03:22 <TD-Linux> there's not really much of a point to a single file opus decoder

03:23 <TD-Linux> there are a lot of dsp instructions you could potentially add to get opus to work on a rv32

03:23 <TD-Linux> see the existing assembly for ideas

03:23 <TD-Linux> sorear, that ti dsp running vorbis is done by having a bunch of hand written assembly using the weird dsp instructions

03:30 <Ultrasauce> dsp core and llvm backend for boneless are obviously the way to go forward here

03:31 <Ultrasauce> but yeah it wasnt really a serious suggestion. cm4 can do it but probably needs 80 mips or something

03:48 <sorear> TD-Linux: the ti dsp was opus, not vorbis, and I assumed exactly as much - a constructive demonstration that a small number of extra instructions are probably enough to do it in a ~40 mips pipeline

06:47 <marcan> just add a decode_opus instruction

06:47 <marcan> problem solved

06:47 <marcan> heck I bet the next IBM mainframe revision will come with that

06:47 <marcan> they already have unicode conversions and have had boneless sprintf as an instruction for decades

06:48 <sorear> It has to fit on a up5k

06:48 <sorear> So the added hardware has a very finite complexity budget

07:02 <marcan> VMDOP - Virtualize Memory and Decode Opus to PCM

07:03 <marcan> GWWGA - Get Whitequark to Write a Glasgow Applet

07:03 <marcan> has anyone started @riscvinstructions yet?

07:05 <whitequark> lmao

11:43 <_whitenotifier-9> [whitequark/Glasgow] whitequark pushed 1 commit to master [+0/-0/±1] https://git.io/fhd5O

11:43 <_whitenotifier-9> [whitequark/Glasgow] whitequark 3526bd4 - access.direct.arguments: fix typo.

11:45 <_whitenotifier-9> [Glasgow] Success. The Travis CI build passed - https://travis-ci.org/whitequark/Glasgow/builds/495945471?utm_source=github_status&utm_medium=notification

11:46 <_whitenotifier-9> [whitequark/Glasgow] whitequark pushed 1 commit to master [+1/-0/±1] https://git.io/fhd5C

11:46 <_whitenotifier-9> [whitequark/Glasgow] whitequark 25aa227 - hardware: revB resistors did not really fix the FXMA brokenness :(

11:49 <_whitenotifier-9> [Glasgow] Success. The Travis CI build passed - https://travis-ci.org/whitequark/Glasgow/builds/495946591?utm_source=github_status&utm_medium=notification

11:49 <whitequark> so yeah revB is as fucked as revA

14:55 <apo> D:

14:56 <apo> oh hey

14:56 <whitequark> what

14:57 <whitequark> marcan wasn't involved in revB and his work is excellent

14:57 <marcan> you can kick him for me :p

14:57 <whitequark> what's his problem i don't get it

14:57 <apo> just saying hi

14:57 <marcan> he's just trolling

14:57 <marcan> I know him

14:57 <whitequark> fine

15:50 <_whitenotifier-9> [whitequark/Glasgow] whitequark pushed 1 commit to master [+0/-0/±1] https://git.io/fhdpe

15:50 <_whitenotifier-9> [whitequark/Glasgow] whitequark 4653a6a - applet.hd44780: require revC0+.

15:52 <_whitenotifier-9> [Glasgow] Success. The Travis CI build passed - https://travis-ci.org/whitequark/Glasgow/builds/496060806?utm_source=github_status&utm_medium=notification