whitequark changed the topic of #prjbureau to: Microchip (Atmel) ATF15xx CPLD reverse engineering · code https://github.com/whitequark/prjbureau · docs https://whitequark.github.io/prjbureau/ · logs https://libera.irclog.whitequark.org/prjbureau
<whitequark> yes
<balrog> whitequark: with the RE nearly done, what is the next step — nextpnr integration?
<whitequark> i decided i won't work on CPLD synthesis and pnr
<whitequark> there should be at least some hill i'm not dying on
<balrog> That's fair — I'm assuming that it would require a fair amount of work to get symbiflow/nextpnr to work well with CPLDs
<whitequark> nextpnr isn't suited for CPLDs anyway
<balrog> ahhhh
<balrog> hmm... Atmel's WinCUPL comes with a fitter that takes EDIF input
<whitequark> yes it's used during RE
<whitequark> it's very bad
<balrog> :(
<balrog> (it seems like Xilinx and Altera are both abandoning their CPLD lines, and no one else seems interested in tooling at all)
<whitequark> yes
<whitequark> Microchip will make ATFs for as long as anyone buys them
<whitequark> and there is a surprisingly large market for those
<balrog> Microcip is the sole remaining manufacturer of PAL/GAL compatible chips and 27-series EPROMs
<balrog> (well, OTP PROMs, but same idea)
<balrog> Microchip*
<balrog> I need to clean my 'h' key
<balrog> are a lot of the buyers just using them to replace Altera MAX 7000 CPLDs with pof2jed?
<whitequark> not sure
<whitequark> but consider Microchip policy
<whitequark> you can still buy ATF15xxBE chips, they'll just have to make them for you on demand
<whitequark> for ATF15xxAS, they have a line from 1999 or so that does basically nothing other than these ICs
<balrog> I guess there isn't enough demand for them to have rochester electronics to stock BEs
<whitequark> and it gets quite a lot of use, look at Mouser stock for them
<balrog> how does the Atmel fitter compare to e.g. using old Quartus + POF2JED?
<whitequark> never used either
<whitequark> the only thing i used the Atmel fitter for is RE
<balrog> I see
<whitequark> <whitequark "for ATF15xxAS, they have a line "> so they don't plan to take them out of production, there's nothing else they can make on that line
<whitequark> and they have enough customers that it's economically sensible
<whitequark> (i asked a friend to talk to a Microchip distributor)
<whitequark> for ATF15xxBE they have 1 (one) customer
<whitequark> so you can only buy them via microchipdirect
<balrog> do they sometimes have extra stock so that you can get a few for testing?
<whitequark> also, apparently microchip is interested in getting *more* customers for ATF15xxAS
<balrog> the way to do that is to fund an open fitter so that people can use them with verilog
<whitequark> yes. so if someone makes a modern toolchain you may easily convince microchip to promote it
<balrog> there is a market in devices that interfacing with 5V bus
<whitequark> most likely the people who wrote the old fitter are all retired by now
<balrog> I wish Microchip was more FOSS friendly with their CPUs
<whitequark> it's been 20 years for the bulk of the code
<balrog> but I see they went paid-GCC for AVR :(
<balrog> (Atmel worked with avr-gcc upstream)
<balrog> their PIC24 and PIC32 compilers are hacky forks of an ancient GCC version
<balrog> I wonder if it's the same for POF2JED
<balrog> (regarding the authors being retired)
<whitequark> quite possibly
<balrog> though — does the RE here allow taking apart an atf15xx .jed?
<whitequark> yes
<balrog> I could POF2JED a file and then disassemble the JED
<whitequark> i never got around to making a jed2v converter, well, not a complete one
<balrog> I'm screwing around with this POF: http://wiki.pldarchive.co.uk/index.php?title=Kurzweil_Sampling_Option_(SMP-2K/2R)
<whitequark> but it would be not hard
<balrog> has anyone played around with making an open source fitter suitable for CPLDs?
<balrog> (or a framework for one)
balrog has quit [Quit: Bye]