| markos | ok, one more question before I hit the bed, I have two maddlds, same elements, but I need the second to multiply the elements *in reverse order*, if first is sum_0^N{A[i]*B[i]}, the 2nd should be sum_0^N{A[i]*B[N-i]} | 00:13 |
|---|---|---|
| markos | is that a case of svindex? | 00:13 |
| markos | basically the second would have A index ascending, but B index descending | 00:14 |
| markos | or is it easier to just reverse B (and how should I do that?) | 00:19 |
| markos | gn | 00:34 |
| lkcl | yeah that'll be an svindex again | 01:00 |
| lkcl | this time you would set rmm=0b00100 (if you wanted to reverse-order on RC) | 01:01 |
| lkcl | you should be able to create the reversed-indices with "sv.svstep/mrr" | 01:04 |
| lkcl | although, i am drumming my fingers here at the number of times inversion is coming up, i think it may be time for a 64-bit version of svshape | 01:05 |
| lkcl | but the last time i tried that it really did not go well | 01:05 |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has quit IRC | 06:58 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has joined #libre-soc | 07:08 | |
| *** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has quit IRC | 08:04 | |
| *** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has joined #libre-soc | 08:05 | |
| *** lkcl <lkcl!lkcl@freebnc.bnc4you.xyz> has quit IRC | 10:10 | |
| *** ghostmansd <ghostmansd!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has joined #libre-soc | 10:12 | |
| *** lkcl <lkcl!lkcl@freebnc.bnc4you.xyz> has joined #libre-soc | 11:36 | |
| markos | lkcl, know what would be good? | 12:01 |
| markos | an SVP64 cookbook, how to solve certain problems using SVP64, showcasing the new instructions | 12:01 |
| markos | eg. how to produce inverse indices | 12:02 |
| markos | I could write already a few entries there based on the functions done so far | 12:02 |
| markos | could be in the form of a knowledge-base, FAQ, or wiki, but preferably something online | 12:03 |
| markos | this is definitely what many developers will look for anyway and what will actually win them over | 12:04 |
| markos | see exactly how it's done, vs reading about the instructions in a spec | 12:04 |
| *** ghostmansd <ghostmansd!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has quit IRC | 12:08 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has quit IRC | 12:27 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@176.59.160.232> has joined #libre-soc | 12:28 | |
| *** octavius <octavius!~octavius@146.147.93.209.dyn.plus.net> has joined #libre-soc | 14:18 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@176.59.160.232> has quit IRC | 15:07 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@176.59.164.111> has joined #libre-soc | 15:08 | |
| octavius | markos, if a an iwiki section hasn't been started, a svp64cookbook.mdwn could be made on the site. Then using the same workflow we have for the pandoc/latex pdf generation to make a portable doc | 15:23 |
| lkcl | yes great idea | 15:24 |
| octavius | To use pandoc_img.py, would it have to be under the "openpower" dir? | 15:25 |
| octavius | In the Makefile rules I see the --filter arg, so I guess a dir could be specified | 15:28 |
| octavius | I don't know if you want to have the cookbook and spec in the same dir, would start to get a little busy? | 15:29 |
| markos | I would put them in separate dirs | 15:29 |
| markos | but great idea | 15:29 |
| markos | and easy to expand | 15:29 |
| lkcl | https://libre-soc.org/openpower/sv/rfc/ - is an automatic "collation" | 15:30 |
| lkcl | so the ideas can go one per page under a subdir | 15:30 |
| lkcl | then "appear" magically in a top-level "listing" page | 15:31 |
| markos | can we have categories as well? | 15:31 |
| markos | eg generic for loops, dot-products, cross-products, how to use predicates, how to use indices, remaps, etc | 15:33 |
| lkcl | as tags, yes | 15:46 |
| *** octavius <octavius!~octavius@146.147.93.209.dyn.plus.net> has quit IRC | 16:19 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@176.59.164.111> has quit IRC | 17:17 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has joined #libre-soc | 17:17 | |
| *** ghostmansd <ghostmansd!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has joined #libre-soc | 17:38 | |
| *** ghostmansd <ghostmansd!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has quit IRC | 17:58 | |
| *** ghostmansd <ghostmansd!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has joined #libre-soc | 18:18 | |
| *** ghostmansd <ghostmansd!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has quit IRC | 18:42 | |
| markos | lkcl, I remember you mentioning graphcore a while ago: https://www.uktech.news/deep-tech/graphcore-microsoft-deal-20221010 | 19:10 |
| markos | basically it's gameover for graphcore | 19:10 |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has quit IRC | 19:29 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@176.59.162.131> has joined #libre-soc | 19:33 | |
| lkcl | that was on the cards pretty much straight away | 19:51 |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@176.59.162.131> has quit IRC | 19:51 | |
| lkcl | anything that specialist / focussed is a serious risk | 19:51 |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has joined #libre-soc | 19:52 | |
| *** jab <jab!~jab@courtmarriott2.wintek.com> has joined #libre-soc | 20:30 | |
| *** josuah <josuah!~irc@46.23.94.12> has quit IRC | 21:34 | |
| *** josuah <josuah!~irc@46.23.94.12> has joined #libre-soc | 21:36 | |
| *** octavius <octavius!~octavius@230.147.93.209.dyn.plus.net> has joined #libre-soc | 23:09 | |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!