futarisIRCcloud has quit [Quit: Connection closed for inactivity]
kgugala_ has joined #yosys
kgugala has quit [Ping timeout: 246 seconds]
dys has joined #yosys
emeb_mac has quit [Quit: Leaving.]
kraiskil has joined #yosys
kraiskil has quit [Ping timeout: 260 seconds]
jakobwenzel has joined #yosys
kraiskil has joined #yosys
kraiskil has quit [Ping timeout: 258 seconds]
kgugala_ has quit [Read error: Connection reset by peer]
kgugala has joined #yosys
kraiskil has joined #yosys
kraiskil has quit [Ping timeout: 260 seconds]
Asu has joined #yosys
kraiskil has joined #yosys
Asu has quit [Remote host closed the connection]
Asu has joined #yosys
citypw has joined #yosys
N2TOH has joined #yosys
captain_morgan4 has joined #yosys
N2TOH_ has quit [Ping timeout: 265 seconds]
cr1901_modern has quit [Ping timeout: 265 seconds]
captain_morgan has quit [Ping timeout: 265 seconds]
corecode_ has joined #yosys
corecode has quit [Ping timeout: 265 seconds]
smarter has quit [Ping timeout: 265 seconds]
smarter has joined #yosys
craigo has quit [Ping timeout: 265 seconds]
cr1901_modern has joined #yosys
<lambda>
claire isn't usually in here, right? I'd really like to start working on memory stuff, but without knowing what she wants it to look like, I fear anything I do might be for naught
<Lofty>
lambda: correct, but you can ask her on Twitter or by email
<whitequark>
why not use github for discussing code?
<lambda>
whitequark: that's where the discussion stalled a few days ago, not sure if she even saw it
<whitequark>
yeah sometimes that happens
<lambda>
eh, guess I'll poke her on twitter in a day or two
bubble_buster_ has joined #yosys
rrika_ has joined #yosys
[Ristovski] has joined #yosys
Asuu has joined #yosys
asinghan2 has joined #yosys
withig has joined #yosys
mwk_ has joined #yosys
emilazy_ has joined #yosys
tmiw_ has joined #yosys
bubble_buster has quit [Ping timeout: 246 seconds]
Xark has quit [Ping timeout: 246 seconds]
rrika has quit [Ping timeout: 246 seconds]
nengel has quit [Ping timeout: 246 seconds]
mwk has quit [Ping timeout: 246 seconds]
markus-k has quit [Ping timeout: 246 seconds]
knielsen has quit [Ping timeout: 246 seconds]
emilazy has quit [Ping timeout: 246 seconds]
asinghan1 has quit [Ping timeout: 246 seconds]
tmiw has quit [Ping timeout: 246 seconds]
Forty-Bot has quit [Ping timeout: 246 seconds]
bubble_buster_ is now known as bubble_buster
kristian1 has joined #yosys
emilazy_ is now known as emilazy
pacak has quit [Ping timeout: 264 seconds]
kgugala has quit [Ping timeout: 264 seconds]
cyrozap has quit [Ping timeout: 264 seconds]
DaKnig has quit [Ping timeout: 264 seconds]
Ristovski has quit [Ping timeout: 264 seconds]
Asu has quit [Ping timeout: 264 seconds]
indy has quit [Ping timeout: 264 seconds]
kristianpaul has quit [Ping timeout: 264 seconds]
anuejn has quit [Ping timeout: 264 seconds]
knielsen_ has joined #yosys
Forty-Bot has joined #yosys
indy_ has joined #yosys
kgugala has joined #yosys
mwk_ is now known as mwk
anuejn_ has joined #yosys
[Ristovski] is now known as Ristovski
emeb has joined #yosys
Asuu has quit [Remote host closed the connection]
Asu has joined #yosys
futarisIRCcloud has joined #yosys
kristian1 has quit [Quit: WeeChat 2.8]
kristianpaul has joined #yosys
jakobwenzel has quit [Quit: jakobwenzel]
Asu has quit [Remote host closed the connection]
Asu has joined #yosys
Asu has quit [Remote host closed the connection]
Asu has joined #yosys
Asu has quit [Remote host closed the connection]
Asu has joined #yosys
jakobwenzel has joined #yosys
jakobwenzel has quit [Remote host closed the connection]
citypw has quit [Ping timeout: 240 seconds]
withig is now known as nengel
az0re has quit [Remote host closed the connection]
N2TOH has quit [Ping timeout: 256 seconds]
az0re has joined #yosys
dys has quit [Ping timeout: 256 seconds]
knielsen_ is now known as knielsen
kgugala has quit [Quit: -a- Connection Timed Out]
kgugala has joined #yosys
craigo has joined #yosys
<thardin>
has anyone here fiddled with the MAX11300 by any chance?
<thardin>
20-channel PIXI mixed-signal thingy
<thardin>
nm I figured it out. the R/W bit is 0 for write, 1 for read
<az0re>
daveshah: Wow, that is really fucked up
<az0re>
WTF is Lattice thinking
<az0re>
They should have gone all-in on open source tooling
<az0re>
They should be doing the job you have done/are doing, or at least helping you
<az0re>
Almost certainly a decision by some idiot MBA
<az0re>
I think you should play hardball with them. Declare your intention to stop work on Lattice products and do everything you can to assert your IP rights, work with Claire, etc. to remove your work from future versions of Yosys and SymbiFlow
<az0re>
If they want to poke you in the eye for doing great work to expand their market for them, fuck them
<az0re>
Maybe it's time to consider GPL for Yosys
<az0re>
Bespoke commercial licenses available for purchase from Symbiotic EDA
<daveshah>
I don't think Lattice will really care if Yosys et al loses support for them - they might even be happier - but I'm definitely not working on support for the new lattice parts any more
<daveshah>
If Yosys goes GPL then I will maintain an ISC fork, that doesn't solve anything
<daveshah>
I'm waiting to see what happens with various contacts first, maybe we can get a U-turn or better from Lattice
<az0re>
Do you know who actually made the decision?
<daveshah>
No idea
<daveshah>
I've had no direct contact with Lattice yet
<az0re>
I understand completely how you feel about the license. But this is basically your only leverage. They benefit from the work you did and are now flipping you the bird. IMO you should play your cards.
<az0re>
You've gotta impose some sort of cost as a deterrence, even if they don't see it as a deterrent right now
<daveshah>
A few contacts who buy in at least medium volume from Lattice are trying to see what they can do
<az0re>
Cool
<thardin>
use AGPL to really piss them off
<az0re>
But I doubt they'll be able to do anything
<daveshah>
I don't see how GPL/AGPL solves anything here
<az0re>
Well, there are two ideas here:
<az0re>
1) Change licenses
<daveshah>
It seems like Lattice _don't_ want people to use Yosys so making they won't care if we make it harder to use Yosys
<az0re>
2) Remove lattice support from future versions of Yosys/SymbiFlow
<daveshah>
I think the only way to success is to try and explain a tangible link between open source tools and sales
<az0re>
Correct, they don't like this alternative implementation. But they do like the increased revenue from having a part compatible with a fully open source toolchain. They may not recognize the link. Make it clear.
<daveshah>
At the end of the day that is all they care about
<az0re>
Right
<az0re>
If that open source toolchain no longer supports their chips, people buying for open source compatibility will stop buying
<az0re>
You don't have to remove support from old versions, even if it were possible
<az0re>
Just the fact that support will decay, no new features will be added, etc. will be enough of a deterrent
<daveshah>
Well, that will happen by default now
<az0re>
For new chips, yes
<daveshah>
I can't add support for any of the further fancy new FPGAs supposed to be coming out soon
<az0re>
But if current chips are not supported in new Yosys versions
<thardin>
you could try and ask them whether they're a hardware company or a software company
<thardin>
get them to focus on the hardware side
<az0re>
I think that would go nowhere. I am not privy to the dicussion about making these changes, but IIUC FPGA vendors are so clammy about specs because they're worried about patent lawsuits and think that it will be easier to find material to litigate with an open implementation of a toolchain
<az0re>
So, they're a hardware company, but they want to control relevant software, too
<daveshah>
There's a perennial worry of people making bitstream compatible clones, too
<az0re>
Actually that leads to another potential point of leverage: Tell them you're pissed, why, and that you will now spend all your reverse engineering effort looking for patent infringement, which you will happily forward to Xilinx, Intel, and MicroSemi
<daveshah>
People in glass houses shouldn't throw stones, god knows what the patent situation is for Yosys/nextpnr
<az0re>
I know what the patent situation is for Yosys/nextpnr
<az0re>
Irrelevant for noncommercial uses
<thardin>
also irrelevant for anywhere but the US
<thardin>
or at least in EU
<sorear>
lattice announced $97 million in revenue in 2020Q1
<az0re>
Yosys and nextpnr would also only be vulnerable to software patent claims
<sorear>
how much of that do you think was from people using nextpnr
<az0re>
I don't know what Lattice's patent portfolio looks like, but I doubt they have many software patents
<thardin>
what a swell system this is where groups of people can't share work because reasons
<smkz>
whats the state of open source toolchain support for xilinx parts; i had been considering using an ecp5 for a design because of open source toolchain support but with this stuff going on i'm wondering if i should reconsider ;;
<daveshah>
It's more experimental, but improving
<thardin>
I got the arty 35t board, seems there's support for htat
<az0re>
daveshah: Your GitHub profile pic IIRC is an Artix-200T. is that supported?
<daveshah>
That's from years ago! But X-ray should support it now, I think a few nextpnr patches would be needed for it to work
<daveshah>
I've been more working on improving the general scalability at the moment, rather than arch-specific stuff
<az0re>
Cool
<az0re>
Actually I have one sitting around. Maybe in a couple weeks I'll try to start using it. If I encounter any problems, I'll be sure to let you know.
* smkz
nods
<smkz>
what's the biggest (in terms of logic / dsp and maybe serdes) xilinx fpga that's supported or will soon be supported by yosys/nextpnr (or other open source tools)
<daveshah>
If you don't care about open bitstream generation yet, vu13p soon
* smkz
nods
<az0re>
And what about with open bitstream generation? End-to-end open source?
<daveshah>
xc7a200t for now
<daveshah>
I think X-ray should get k325 support fairly soon
<az0re>
Got it
<az0re>
Thanks :)
<az0re>
I was previously interested in Lattice because they seemed to "get it" and be on board the open source train. But now that it's clear they are not, well, if I'm going to be working with assholes, I'd at least like to work with the assholes with the best products.
<smkz>
tbh that thing with lattice makes me really sad because the open source toolchain (yosys and arachne-pnr) was pretty much what got me into fpgas with an ice40 ;-;
<az0re>
I don't understand this business decision.
<daveshah>
I honestly have no idea what happened at Lattice. There are/were definitely senior people on board, but also some senior people against open tools
<az0re>
Now what do they have to differentiate them in the market?
<sorear>
right hand, left hand, etc etc
<thardin>
does anyone have a link to info on what exactly lattice is doing? I'm missing enough backlog
<daveshah>
One possible reason for Lattice including this is they are trying to promote their parts for platform security
<thardin>
ah
<thardin>
yeah that's some bs
<daveshah>
I guess maybe they think open bitstreams will make their parts "less secure"
<thardin>
I guess they don't understand that security through obscurity doesn't work
<daveshah>
I don't know if this is true. But currently this SDK targets their security-oriented part (xo3d)
<az0re>
If so, that's extra hilarious, as what thardin says is true: If they think lack of open source toolchains helps to secure their products, the security of their products should be *less* trusted
<az0re>
Because clearly they don't know what they're doing lol
<thardin>
yep
<az0re>
Xilinx gets pilloried for the constant breaks in its bitstream decryption
<az0re>
But Ithey're getting their act together and working on actually proving the invulnerability of their implementation to various attacks
<az0re>
s/Ithey/they/
<az0re>
Lattice instead gets really scared they don't have a secure product, so they take legal means to deter people from looking
<thardin>
imagine if the engineers at xilinx, lattice etc had a say instead of the bean counters
<az0re>
I mean... if you're right daveshah, and you quite possibly are, Lattice is just a joke of a company
<daveshah>
It's a total guess. But if this isn't just a legal team gone rampant (and it seems too technically specific) then there will be a reason
<az0re>
BTW looks like relevant stakeholders at LatticeSemi are Esam Elashmawi (Chief Marketing & Strategy Officer), Byron Milstead (General Counsel), and Mark Nelson (VP of Worldwide Sales).
<thardin>
stakeholders should only means those who stake vampires, not those who are
<az0re>
Might be worth shooting them an email: "Are you sure you want to declare war on the open source community just because you're scared of how shitty your products might be?" Make sure to include legally-not-a-threat threats about licensing changes, removal of support from Yosys/nextpnr and revenue impacts it will have, and looking for security flaws and selling them on black market, patent infringement and giving evidence to competitors, etc.
<az0re>
Probably won't change their minds if they all decided on this, but if some of them haven't heard of this decision and weren't involved, it wil lraise some red flags.
<somlo>
az0re: OTOH, maybe a "nice" letter inquiring about whether they're sure they want to alienate the FOSS market (which preferred them *because* of their insightful and friendly attitude so far)
<thardin>
might want to be more diplomatic at first
<thardin>
what somlo said
<az0re>
I mean, sure, if you want to waste time. But there's no need for pleasantries.
<az0re>
This is a business decision.
<thardin>
something like "don't damage the goodwill you've accrued"
<tpb>
Title: An uproar over the Fedora Git forge decision [LWN.net] (at lwn.net)
<somlo>
it's "multiple personality" and "distributed decision making" :)
<az0re>
Yep that's a great comment
kraiskil has quit [Ping timeout: 260 seconds]
<az0re>
I did not mean to anthropomorphize LatticeSemi with "you". I should have written "y'all" and "y'all're".
<somlo>
I wasn't trying to imply you're even antropomorphizing them :) It's just that I'm (maybe naively) hoping this is an example of where someone in charge can step in and say "You done screwed up, now go fix it" :)
<az0re>
Hopefully there's someone--maybe CC the CEO, too--who can think for 5 seconds and put a stop to this idiocy
* sorear
is unconvinced people who care about OSS make up more than 0.1% ($100K this quarter) of Lattice's revenue
<sorear>
if you can convince lattice they have any reason to care, maybe you can also convince me?
<thardin>
how many people use lattice's tools?
<daveshah>
the main reason for them to care is brand recognition
<daveshah>
xilinx and intel spend loads on free dev boards and material for unis
<thardin>
are any of these tools scriptable in a normal terminal? that is, Makefile-compatible?
<thardin>
you could make a CI argument
<daveshah>
yes, although that can be done easily enough with the Lattice tools
<daveshah>
the main problem with the Lattice tools and CI is the MAC locking
<thardin>
even on debian?
<daveshah>
for private CI that tends to work though
<thardin>
ah. that sounds annoying
<daveshah>
Probably
<thardin>
not that MAC addresses are unique
<daveshah>
or hard to change
<daveshah>
but you wouldn't steal a car; you wouldn't steal a gratis software license
<somlo>
Lattice's *other* problem with uni ECE curricula (besides licensing and overall clunkiness of the toolchain) is the lack of a decently equipped dev board
<bwidawsk>
I thought customers of the fpga vendors would want to be able to debug their own issues, optimize their own flows, and have mobility between hardware...
<bwidawsk>
that turned out to not be as true as I thought
<thardin>
stockholm syndrome or something
<somlo>
the want those things, they're just too far zoomed in to know it :)
<bwidawsk>
what I ran into was the shops get engineers who are good at *A* toolchain
<bwidawsk>
the cost of changing is too high
<bwidawsk>
so yeah, somlo, too zoomed in
<bwidawsk>
daveshah: sorry this was your plight
<daveshah>
oh don't worry, luckily there are other fish in the sea
<daveshah>
lattice's new parts were pretty overpriced anyway
<thardin>
woop, seems a HX1K is sufficient for the project I'm working on
<thardin>
52% utilization for the most important parts of the design
<thardin>
67 MHz even
<Lofty>
daveshah: so what now? Xilinx?
<Lofty>
And hope for a Lattice retraction?
<daveshah>
Basically, yes
<Lofty>
And presumably prjoxide is on hold?
<Lofty>
Though the only part it targeted was CrossLink, right?
<daveshah>
That was for the CrossLink NX
<daveshah>
It will indeed be on hold
<daveshah>
I was hoping to continue it for the new iCE parts whenever they are released
<daveshah>
But the tools for those will almost certainly be under this new license
<Lofty>
We'll see, I suppose
<sorear>
and even if OSS was a driving factor, they can't count on exclusivity for any length of time
Asuu has joined #yosys
Asu has quit [Ping timeout: 260 seconds]
<awygle>
the security thing makes a fair amount of sense tbh
<awygle>
(not like, technically, but if they were _only_ going to include that clause on _some_ parts, it'd be the security ones)
<awygle>
i will hold out hope that when the ECP6 comes out there's no such issue
<awygle>
i really don't want to use xilinx parts :(
<whitequark>
daveshah: so this actually makes me more likely to spend time REing those parts
<whitequark>
and, naturally, openly release cracks for Diamond as a punitive measure
<whitequark>
or whatever is the new tool
<daveshah>
Radiant
<daveshah>
This new thing is Propel, which ironically is little more than a bundle of open source software tooling
<whitequark>
ah right
<daveshah>
As far as I know the new parts will still use Radiant in general, Propel is just their IP integrator and SDK
<az0re>
whitequark: +1
<daveshah>
But I suspect Radiant will use this new license for future versions too
<az0re>
I'll help if I can
emeb has left #yosys [#yosys]
emilazy has quit [Ping timeout: 246 seconds]
lukego has quit [Ping timeout: 244 seconds]
ktemkin has quit [Ping timeout: 272 seconds]
sorear has quit [Ping timeout: 272 seconds]
litghost has quit [Ping timeout: 272 seconds]
bubble_buster has quit [Ping timeout: 244 seconds]
svenn has quit [Ping timeout: 256 seconds]
flammit has quit [Ping timeout: 256 seconds]
y2kbugger has quit [Ping timeout: 260 seconds]
thoughtpolice has quit [Ping timeout: 260 seconds]
rohitksingh has quit [Ping timeout: 252 seconds]
benreynwar has quit [Ping timeout: 256 seconds]
pointfree has quit [Ping timeout: 260 seconds]
y2kbugger has joined #yosys
ovf has quit [Ping timeout: 260 seconds]
rjeli has quit [Ping timeout: 260 seconds]
futarisIRCcloud has quit [Ping timeout: 252 seconds]
litghost has joined #yosys
<az0re>
netsplit?
mithro has quit [Ping timeout: 272 seconds]
daveshah has quit [Ping timeout: 272 seconds]
<az0re>
irccloud disconnected or something?
emilazy has joined #yosys
daveshah has joined #yosys
rjeli has joined #yosys
rohitksingh has joined #yosys
svenn has joined #yosys
bubble_buster has joined #yosys
mithro has joined #yosys
<awygle>
i get the impulse but that feels like a step 3 or 4 to me? would personally like to give them a chance to retract or explain their idiocy.
ovf has joined #yosys
<daveshah>
We shall see
thoughtpolice has joined #yosys
<daveshah>
A few people have contacted them now
benreynwar has joined #yosys
lukego has joined #yosys
futarisIRCcloud has joined #yosys
<awygle>
the messaging of "you were supportive then you stopped, so now we'll punish you, even though we haven't punished people who were never supportive" seems a bit iffy to me. but my opinion is moot since realistically i'm not gonna do either of those things myself anyway
<whitequark>
awygle: I have, actually
<whitequark>
even got a C&D for it
<whitequark>
anyway, yes, they will get some time
<awygle>
oh, well then
emeb_mac has joined #yosys
<whitequark>
awygle: I guess the bottom line is that this feels like betrayal, and is harder to tolerate than the approach of someone like Xilinx who has always been openly hostile
<Lofty>
As I posted, it's a trust thing to me at least
<daveshah>
To be fair even Xilinx seem to be moving in the right direction at the moment
<Lofty>
If this gets past without even advanced notice, what stops it from happening again?
<daveshah>
Which makes Lattice behaving like this even more shocking
<awygle>
whitequark: yeah, i can understand that. i come down a different way, like i said, but that take certainly makes sense.
<Lofty>
Meanwhile at intel: *crickets*
<whitequark>
awygle: I'm going to tone it down to "let's see what happens and keep that option in mind as something we can do if they prove their hostility after all and this turns out to not be a misunderstanding."
ktemkin has joined #yosys
<awygle>
:thumbs_up: (i've become an emoji-user... oh dear)
sorear has joined #yosys
flammit has joined #yosys
pointfree has joined #yosys
<daddesio>
I'm late to the discussion, but I've heard chipmakers like working with students (giving samples), because when said students go on to engineering companies they tend to stick with the components they're already familiar with and use them in new designs. I would have never played around with Lattice if it weren't for symbiflow.
<daddesio>
lattice isn't tied to military or automotive, they appear in lots of general devices I think (the new siglent scopes for instance)
<az0re>
Lattice would like to be used in military parts if they are not already (I think they are, but obviously not in the same applications as Xilinx devices)
<az0re>
It's a big market
<az0re>
And don't forget CFIUS refusing a sale to the Chinese
<tpb>
Title: CFIUS blocks sale of Lattice Semiconductor to Canyon Bridge Fund - MarketWatch (at www.marketwatch.com)
<az0re>
"The national security risk posed by the transaction relates to, among other things, the potential transfer of intellectual property to the foreign acquirer, the Chinese government's role in supporting this transaction, the importance of semiconductor supply chain integrity to the U.S. government, and the use of Lattice products by the U.S. government"
adjtm has quit [Remote host closed the connection]