kristianpaul has quit [Read error: Connection reset by peer]
kristianpaul has joined ##openfpga
<pie_>
is this because intel bought...wait no what
<pie_>
did intel buy altera or xilinx anno?
tannewt has quit [Read error: Connection reset by peer]
edmoore has quit [Ping timeout: 268 seconds]
esden has quit [Ping timeout: 264 seconds]
tannewt has joined ##openfpga
eddyb[legacy] has quit [Ping timeout: 240 seconds]
eddyb[legacy] has joined ##openfpga
esden has joined ##openfpga
edmoore has joined ##openfpga
<mwk>
pie_: intel bought altera long time ago; amd bought xilinx, like, this week
<pie_>
ok right
<pie_>
so did they buy xilinx because intel bought altera? :P
tlwoerner has quit [Quit: Leaving]
cr1901_modern has quit [Ping timeout: 260 seconds]
cr1901_modern has joined ##openfpga
<implr>
possible? also maybe scared of nv buying arm & mellanox
<implr>
it is a bit of a strange transaction, they overbid by a lot
<implr>
and amd isn't exactly swimming in cash
<implr>
they paid off most(all?) of their debt recently iirc, presumably with ryzen profits
<implr>
but they aren't like nv or apple with actual tens of billions of $ in the bank
kristianpaul has quit [Read error: Connection reset by peer]
kristianpaul has joined ##openfpga
jeanthom has quit [Ping timeout: 240 seconds]
Degi has quit [Ping timeout: 264 seconds]
Degi has joined ##openfpga
_whitelogger has joined ##openfpga
Bike has quit [Quit: Lost terminal]
<kc8apf>
I've heard a theory that AMD wants Xilinx's serdes IP
<sorear>
somehow I assumed there was a single serdes IP company that everyone used via a comically long sequence of sublicenses
kristianpaul has quit [Read error: Connection reset by peer]
kristianpaul has joined ##openfpga
<scientes>
is there something I ran read about doing DMA from a FPGA attached via PCIe?
<scientes>
and also syncronization
<scientes>
I want to implement a queue
emeb_mac has quit [Quit: Leaving.]
<whitequark>
azonenberg: pong. i've been in too much pain to set up the scope. i still have the notification in my email though
jeanthom has joined ##openfpga
Asu has joined ##openfpga
_whitelogger has joined ##openfpga
jeanthom has quit [Ping timeout: 260 seconds]
s_frit_ has quit [Remote host closed the connection]
s_frit has joined ##openfpga
<implr>
kc8apf: $35B for a serdes ip seems a bit overkill
<whitequark>
xilinx serdes ip isn't even good, is it
finsternis has joined ##openfpga
s_frit has quit [Remote host closed the connection]
s_frit has joined ##openfpga
_franck_ has quit [Read error: Connection reset by peer]
_franck_ has joined ##openfpga
_franck_ has quit [Ping timeout: 256 seconds]
_franck_ has joined ##openfpga
_franck_9 has joined ##openfpga
_franck_ has quit [Ping timeout: 268 seconds]
_franck_9 is now known as _franck_
oeuf has quit [Read error: Connection reset by peer]
<azonenberg>
whitequark: $35B for a serdes ip is indeed overkill. lol
<gruetzkopf>
also, AMDs PCIe serdes aren't bad or anything
CounterPillow has joined ##openfpga
egg has joined ##openfpga
OmniMancer has quit [Quit: Leaving.]
emeb_mac has joined ##openfpga
kristianpaul has quit [Read error: Connection reset by peer]
kristianpaul has joined ##openfpga
<kc8apf>
AMD's I/O chiplet is still on 12nm and getting rather old. Looking forward, they need PAM4 for PCIe gen 6. They could keep buying Synopsis IP but I/O is becoming the differentiator that frequency was years ago
emeb has joined ##openfpga
<azonenberg>
kc8apf: speaking of pcie, holy moley
<azonenberg>
it looks ilke the pcie gen 5 (not 6) loss budget is 36 dB
<azonenberg>
36 dB of insertion loss between tx and rx
<kc8apf>
Yup. Gen4 already has a lot of retimers used on larger server designs. Gen5 will probably make them common.
<kc8apf>
For 200BASE-C4, we're finding it easier to use flyover cables from QSFP-DD cage to right near a switch ASIC as cable loss is lower than even the fanciest of PCB materials
<whitequark>
what does "BASE" mean btw?
<whitequark>
"baseband"?
<kc8apf>
Ethernet standard. I missed a G 200GBASE-CR4
<whitequark>
nono
<whitequark>
i know that it refers to ethernet
<agg>
the BASE is baseband but I don't think there's any 'passband'/modulated versions?
<agg>
there's AUI
<whitequark>
i want to know where specifically the letters "BASE" originate from, because my copy of 802.3 doesn't say
<kc8apf>
I was just going to look at 802.3. must be even earlier
<agg>
and there are indeed some passband variants like 10PASS-TS
<azonenberg>
Yeah but the vast majority of ethernet-over-twisted-pair is baseband signalin
<azonenberg>
hence the name
<azonenberg>
kc8apf: and yes coax is generally lower loss than most pcb materials
<azonenberg>
This is a challenge i'm running into with my probe designs, trying to keep loss on the PCB as low as possible
<kc8apf>
10BROAD36 seems like a fun thing to implement
<agg>
wikipedia's explanation of the unidirectional thing sounds mad
<agg>
broadcast packets would go from you to the end of the line in one direction, which would demodulate them and re-transmit on another frequency back in the other direction?
<kc8apf>
That's not that different from how DOCSIS works
<agg>
except in DOCSIS you don't normally need to talk to the other nodes, right?
<kc8apf>
Except the whole intentionally limiting broadcast
<kc8apf>
Right. DOCSIS improves things like ARP handling
<kc8apf>
I might have to read DOCSIS again
<gruetzkopf>
i have a bunch of cisco LRE (Eth-over-VSDL1) equipment
<gruetzkopf>
still haven't managed to get my hands on 10BASE-T1L
<gruetzkopf>
(and i'm fairly happy that i got cost for one 40GE link down to just above 40€ ;))
emeb has left ##openfpga [##openfpga]
<omnitechnomancer>
Would you not use twinax for high speed digital pairs?
<kc8apf>
yes, that what -C means in 100GBASE-C4
<gruetzkopf>
thats QSFP28 DAC ?
<azonenberg>
I want to get hold of some 100base-T1 gear to play with sniffing it at some point