Tuesday, 2023-04-04

*** gnucode <gnucode!~gnucode@user/jab> has quit IRC03:01
programmerjakelkcl: my point in https://bugs.libre-soc.org/show_bug.cgi?id=1015#c28 is that both ls006 and int_fp_mv have large sections that shouldn't be in the other one, so having either include the other doesn't give the desired outcome -- we don't want all of fmvis/fishmv in ls006 so ls006 can't include int_fp_mv -- we also don't want all of the rfc boilerplate in int_fp_mv, so int_fp_mv can't include ls006 either05:44
programmerjakeincluding stuff works fine for ls010 because there aren't large sections of those markdown files that shouldn't be in ls01005:45
*** octavius <octavius!~octavius@> has joined #libre-soc11:13
lkclprogrammerjake, in the "massive kludge" the operands coming in to the pipelines are already sorted out13:49
lkclinversion - of both the input and the output - is covered by the CSV entries13:50
lkclso it can be synthesised down to a spectacularly-small number of lines13:50
lkclbut it'll do for now13:50
lkclall the information *is* right there, in the CSV files, which originally came from microwatt, so you've effectively duplicated (re-discovered, time spent doing so and complexified the code) the work done by the microwatt team13:51
lkcl A-inversion is the "inv A" column13:54
lkcl output inversion13:54
lkcl        is the "inv out" column13:54
lkclcarry-in equal to 0 or 1 or CA is the13:55
lkcl        "cry in" column13:55
lkclall of that is pulled over into info.op_fields (self.instrs[ins_name].op_fields)13:57
lkclwhere the frickin 'ells the bugreport14:03
lkclprogrammerjake, to make finding things easier can i ask you to put the bugreport number into the commit message?14:04
lkcldoesn't have to be every single message, just enough times so that it shows up in "git log" at least approximately close by14:04
lkcli'm having difficulty finding (by keyword search) the ca/ov bugreport14:05
lkclahhh because it got closed14:05
lkcli need to now focus on getting LD/ST Data-Dependent Fail-First in as fast as possible14:09
lkclghostmansd[m], i'll handle the power_insn.py modifications, we'll do a follow-up binutils bugreport later14:10
lkclthe next ISA WG meeting is wed 19th and i absolutely have to have the SVP64 specification submitted before then14:10
lkclghostmansd[m], the format of power_insn.py is really clear, effing good work14:20
ghostmansd[m]Thanks Luke!14:30
ghostmansd[m]I hope I can return to this soon, but I'm really overwhelmed by the regular work :-(14:31
lkclno problem - hey it's good to actually have work14:32
*** octavius_ <octavius_!~octavius@> has joined #libre-soc16:11
*** octavius <octavius!~octavius@> has quit IRC16:13
*** octavius__ <octavius__!~octavius@> has joined #libre-soc16:16
*** octavius_ <octavius_!~octavius@> has quit IRC16:18
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC16:50
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc16:56
*** octavius__ <octavius__!~octavius@> has quit IRC17:10
*** octavius <octavius!~octavius@> has joined #libre-soc17:56
octaviuslkcl, on page 4 of ls010.pdf, the pseudo-code modifies the int_regfile.halfs struct member, however 'halfs' has not been defined in the earlier struct definition17:58
octaviusI guess it's meant to say 'hwords' instead17:59
*** octavius <octavius!~octavius@> has quit IRC18:03
*** octavius <octavius!~octavius@> has joined #libre-soc18:03
octaviusI don't know if my IRC client is not working, but seems the IRC archive hasn't been updating18:04
octaviusOh good, the archives are working. I made a bug comment anyway18:06
*** gnucode <gnucode!~gnucode@user/jab> has joined #libre-soc19:28
ghostmansd[m]Holy cow, it's been a long since I updated svp64 branch for binutils.19:48
lkcloctavius, good catch - it's supposed to be the same phrases used in the Power ISA v3.1 spec19:48
ghostmansd[m]They even got new operand types.19:48
lkclghostmansd[m], :)19:48
ghostmansd[m]I lost a lot of time trying to get what was changed... Hopefully this is rebased now.19:48
lkclfigure 97, p25819:48
lkclmeeting's moved an hour earlier, to 10mins time19:49
ghostmansd[m]They got something creepy called dm registers.19:49
lkclghostmansd[m], niiice. what are those? moo?19:49
ghostmansd[m]Dense Math Registers19:49
lkclcesar, toshywoshy, programmerjake, octavius, jn, markos, sadoon[m] ^19:50
lkclthat'll be Power ISA v3.1 - to do with MMA i'm guessing19:50
*** octavius <octavius!~octavius@> has quit IRC22:13
*** gnucode <gnucode!~gnucode@user/jab> has quit IRC22:25
*** gnucode <gnucode!~gnucode@user/jab> has joined #libre-soc22:31
sadoon[m]Just woke up :(23:43
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC23:45

Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!