Monday, 2022-07-04

lkcloctavius, mooornin. you saw i added a part ii to the pdf?10:37
octaviusmorning lkcl, yes it's quite impressive!10:38
lkclneeds an inline-include program which does the same job as ikiwiki [[!inline ...]]10:40
octaviusinline-include? What are you including?10:40
lkcl[[!inline pagenames="gf_reference/" raw="yes"]]10:44
octaviusAh, you can pull in text files directly, cool10:45
lkcli'm going to pre-process it, it's easier10:47
lkcloctavius, do put in RFPs *today*10:55
lkclcesar, ^10:55
lkclprogrammerjake, ^10:55
octaviusregardless of completion?10:56
programmerjakeI'll do it when I get up's 3am now...10:56
lkclprogrammerjake, ack :)10:56
lkcloctavius, no, only if completed10:56
lkcloctavius, can you update the link-creating-program for the pdf?11:05
lkcladd [[sv/vector_ops]] for example11:08
lkclsee simple_v_spec.tex for the names of chapters11:08
lkcloh - after doing RFPs :)11:08
cesarlkcl: I really did some work (FSM CompUnit proof) on , so I will send a RFP for that one. For the  MultiCompUnit proof ( ) I'd like to work a few more days on it. But I can send a RFP anyway, if you prefer.11:36
programmerjakeI probably got fma working for f16 and round towards zero...bitwuzla's been running for 8min so usually stops quickly if the HDL is broken somehow11:36
lkclcesar, no don't send in an RFP for work not already done! that's one of the conditions of the NLnet grants, things have to be 100% completed12:08
lkclprogrammerjake, wha-hey!12:08
cesarGot it, but then I'd like to work on them for a few more days. Won't be long, but it's not today, sorry.12:15
lkclcesar, no problem - do put in 342 though12:30
programmerjakegn all12:35
lkclprogrammerjake, night12:37
lkcloctavius, ya get the general drift?12:37
octaviusYeah, want me to do the rest?12:40
* octavius afk for 15 min12:41
lkclyes please12:45
lkcland add more to Part III12:45
octaviuslkcl, do you want me to add all the remaining pseudo code mdwn files from isa/ dir?14:00
octaviusI noticed a few you had in Part II (svfparith and svfixedarith)14:00
octaviusI'll re-arrange the Makefile rules to follow the contents order, otherwise it gets pretty confusing XD14:01
lkcloctavius, yes please, add the *existing* Power ISA ones from isa/ dir14:11
lkclthey're listed oooonn....14:11
octaviusAlso I noticed fcvt is unused (SV FP Conversion)14:11
octaviusthanks lkcl, will do14:12
lkclyes. don't add that14:14
lkcloctavius, git pull, i just added a define for ../../openpower-isa/....14:16
lkclthe Makefile should now work *without* having to build absolutely everything14:17
lkclif one single .mdwn file is modified it should pull in only what's needed14:17
lkclbut it's not perfect, because of some of the [[!inline]] includes14:17
octaviusThat's what I thought you were planning14:17
lkclthere is a way to deal with that, by getting the ./ program to actually output the files it detects14:18
lkclbtw don't move the svfixedarith / svfparith to Part III14:19
* lkcl added bitmanip pseudocode14:26
octaviuswait, I thought I'm meant to add them?14:27
octaviusI'm gonna have git fun again14:29
octaviusI've set rebase to true14:32
lkclthat one i deliberately added into Part II *not* Part III14:36
lkclbecause it's new.14:37
lkclPart III is *only* for *existing* Power ISA v3.0B14:37
lkclthe list of which is on14:37
lkclbcd .... system14:37
lkcloh and fpload ... fpcvt14:37
octaviusPhew...324 pages XD15:24
lkcloctavius, i know :)15:27
lkclthat's very deliberate, to include everything15:27
octaviusYes ;)15:33
octaviusAdded all the instructions from the spec15:42
octaviusI'll go for walk, gotta give my eyes a break :)15:42
tinybronca[m]This may be a bit OT, but I have a general Power ISA question... do any Power ISA CPUs made by anyone have loadable microcode like intel CPUs have?18:54
lkcltinybronca[m], hi, there's only IBM POWERn/nn processors available19:54
lkclif you ask in either #talos-workstation or #openpower you'd be more likely to get an accurate answer19:55
tinybronca[m]I did not know of these channels, thank you!19:55
lkclbecause there will be people there with more knowledge of IBM's specific product lines - the POWER8, POWER9 and POWER1019:56
lkclthis is also a gateway to them:
lkclnot #talos-workstation though, that's for raptor TALOS-II systems19:57
lkcloctavius, now you know why my eyes went prism!20:02
octaviusOh you mean the computer overuse?20:03
octaviusLuckily for me, my eyes get strained very quickly, so I have to take breaks sooner20:03
octaviusPaper's lovely :D20:04
lkclrequired for the [[sv/av_opcodes]]20:04
lkclmooore muhahahah20:04
lkclcan you put that (and others) into
lkclineptly-named program, i know20:04
lkcl [[sv/vector_ops]]20:05
octavius* (I was thinking about that name earlier....) *20:05
octaviussure, will do20:05
octaviusYou know what I just realised lkcl? We need to add a dependency on pandoc_img, otherwise the .tex files won't regenerate20:16
octaviusDo I have to add this dependency to every file?20:16
octavius* rule (not file)20:16
octaviuslkcl, can you check why the av_opcodes link is not working? Last time I added links I didn't really understand what the proper link names are supposed to be20:20
octavius"Just worked" for most of the links I added before XD20:20
octaviusI've seen that mv_swizzle and mv_vec became mv.swizzle etc. but not sure what needs to be done for the av_opcodes21:04
octavius...I'm a numpty21:05
octaviusAdded the pandoc_img dependency to pseudo_bitmanip (not bitmanip, which has a link to av_opcodes)21:09
lkcloctavius, yyep ya do22:07
lkclgenerate the latex then inspect it, you'll see an auto-generated hyperlink.  just copy that.22:08
octaviusbut now has an error (Error in $: Failed reading: not a valid json value)22:08
lkcldeep joy. ya broke it!22:09
lkclremove whatever-you-added, one line at a time.22:09
octaviusI didn't make changes in that area. The difference is that the bitmanip rule is now being executed22:11
octaviusAnd looking at the code, and I don't understand why JSON is involved. Are you converting the mdwn to json?22:12
octaviusJust tried converting another file, the has the same issue22:16
lkclthat's how pandoc filters work22:18
lkcl$ pandoc --version22:19
lkclCompiled with pandoc-types, texmath, skylighting
lkcli'm not getting any errors22:20
octaviuspandoc 2.2.1 Compiled with pandoc-types, texmath 0.11.1, skylighting 0.7.522:20
octaviusI'm using debian 10 as well22:21
lkclcan you back-track through the commit history, git log git checkout blahblah22:22
lkcltomorrow. 22:30 now22:25
octaviusOk, I just git reset --hard to my latest commit. It now generates the text files22:25
octaviusNot sure what happened there, some junk files not cleared?22:25
octaviusYou mean the meeting tomorrow? Or is there something else?22:26
lkclit's late22:28
octaviusAh ok, gn22:29

Generated by 2.17.1 by Marius Gedminas - find it at!