Monday, 2019-01-14

*** tpb has joined #symbiflow00:00
*** TT_ has joined #symbiflow01:49
*** TT_ has quit IRC02:54
*** citypw has quit IRC03:08
*** citypw has joined #symbiflow03:47
*** _whitelogger has quit IRC05:24
*** _whitelogger has joined #symbiflow05:27
*** Rahix has quit IRC08:21
*** Rahix_ has joined #symbiflow08:21
*** kraiskil has joined #symbiflow08:38
*** kraiskil has quit IRC08:46
*** Rahix has joined #symbiflow08:53
*** Rahix_ has quit IRC08:54
*** kraiskil has joined #symbiflow08:59
nats`litghost, I closed the bug I opened about High ram usage on 072, we could also close the #171 is everything is fine for build directory09:24
*** citypw has quit IRC09:33
*** kraiskil has quit IRC09:40
*** citypw has joined #symbiflow09:51
*** kraiskil has joined #symbiflow09:51
*** citypw has quit IRC09:57
*** kraiskil has quit IRC09:58
*** kraiskil has joined #symbiflow10:11
*** kraiskil has quit IRC12:49
*** _whitelogger has quit IRC14:00
*** _whitelogger has joined #symbiflow14:03
*** kraiskil has joined #symbiflow14:11
*** kraiskil has quit IRC15:06
*** citypw has joined #symbiflow15:07
nats`digshadow, any details about the "part-yaml" ?15:15
nats`I could take a look into that to close the #171 if it's not too complicated15:15
digshadownats`: I think I have a fix already for it15:17
digshadowbut I can direct you to it if yo uwant15:18
digshadowI'm just waiting for a run to complete15:18
nats`if you have the fix go ahead :)15:18
nats`it's just that #171 didn't have any mention of that15:18
nats`only about 07x fuzzer in the end15:18
*** kraiskil has joined #symbiflow15:18
*** mgielda has quit IRC15:19
digshadowyeah its an error in the ticket, not your fault15:21
digshadowI did a quick sweep to make sure and found that15:21
digshadowyours was definitely the hard part though, this is a relatively easy fix by comparison15:21
nats`no problem :)15:22
digshadownats`: odd question, but can you try running the yaml fuzzer as is and see if it works for you15:29
digshadowoh wait I think I know the issue15:29
nats`digshadow, not at homs but I can this evening for sure15:37
digshadowI needed to rebuild tools15:37
*** citypw has quit IRC15:38
nats`if you rerun at one time the full fuzzers set can you check that the results are good for artix and kintex ?15:42
nats`for kitnex I mean sorry15:42
nats`artix should be good15:43
*** Rahix has quit IRC16:10
*** Rahix has joined #symbiflow16:13
*** Rahix has quit IRC17:19
*** kraiskil has quit IRC18:33
nats`litghost, I suggested to change the order of fuzzer but digshadow  had some point against that19:01
litghostWell right now 072-074 take a really long time, I believe that was the objection19:14
nats`oky I'll take a look19:38
nats`I couldn't yesterday19:38
*** somlo has quit IRC19:43
nats`litghost, I'm working on the 074 optimisation I'll do simple and copy what I did for 072 we will see later if more intelligent optimization can be done20:23
nats`anyway if by chance you have the file resulting from a 074 run I want them :)20:24
nats`I'll check my code using them20:24
*** somlo has joined #symbiflow20:28
litghost074 outputs are here
tpbTitle: prjxray-db/artix7 at master · SymbiFlow/prjxray-db · GitHub (at
litghostand site_type_*.json20:56
litghostand tileconn.json20:56
litghostthose are the outputs from 07420:56
nats`there should be acsv too21:03
nats`there should be a csv too21:03
litghostThe CSV is used during processing21:04
litghostit is not output21:04
litghostThe CSV is simply a work index21:05
nats`sure but maybe it'll be easier to use to check if my new code generates error21:07
nats`but let's try with final output first21:07
*** daveshah_ has joined #symbiflow21:11
*** daveshah has quit IRC21:18
*** xobs has quit IRC21:18
*** daveshah_ is now known as daveshah21:18
*** xobs has joined #symbiflow21:19
nats`litghost, wow 074 is worse !21:23
litghostit is :)21:23
nats`there are only 18k tiles but so much call inside that leak everywhere !21:27
nats`I splitted in for but 64 block is too much for my ram I try with 12_21:28
nats`oky corrected a problem in my script22:01
nats`so the first part for tiles seems working and way faster22:01
nats`I need to add the same for nodes and then merge files22:01
litghostsounds good22:02
litghostthe outputs should be json5 files, so it should just be a simply load(a), load(b), dict.merge22:02
nats`I shouldn't need to concat json22:07
nats`only the csv22:07
nats`I work tile by tile22:07
litghostah, sure22:14
nats`15 minute for the tile22:21
nats`I need to do the second part with nodes22:21
nats`may sound stupid but the csv header is important ? :D22:22
litghostthe post-processing code uses csv.DictReader, so yes22:38
nats`oky :)22:39
nats`I'm trying to focus but my stupid neighboor are making a party again -_-22:39
nats`litghost, few things to correct I think I'll PR tomorrow evening on the 07423:49
nats`maybe something like a 3 or 4x gain in speed23:49
nats`and running on my 16GB computer :)23:49
nats`someone will have to check it anyway to be sure23:50
nats`good night folks !23:50
litghostnats: great!23:56

Generated by 2.13.1 by Marius Gedminas - find it at!