed the process along you'd close pass transistors starting at the tip of the U then moving up to the open end
<azonenberg>
and increase the frequency
<azonenberg>
Then to control those pass transistors you'd have a shift register with a single bit set in it
<azonenberg>
stretching in parallel with the U
<azonenberg>
and your phase comparator would shift that bit left or right to increase/decrease the oscillator frequency
<azonenberg>
in reality it wouldnt be nice and linear like that, it's probably an S-curve to make more efficient use of space
<azonenberg>
and maybe a decoder instead of a shift reg
<azonenberg>
but same idea
carl0s has joined ##openfpga
mifune_ has joined ##openfpga
eduardo has joined ##openfpga
mifune has quit [Ping timeout: 255 seconds]
carl0s has quit [Ping timeout: 255 seconds]
<Lord_Nightmare>
azonenberg: so an oscillator made of a PLL and a hugely long shift register in a loop which it clocks, which can have different things preloaded to it?
<Lord_Nightmare>
with different selectable taps to 'shorten' the ring?
maaku has quit [Ping timeout: 245 seconds]
maaku has joined ##openfpga
maaku has quit [Ping timeout: 245 seconds]
maaku has joined ##openfpga
<cr1901_modern>
Lord_Nightmare: DCM is "purely digital". This is why I have questions in the first place about how to get a clean multiplied clock from one :P
<cr1901_modern>
Sure a ring buffer can do it, but I don't really trust those things
<nats`>
shit I wonder why I struggle placing ddr input behind an IO
<azonenberg>
Lord_Nightmare: not a PLL
<azonenberg>
a PLL uses a VCO as the oscillating element
<azonenberg>
a DLL uses a ring oscillator
<azonenberg>
The shift register is just an easy means of selecting taps
<Lord_Nightmare>
couldn't you use an LFSR as an oscillator as well, by detecting the 'all zeroes except 1 bit' case?
<Lord_Nightmare>
that way you only set up the xor taps and optional invert at the end
<Lord_Nightmare>
an you can get any power of 2 minus one length oscillation
<Lord_Nightmare>
plus a bunch of suboptimal lengths of various length
<Lord_Nightmare>
but it sounds like a DLL is fully asynchronous and the whole fact that it oscillates at all is what allows it to be used as an oscillator source
<azonenberg>
it's not just the oscillation
<azonenberg>
remember it phase locks to an incoming signal and multiplies
<azonenberg>
So for example, to double an incoming signal you put a 2x divider on the output of the ring
<azonenberg>
then adjust taps as needed in order to phase lock the output of the divider to your reference clock
<azonenberg>
so every two rising edges of the ring you get one rising edge on the input
<azonenberg>
a LFSR oscillator cannot be easily adjusted by +1 or -1 tap
<azonenberg>
in other news, just ordered a bunch of additional RF cables and adapters to replace stuff i hacked together with BNC flying leads
<azonenberg>
so i should have cleaner signals in my test environment in a week or so
mifune_ has quit [Ping timeout: 255 seconds]
carl0s has joined ##openfpga
<nats`>
uhhmmmmmmmm azonenberg I made something not sure it's clean :D
<nats`>
I have this hold violation on the rgmii data
<nats`>
I added a delau
<nats`>
delay
<nats`>
seems to pass
maaku has quit [Ping timeout: 245 seconds]
maaku has joined ##openfpga
ZipCPU has quit [Ping timeout: 245 seconds]
ZipCPU|Laptop has quit [Ping timeout: 264 seconds]