10:26 pendingchaos: imirkin: shouldn't there be a "offset += insn->src(s).get()->reg.data.offset" in TargetNVC0::insnCanLoadOffset() like in TargetNV50::insnCanLoadOffset()?
16:07 imirkin: pendingchaos: yep
19:04 rhyskidd: what is the best programmatic way in envytools to list all the possible "variants" for rnndb?
19:04 rhyskidd: i'm looking to ensure I constrain a mmio range to all pre-Pascal cards
19:06 rhyskidd: is it to look at the table in generated cgen/gpu.c?
19:06 mwk: rnndb-generated/nvchipsets.xml
19:06 mwk: yeah, the .c file works just as well
19:08 rhyskidd: and is the interval notation [a,c) to indicate an interval from a to c that is inclusive of a but exclusive of c?
19:09 mwk: yes
19:10 rhyskidd: so GF100:GP100 is an interval that is inclusive of Fermi but exclusive of Pascal?
19:10 mwk: yes
19:10 rhyskidd: great, thanks mwk :)
20:07 rhyskidd: ^ skeggsb_ you might be interested in the above. nouveau doesn't yet make use of the memory clock command shadow scripts from the vbios that nvgpu uses
20:07 rhyskidd: nouveau does use the non-command scripts at the moment
20:11 Subv: in envytools, is gf100_m2mf.xml the method layout for engine 0xB0B5 in maxwell?
20:12 mwk: no
20:12 mwk: that would be gk204_copy.xml
20:13 mwk: or something like that
20:13 mwk: different engine
20:13 mwk: er, gk104_copy.xml
20:56 Subv: ah, thanks!
23:07 mooch: hey, does anybody know what the different copy modes here do? https://github.com/envytools/envytools/blob/master/rnndb/fifo/gk104_copy.xml#L23
23:08 mooch: also, where in nouveau is the copy mode set? i can't even see any code for this object class
23:24 mooch: oh, damn, nouveau just uses unk2 all the time
23:24 mooch: crap
23:32 mooch: does anybody know what unk1 and copy mode 3 do?