ayjay_t has quit [Read error: Connection reset by peer]
ayjay_t has joined ##openfpga
* sxpert
discovers wierd ass chips that he never heard of
<tnt>
azonenberg: why the TLK10232 if you're not using the secnd channel ? Isn't the TLK10031 the same but with only 1 channel ?
<azonenberg>
tnt: Because i have about 15 TLK10232 samples sitting around my lab waiting to get used :p
<sorear>
I did not realize you were misusing a 10G PHY in quite this way
<azonenberg>
i'd likely respin if this went into production
<azonenberg>
sorear: lol
<azonenberg>
Can you think of a better way to recover a clock off a high-speed serial signal? nobody sells CDR PLLs as a discrete component
<azonenberg>
at least not past a few hundred Mbps
<azonenberg>
so you basically have to use a full serdes IP and discard the data
<azonenberg>
and fpga serdes are out b/c too much jitter on the clock tree etc
<tnt>
azonenberg: otoh, after checking pricing it's really not that much cheaper.
<azonenberg>
tnt: yeah power is the bigger issue
<azonenberg>
power numbers for the tlk10232 with various blocks disabled are hard to find so i have to overdesign the psu to handle the whole thing
<azonenberg>
Good news is, i can build an external CDR circuit on a different board and feed it into the external-clock input for testing
<sorear>
so I read "fun features like CDR trigger" to imply data-dependent triggering, possibly with a shift register + comparator
<azonenberg>
without a huge impact on performance
<azonenberg>
sorear: For the moment the triggering is only clock recovery, there's no data dependence
<azonenberg>
Pattern trigger at multiple Gbps is nontrivial
<azonenberg>
but might be a wishlist feature for way further out
<azonenberg>
the big thing is, the latency through the tlk10232 is unpredictable so i can't use it as a trigger source
<azonenberg>
i'd have to build my own trigger circuit that ran at full line rate
<azonenberg>
I could do it, using a shift register + comparator, however you'd be looking at $15ish per bit of the shreg plus $12ish for each comparator, plus passives and PCB area
<azonenberg>
more importantly, as soon as your trigger is no longer free-running you can't lock a PLL to it, and your sample rate drops from 1 Tsps to 90 Gsps with much higher jitter
* sxpert
knew he was in wierdo-land when he read SiGe on one of the datasheets ;-)
<azonenberg>
Lol
<azonenberg>
This board has $160 of comparators on it
<sxpert>
yeah, saw that
<azonenberg>
And $58 of discrete DFFs
<sxpert>
had a few links to mouser on the search ;)
<azonenberg>
The 74xx equivalents for 10G data have double digit price tags :p
<sxpert>
12 EUR flipflops
<azonenberg>
Yep
<sxpert>
I wonder what those trans-ocean cable termination systems are made of
<azonenberg>
SiGe? :p
<sxpert>
well the linerate is like 1+ Tbit/s
<azonenberg>
yeah but that's after WDM
<azonenberg>
its not one wavelength carrying that much
<sxpert>
ah yeah
<azonenberg>
it's DWDM and probably PAM4 on each wavelength (actually idk, PAM4 seems to be just hitting mainstream now)
<sxpert>
it's like 100G "coherent" whatever that means
<sxpert>
probably 404x25Gbit/s
<sxpert>
argh
<sxpert>
40*4*25Gbit/s
<sxpert>
then you need freaking lazers to shoot that across the ocean
<sxpert>
hah, kintex ultrascale+ has "integrated 100G Ethernet MAC with RS-FEC and 150G Interlaken cores"
<sxpert>
that will do it
<davidc__>
heh, I saw a pricetag for a small multi 100Gbit WDM system. I'm fairly sure they are made of unobtainium
<sxpert>
heh
<sorear>
world market for maybe 6 multi 100Gbit WDM systems
<sorear>
(yes, I know that's massively out of context, but the out of context quote has its own history now)
<davidc__>
actually, the market is much much larger than that
<davidc__>
anyone that operates a backbone uses them
<sxpert>
sorear: either that, or vibranium from wakanga
<sxpert>
sorear: yeah lol...
<sxpert>
sorear: then there is the similar "nobody needs more that 10Gbit/s"
emeb_mac has quit [Ping timeout: 255 seconds]
lain has quit [Quit: kthxbai]
<sxpert>
ABC: Warning: The network is combinational (run "fraig" or "fraig_sweep").
<sxpert>
what should I do ?
indy has joined ##openfpga
unixb0y has quit [Ping timeout: 255 seconds]
unixb0y has joined ##openfpga
Flea86 has quit [Read error: Connection reset by peer]
<daveshah>
sxpert: absolutely nothing, it's an artefact of how Yosys interacts with ABC
<sxpert>
ah ok
* sxpert
is cutting up his bigger modules into smaller ones
<sxpert>
trying to make that all more readable
rohitksingh has joined ##openfpga
rohitksingh has quit [Ping timeout: 246 seconds]
scrts has quit [Ping timeout: 255 seconds]
rohitksingh has joined ##openfpga
Laksen has joined ##openfpga
Asu has joined ##openfpga
Asu has quit [Remote host closed the connection]
kuldeep has quit [Remote host closed the connection]
kuldeep has joined ##openfpga
rohitksingh has quit [Ping timeout: 255 seconds]
Miyu has joined ##openfpga
scrts has joined ##openfpga
somlo has quit [Quit: Leaving]
somlo has joined ##openfpga
emily has quit [Remote host closed the connection]
emily has joined ##openfpga
<_whitenotifier-9>
[whitequark/Glasgow] whitequark pushed 1 commit to master [+1/-0/±2] https://git.io/fhbWG
<_whitenotifier-9>
[whitequark/Glasgow] whitequark b9de91e - applet.yamaha_opl: add a web interface.