Tuesday, 2019-01-08

*** tpb has joined #symbiflow00:00
*** citypw has joined #symbiflow03:34
*** litghost has quit IRC05:35
*** kraiskil has joined #symbiflow07:46
nats`good morning :)08:07
*** sorear has quit IRC08:11
*** daveshah has quit IRC08:12
*** sorear has joined #symbiflow08:12
*** daveshah has joined #symbiflow08:12
*** Jim_GM0UIN has joined #symbiflow08:22
*** JimGM0UIN has quit IRC08:24
*** kgugala has joined #symbiflow08:29
*** tmichalak has joined #symbiflow08:29
*** kraiskil has quit IRC08:30
*** acomodi has joined #symbiflow08:31
*** mgielda has joined #symbiflow08:33
digshadowhey nats`09:03
digshadowHows it going?09:03
digshadowIf someone is looking for an entry level prjxray issue, this one is probably easy to fix: https://github.com/SymbiFlow/prjxray/issues/45009:03
tpbTitle: Missing *LUT bits in DB · Issue #450 · SymbiFlow/prjxray · GitHub (at github.com)09:03
nats`fine short night but good :)09:04
nats`I'll continue on fuzzers this evening09:04
*** sxpert has quit IRC09:54
*** sxpert has joined #symbiflow09:54
*** citypw has quit IRC10:04
*** Rahix has joined #symbiflow10:27
*** Rahix has quit IRC10:43
digshadownoopwafel: https://github.com/SymbiFlow/prjxray/pull/45510:52
tpbTitle: addrwidth utility by mcmasterg · Pull Request #455 · SymbiFlow/prjxray · GitHub (at github.com)10:52
digshadowFYI those utility functions should be useful for your project10:52
*** _whitelogger has quit IRC11:51
*** _whitelogger has joined #symbiflow11:54
*** kgugala has quit IRC11:59
*** Rahix has joined #symbiflow12:29
*** Jack__ has joined #symbiflow14:22
*** kgugala has joined #symbiflow14:36
*** tmichalak has quit IRC15:24
digshadowmithro: htmlgen is un-borked15:46
digshadowif we can convince ourselves that database is reasonably stable, might be worth it to regenerate15:46
digshadowthere are a few outstanding issues (mostly related to db conflicts), but the bulk is working again15:47
digshadowmithro: https://photos.app.goo.gl/vw3VpEwwNBJQQsPu615:49
digshadowtop of tree using download latest15:50
digshadowthe red is from: https://github.com/SymbiFlow/prjxray/issues/45615:53
tpbTitle: Different bit names on the same bit position in CLBLL · Issue #456 · SymbiFlow/prjxray · GitHub (at github.com)15:53
mithroI'm still waking up16:18
mithroBut I did start playing with the cloud build stuff that Rick setup a long time ago16:19
mithroI also kicked off a manual build last night16:24
mithrodigshadow: it would be good to do some gardening of the prjxray issue tracker16:36
*** halosghost has joined #symbiflow16:42
halosghosthowdy, folks16:42
mithroHi halosghost16:53
*** litghost has joined #symbiflow16:58
*** Rahix has quit IRC17:00
*** acomodi has quit IRC17:05
*** halosghost has left #symbiflow17:44
nats`ahhh SLICEL doesn't have the "dynamic write port" that's why it can be used only for combinatorial logic !18:08
nats`oups WW sorry18:09
*** kgugala has quit IRC18:30
*** Jack__ has quit IRC18:40
mithrodigshadow: Should 051-imuxlout be working?18:49
nats`I think I made my first patch for fuzzers, is there a way to have someone to test the patch and make me some remark about it to know if it's good enough ?18:59
tpbTitle: All output products should go into "build" dir · Issue #171 · SymbiFlow/prjxray · GitHub (at github.com)19:00
nats`I have a problem with the Fuzzers 072 it seems it get stuck but maybe it's just Vivado not outputing message after some point.20:08
nats`INFO: [Common 17-14] Message 'Vivado 12-2683' appears 100 times and further instances of the messages will be disabled. Use the Tcl command set_msg_config to change the current settings.20:08
nats`mithro digshadow any hint on that ?20:08
nats`(vivado still running 100% in top)20:09
nats`(eating 11GB of ram now)20:25
nats`the process got killed after going over ram limit at 16GB20:35
litghostnats: Ya, fuzzer 072 consumes a lot of memory.  16 GB is typically enough, but we typically run it on a 32 GB or 64 GB system.  I believe there is an issue for attempting to lower the memory usage.20:56
nats`oky feel free to close my issue on the github so20:57
nats`I'll buy few ram stick to reach 3220:57
litghostFeel free to take a shot at lowering the memory usage.  The tcl code was written in a very simple way, but I believe with more care the peak memory usage could be lowered.  This work is likely important when running fuzzer 072 on larger parts.20:58
nats`I'll take a look but I think it's way above my knowledge for now :)20:59
nats`I'm patching simple things for now20:59
nats`anyway in the log vivado is giving some hints apparently20:59
nats`WARNING: [Vivado 12-2548] 'get_pips' without an -of_object switch is potentially runtime- and memory-intensive. Please consider supplying an -of_object switch. You can also press CTRL-C from the command prompt or click cancel in the Vivado IDE at any point to interrupt the command.20:59
litghostThe relevant tcl script is https://github.com/SymbiFlow/prjxray/blob/master/fuzzers/072-ordered_wires/generate.tcl21:00
tpbTitle: prjxray/generate.tcl at master · SymbiFlow/prjxray · GitHub (at github.com)21:00
litghostIt's actually pretty small and understandable, but you would need to understand tcl to work on it21:00
nats`I understand tcl but not really what it call in vivado batch mode but I'll take a look just after finishing the patch of build process for fuzzer21:01
nats`https://github.com/SymbiFlow/prjxray/issues/465 <= can we use this issue to work on that so ?21:02
tpbTitle: Fuzzer 072-ordered-wires RAM explosion · Issue #465 · SymbiFlow/prjxray · GitHub (at github.com)21:02
nats`if you agree I can add a comment with a brief of what we just discussed here21:02
nats`mithro, why did you remove the -j to fuzzer ?21:10
litghostYa, that issue is fine21:23
nats`uhhmmmm something is boring me https://github.com/SymbiFlow/prjxray/issues/17121:26
tpbTitle: All output products should go into "build" dir · Issue #171 · SymbiFlow/prjxray · GitHub (at github.com)21:26
nats`why the issue reference all the different commit I amended in my fork21:26
litghostBecause you have the #171 in the commit description21:29
litghostGithub is being "smart"21:29
nats`but........ I made a lot of testing on my fork exactly to not pollute the main one...21:30
nats`Am I wrong or this is really stupid ?21:30
nats`(yep never really used for working on big projects)21:30
litghostNothing to worry about21:32
nats`oky thanks you're reassuring me :)21:33
nats`for pull request should I make them atomic to the smallest commit possible or group some higher level commit ?21:33
nats`for example i'm fixing different fuzzer build process, should I PR each fuzzer fix or all at once ?21:34
nats`litghost, if I'm understanding correctly it creates a list of the PIPS in two file uphill and downhill22:03
nats`but it doesn't flush while running22:03
litghostIt's unclear where the memory usage is going.  If adding a flush in the loop solves the issues, great.  But that seems wrong to me, as typical buffered file writers flush on newline or # of characters.22:04
nats`I'll dig a little more to understand that22:05
nats`but for what I saw the file are empty while the loop is running22:05
nats`I need to recheck that22:05
litghostWorth trying adding a flush then22:06
tpbTitle: Tcl Built-In Commands - flush manual page (at www.tcl.tk)22:06
nats`but does this mean file will be 16GB+ ?22:06
litghostI'm showing ~2.7G and 3.6G22:07
litghostIf all of that was buffered, that might be enough to cause an OOM on top of the typical vivado usage22:08
nats`ah nop sorry it doesn't stay empty it has crashed before22:08
nats`seeing the size of that vivado stuff I think it's not doable to try a valgrind22:08
litghostYou can try https://www.tcl.tk/man/tcl8.4/TclCmd/memory.htm#M11 "memory usage" to debug it22:10
tpbTitle: Tcl Built-In Commands - memory manual page (at www.tcl.tk)22:10
nats`thanks for the hint I'll take a look22:11
*** mrhat2010[m] has joined #symbiflow22:14
nats`https://webcache.googleusercontent.com/search?q=cache:CsB7PWFna0QJ:https://forums.xilinx.com/t5/Vivado-TCL-Community/Is-it-possible-to-navigate-through-a-path-automatically-in/td-p/757323+&cd=5&hl=fr&ct=clnk&gl=fr I'm wondering why this page is no more available on the website directly22:26
tpbTitle: Solved: Is it possible to navigate through a path automati... - Community Forums3rd Party Header & Footer (at webcache.googleusercontent.com)22:26
nats`when trying a little in tcl shell I see some interesting things22:35
nats`foreach pip [get_pips] {} <= is not really a problem it consume about 2.3GB but foreach pip [get_pips] {foreach dnode [get_nodes -downhill -of_object $pip] {}} <= seems to already be a a big problem22:39
nats`maybe because dnodes is never freed22:39
nats`I'll try something using intermediate value22:39
digshadownats`: back for a bit sorting through things22:51
digshadowfeel free to ping me again if you need something asap, otherwise will sort through messages as I can22:52
nats`no problem22:52
mithronats`: I removed the -j because it was failing, will figure out how to reenable once I figure out why it failed...23:33

Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!