| *** psydroid <psydroid!~psydroid@user/psydroid> has quit IRC | 00:14 | |
| *** charlie5 <charlie5!~charlie5@2001:8003:368f:ed00:df8a:6ef8:2410:1c7e> has joined #libre-soc | 02:55 | |
| *** charlie5 <charlie5!~charlie5@2001:8003:368f:ed00:df8a:6ef8:2410:1c7e> has quit IRC | 04:40 | |
| *** rsc <rsc!~robert@fedora/rsc> has quit IRC | 05:51 | |
| *** rsc <rsc!~robert@fedora/rsc> has joined #libre-soc | 05:51 | |
| *** markos_ <markos_!~markos_@static062038151250.dsl.hol.gr> has joined #libre-soc | 07:37 | |
| *** markos_ <markos_!~markos_@user/markos/x-1838887> has joined #libre-soc | 07:37 | |
| *** gnucode <gnucode!~gnucode@user/jab> has quit IRC | 08:15 | |
| lkcl | programmerjake: please go back and re-read the entirety of my responses under the bugtracker for all the inter-related bugs. | 08:20 |
|---|---|---|
| lkcl | paying particular attention to SVP64 specification writing being the absolute top priority. | 08:21 |
| lkcl | yes this will need you to re-read pretty much everything. this is normal: i used to do it day-in day-out re-reading the same bugreports approximately 20-30 times a day | 08:25 |
| lkcl | as i cannot do that now, and keep forgetting (it drops out of short-term memory recall), you need to do it, and i can then come in *once*, and do a full read-through | 08:26 |
| lkcl | you need to pay attention to: | 08:27 |
| lkcl | * what i said under the InOrder Core bugreport | 08:27 |
| lkcl | * what i said about TestIssuer | 08:27 |
| lkcl | * what i said about DD-FFirst and Vector-Immediates | 08:27 |
| lkcl | * what i said about LD/ST EXTRA-322 | 08:28 |
| lkcl | and you'll find that there's *at least* (if not more) EUR 3,000 worth of under-budgeting for those four tasks | 08:29 |
| lkcl | making attempting to re-assign to FPU completely inappropriate (especially given that it's been very much underestimated: EUR 8,000 allocated when it's about 24,000 worth of work) | 08:30 |
| lkcl | therefore it should be blindingly-obvious that putting the "groundwork" in place in the FP side to put in *another* grant should be the priority. | 08:31 |
| lkcl | but definitely *not* higher than the SVP64 specification (that's absolute ABSOLUTE top priority) | 08:31 |
| lkcl | and yes, if there's a change to the SVP64 specification then: | 08:32 |
| lkcl | * 1) it needs documenting | 08:32 |
| lkcl | * 2) it needs implementing in TestIssuer | 08:32 |
| lkcl | * 3) which means it needs implementing in CSV files, updating insndb, and updating consts then | 08:33 |
| lkcl | * 4) it needs unit tests (in insndb as well as test_caller_*) and then | 08:33 |
| lkcl | * 5) it needs a budget for binutils to be adapted | 08:33 |
| lkcl | there is now one HELL of a lot of dependencies even from one "simple" seeming change | 08:34 |
| lkcl | all of which requires budget time and planning | 08:34 |
| lkcl | and you (i mean all of you) need to be aware of that, know the process, know what is involved, what the chain is, what the costs are. | 08:35 |
| lkcl | * 6) all of the above needs tracking in inter-dependent bugreports | 08:36 |
| lkcl | sorry (2) ISACaller | 08:38 |
| lkcl | then finally when (4) is correct: | 08:38 |
| lkcl | * (7) implement in TestIssuer | 08:39 |
| lkcl | which brings in *another* cascade of dependencies i won't go into | 08:39 |
| *** ghostmansd <ghostmansd!~ghostmans@broadband-90-154-80-239.ip.moscow.rt.ru> has quit IRC | 08:51 | |
| *** ghostmansd <ghostmansd!~ghostmans@broadband-90-154-80-239.ip.moscow.rt.ru> has joined #libre-soc | 09:50 | |
| *** ibot <ibot!~supybot@2a00:1098:82:f::1> has quit IRC | 10:28 | |
| *** ibot <ibot!~supybot@libre-soc.org> has joined #libre-soc | 10:28 | |
| *** psydroid <psydroid!~psydroid@user/psydroid> has joined #libre-soc | 10:35 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-90-154-80-239.ip.moscow.rt.ru> has quit IRC | 11:24 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@176.59.164.251> has joined #libre-soc | 11:25 | |
| lkcl | i've put a TODO list at the top of #961's top-level description https://bugs.libre-soc.org/show_bug.cgi?id=961#c0 | 12:06 |
| lkcl | the FP pipeline's tasks need to be REDUCED *NOT* increased, down to the absolute bare minimum, in order to give budget to much higher priorities | 12:08 |
| *** psydroid2 <psydroid2!~psydroid@user/psydroid> has joined #libre-soc | 13:07 | |
| *** psydroid <psydroid!~psydroid@user/psydroid> has quit IRC | 13:14 | |
| *** psydroid <psydroid!~psydroid@user/psydroid> has joined #libre-soc | 13:16 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@176.59.164.251> has quit IRC | 13:24 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@176.59.54.232> has joined #libre-soc | 13:26 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@176.59.54.232> has quit IRC | 13:48 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-90-154-80-239.ip.moscow.rt.ru> has joined #libre-soc | 13:49 | |
| openpowerbot | [irc] <sadoon[m]1> markos Arnisio Kotsi Me Briam | 14:16 |
| openpowerbot | [irc] <sadoon[m]1> Is this a famous recipe in Greece? | 14:16 |
| openpowerbot | [irc] <sadoon[m]1> I'm having one in a Greek restaurant, quite the meal | 14:17 |
| *** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has quit IRC | 15:54 | |
| ghostmansd | Damn, again this MSB0. Assuming we have field with bits [5, 7, 13], where the sign would be? 5 or 13? | 15:58 |
| *** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has joined #libre-soc | 16:05 | |
| markos_ | sadoon, hahaha | 16:35 |
| markos_ | yes | 16:35 |
| markos_ | lamb is indeed very popular, in various recipes | 16:35 |
| openpowerbot | [irc] <sadoon[m]1> It was quite good | 16:35 |
| markos_ | lamb is my favourite that's for sure | 16:35 |
| openpowerbot | [irc] <sadoon[m]1> So I'm researching what we can do re: RAM | 16:53 |
| openpowerbot | [irc] <sadoon[m]1> there's many more solutions than we had in mind | 16:54 |
| openpowerbot | [irc] <sadoon[m]1> One example specifically targetting slow clocks | 16:55 |
| openpowerbot | [irc] <sadoon[m]1> https://github.com/ultraembedded/core_ddr3_controller | 16:55 |
| markos_ | sadoon, this is fantastic | 17:43 |
| openpowerbot | [irc] <sadoon[m]1> Keep in mind this specific repo is not licensed | 17:47 |
| openpowerbot | [irc] <sadoon[m]1> But just as an example | 17:47 |
| lkcl | ghostmansd, 3 bits --> numbered MSB0 from the left --> do [31-5, 31-7, 31-13] and that "gets rid of" MSB0 | 18:12 |
| lkcl | of course if you have already extracted them to 3 bits MSB0-numbered [0, 1, 2] then to make them LSB0 you want [2-0, 2-1, 2-2] ==> [2,1,0] | 18:13 |
| lkcl | which means that the sign bit *in each case* will *still be the first bit* | 18:14 |
| lkcl | [SIGN BIT, ... other bits ...., LOWEST-ARITHMETIC-BIT] | 18:14 |
| lkcl | LSB0 / MSB0 does *not* turn the *order* of the bits around... just the numbering of them | 18:15 |
| *** psydroid2 <psydroid2!~psydroid@user/psydroid> has quit IRC | 21:08 | |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!