| markos | lkcl, reg. what you asked yesterday about running just one test, just remembered, with gtest you can use a filter to run just one test so no need to actually make a separate binary | 09:09 |
|---|---|---|
| markos | you can give ./libvpx_variance_test --gtest_list_tests | 09:10 |
| markos | and it will list all available tests | 09:10 |
| markos | and you can run it with --gtest_filter=SVP64/VpxVarianceTest.Ref/5 for example | 09:10 |
| markos | and it will run just that test | 09:10 |
| markos | that's what I'm doing all the time for development | 09:11 |
| markos | I don't wait hours to test everything with every change :) | 09:11 |
| lkcl | :) | 10:12 |
| lkcl | ghostmansd[m], moornin (ish) | 10:17 |
| lkcl | are ya good with another semi-pseudo-instruction similar to svshape2? | 10:17 |
| lkcl | you remember we removed /SVM because it wasn't working? | 10:18 |
| * ghostmansd[m] salutes | 10:18 | |
| lkcl | i need to add something close to it (horizontal reducton) for markos | 10:18 |
| lkcl | but this time as a REMAP | 10:18 |
| lkcl | say.... svreduce | 10:18 |
| lkcl | it'll actually sit on top of existing Matrix REMAP, just giving some slightly different setup | 10:20 |
| * lkcl still thinking it over but it will allow arrays of R G B data to be "constructed" down to a single thing: | 10:21 | |
| lkcl | R0 G0 B0 -> OR-together-into result0 | 10:21 |
| lkcl | R1 G1 B1 -> OR-together-into result1 | 10:22 |
| lkcl | ... | 10:22 |
| lkcl | Rn Gn Bn -> OR-together-into resultn | 10:22 |
| * lkcl thinks is that even needed? | 10:23 | |
| lkcl | actually... it's just a matrix-style REMAP with a 1D vector, isn't it? | 10:23 |
| lkcl | so yep scratch that :) | 10:25 |
| lkcl | markos, for doing RGB pixel-merging after some shifting, you can use: | 10:26 |
| lkcl | * the RGB data as Matrix A: row 1: R0 G0 B0 row 2 R1 G1 B1 .... | 10:26 |
| lkcl | * a "blank" (vertical) vector of length 3, with zeros in it. | 10:27 |
| lkcl | * an OR instruction as the base operation | 10:28 |
| lkcl | then **IGNORE** the blank vector of length 3, it was only there to set up the REMAP schedule | 10:28 |
| lkcl | i will try a unit test, now | 10:29 |
| ghostmansd[m] | This is again a self-problem-solving chat | 10:33 |
| lkcl | frickin amazing, isn't it? | 10:33 |
| ghostmansd[m] | I guess that we should patent that eventually | 10:33 |
| lkcl | haha | 10:33 |
| markos | :) | 11:07 |
| lkcl | eurrrgh, going to have to use ternlogi because a RT |= RB is needed | 12:10 |
| lkcl | ah well | 12:10 |
| lkcl | lucky we have one of those | 12:10 |
| lkcl | hooray! | 12:19 |
| lkcl | ngggh trying to get my head round matrix-multiplyyyy | 13:05 |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has quit IRC | 13:09 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@176.59.162.167> has joined #libre-soc | 13:10 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@176.59.162.167> has quit IRC | 13:44 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has joined #libre-soc | 13:45 | |
| lkcl | holy cow that was obscure | 14:08 |
| markos | lkcl, btw, I submitted the abstract, just in case... | 14:16 |
| lkcl | brilliant | 14:16 |
| lkcl | i must do mine as well | 14:16 |
| lkcl | i'm just committing the example for horizontal-or-reduction | 14:16 |
| lkcl | it works by ignoring the B-Matrix entirely | 14:18 |
| lkcl | https://git.libre-soc.org/?p=openpower-isa.git;a=commitdiff;h=b58869c4f2efc7ab4a885e3a1de39fda616ddd57 | 14:19 |
| lkcl | just use those 3 instructions as-is. | 14:19 |
| lkcl | it *will* overwrite (calculate) VL. | 14:19 |
| lkcl | if you want to do 4 vectors worth of (R<<16,G<<8,B<<0) data then use | 14:20 |
| lkcl | svshape 3,4,1,0,0 | 14:20 |
| lkcl | if you want to do 8, use | 14:20 |
| lkcl | svshape 3,8,1,0,0 | 14:20 |
| lkcl | markos, i made the example more generic so you can see where to drop in the dimension sizes X,Y | 14:27 |
| lkcl | if you need to OR-reduce ARGB data then it should be obvious | 14:28 |
| markos | great | 14:28 |
| lkcl | oh - you can also use the same trick for horizontal-multiply-reduction | 14:28 |
| markos | I have to finish something on arm first -a few hundred lines dct again :)- and then I'll start on av1 | 14:28 |
| lkcl | as well as horizontal-add-reduction | 14:28 |
| lkcl | ack | 14:28 |
| lkcl | joooy :) | 14:28 |
| markos | already miss svp64 :D | 14:28 |
| lkcl | haha | 14:28 |
| lkcl | sorry :) | 14:29 |
| lkcl | bear in mind that for some unknown reason it's doing *inner* product here not outer-product. | 14:32 |
| lkcl | i have no idea why | 14:33 |
| lkcl | matrix-mul was supposed to do an outer-product | 14:33 |
| lkcl | so as to be able to hit back-end pipelines with a series of non-dependent (interleaved) mul-adds | 14:33 |
| lkcl | i think it's down to the fact that the B-Matrix is (1 x y) dimension size | 14:33 |
| lkcl | oh well | 14:34 |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has quit IRC | 14:38 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@176.59.162.225> has joined #libre-soc | 14:39 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@176.59.162.225> has quit IRC | 14:57 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@176.59.160.214> has joined #libre-soc | 14:58 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@176.59.160.214> has quit IRC | 16:02 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has joined #libre-soc | 16:02 | |
| *** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has quit IRC | 16:51 | |
| *** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has joined #libre-soc | 16:52 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has quit IRC | 17:23 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@176.59.42.233> has joined #libre-soc | 17:24 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@176.59.42.233> has quit IRC | 17:43 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has joined #libre-soc | 17:43 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has quit IRC | 18:19 | |
| *** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-188-32-220-156.ip.moscow.rt.ru> has joined #libre-soc | 18:57 | |
| *** Veera <Veera!~veera@117.243.18.142> has joined #libre-soc | 19:02 | |
| Veera | Hi | 19:02 |
| programmerjake | hi | 19:05 |
| Veera | is lkcl near? | 19:07 |
| programmerjake | lkcl: ^ | 19:18 |
| programmerjake | he was around 1hr ago... | 19:18 |
| programmerjake | idk if he's available rn | 19:19 |
| *** Veera <Veera!~veera@117.243.18.142> has quit IRC | 19:24 | |
| *** Veera <Veera!~veera@117.202.231.204> has joined #libre-soc | 19:25 | |
| *** Veera <Veera!~veera@117.202.231.204> has quit IRC | 19:43 | |
| *** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has quit IRC | 20:15 | |
| *** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has joined #libre-soc | 20:17 | |
| *** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has quit IRC | 20:19 | |
| *** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has joined #libre-soc | 20:20 | |
| *** yambo <yambo!~yambo@69.146.1.110> has quit IRC | 22:23 | |
| *** yambo <yambo!~yambo@69.146.1.110> has joined #libre-soc | 22:35 | |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!