Monday, 2023-04-10

*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has joined #libre-soc00:08
*** libredev <libredev!> has joined #libre-soc06:54
*** octavius <octavius!> has joined #libre-soc07:35
*** markos <markos!> has quit IRC09:05
lkcloctavius, great feedback on ls012 - look in the original csv file optable.csv09:21
lkclbut also, run the python script manually to generate the markdown tables, it's not automatic09:22
lkclare you looking at the version where i've added the register profiles?09:24
lkclbear in mind i'm doing 20+ edits a day sometimes so please make sure you always pull and rebuild then check that what you're about to write is still relevant09:25
*** markos <markos!> has joined #libre-soc10:06
*** ghostmansd <ghostmansd!> has joined #libre-soc11:13
lkcloctavius, done
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC12:46
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc12:46
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC13:05
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc13:13
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC13:25
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc13:25
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC13:32
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc13:32
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC15:30
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc15:31
*** octavius <octavius!> has quit IRC16:16
*** ghostmansd <ghostmansd!> has quit IRC17:35
*** ghostmansd <ghostmansd!> has joined #libre-soc17:36
ghostmansdlkcl, it seems that I cannot avoid doing at least some of the scope of task 1003 in scope of 976. One thing pulls another, that one pulls yet another one, and so on...17:43
ghostmansdIn order to make things consistent, I have to update headers and sources which we generate...17:43
ghostmansdAnd, well, this beats the shit out of modes we deleted in Python classes... :-D17:43
ghostmansdThis brings changes to disassembler and assembler... sigh.17:44
ghostmansdI'll try to keep the scope as tigh as possible, but some things are simply inevitable, unless I rewrite parts of autogenerated code, and this is certainly not the way I want to follow.17:44
lkclghostmansd, well, it is what it is.18:11
lkclbtw do see i updated power_insns.py18:12
lkcland i also just removed (yesterday) RC1 mode from LD/ST DD-FFirst18:13
ghostmansd[m]lkcl, RC1 changes... They are not yet pushed, are they?18:15
ghostmansd[m]The last commit I see from you is Apr. 618:16
lkclspecification page - which is always the first place i edit (and then re-read 20-30 times) - is always pushed first18:27
lkclas i made the removal only last night from the specification page, i haven't got round to editing power_insn.py18:30
lkcl(or updating which is always a pig, sigh)18:30
ghostmansdah OK, I'll deal with it later then18:31
ghostmansdI'm updating modes tables18:32
programmerjakelkcl: i'm planning on updating the list of fptrans insn priorities today, some insns you marked as lower priority actually need to be high priority (e.g. min/max insns because min/max reductions need them and are pretty common)18:33
programmerjakealso cuz fmin/fmax take like 10 insns to implement properly which is a lot18:35
lkcli went off of the transcendentals page, and that wasn't on it.18:36
lkclso please do update the transcendentals page *first* / simultaneously18:37
programmerjakeuuh, it should be, iirc i added it...18:37
programmerjakenot the fmin/max expense estimate, but the fmin/max insns themselves should be18:37
lkclyes you'd added them, no you hadn't added what their priority is (which "optional subset")18:38
*** octavius <octavius!> has joined #libre-soc18:41
lkcland guess where i copied the priorities from into the csv file? :)18:41
programmerjakeimho they're quite cheap to implement in hw (about comparable to fp compare + some muxes) so should be in a subset on their own, most the other fptrans ops are waay more expensive18:43
programmerjakemin/max and other newer will also need to be added to the lists of which insns other gpus have18:50
programmerjakenewer ops18:50
programmerjakeoh, do note that pow is commonly used in 3d graphics for computing specular highlights, so does need to be fast18:52
programmerjakeit's even in opengl 1.018:59
programmerjakewell, actually which is closely related19:06
*** knar <knar!~knar@2405:201:d032:9048:9a3:fb59:f4af:1d38> has joined #libre-soc19:36
knarhi again,19:43
knarI was going through this [] for compiling cavatools but ig the repos it links to has restructured a bit which i handled but i keep getting make errors. so ig my question would be is that the best link to get cavatools19:43
openpowerbot_[slack] <github> signin19:43
*** knar <knar!~knar@2405:201:d032:9048:9a3:fb59:f4af:1d38> has quit IRC19:48
*** octavius <octavius!> has quit IRC20:08
lkclknar, as best i know, yes.20:28
lkclprogrammerjake, do put that all into the notes. every instruction needs justification, and that's good stuff20:29
programmerjakewill do, later today20:43
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has quit IRC22:10
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has joined #libre-soc22:16

Generated by 2.17.1 by Marius Gedminas - find it at!