Tuesday, 2019-07-02

*** tpb has joined #symbiflow00:00
hackerfooHacking warning messages to do my work: https://github.com/HackerFoo/symbiflow-arch-defs/commit/9e4aff473a88c2c69d145df4c5e31444f5daa25e#diff-34a17580d74080cc27f07f7f801454a000:28
tpbTitle: WIP · HackerFoo/symbiflow-arch-defs@9e4aff4 · GitHub (at github.com)00:28
*** tweakoz has joined #symbiflow02:01
*** citypw has joined #symbiflow02:17
*** _whitelogger has quit IRC02:26
*** _whitelogger has joined #symbiflow02:29
*** space_zealot has quit IRC05:05
*** citypw has quit IRC05:08
*** proteusguy has joined #symbiflow05:14
*** lethalbit has quit IRC05:50
*** lethalbit has joined #symbiflow06:03
*** noahh has joined #symbiflow06:34
*** noahh has quit IRC06:47
*** citypw has joined #symbiflow06:48
*** Maya-sama has joined #symbiflow07:04
*** Bertl_oO is now known as Bertl_zZ07:13
*** vitamin-q[m] has left #symbiflow07:33
*** kgugala has joined #symbiflow07:45
*** lopsided98 has quit IRC08:03
*** tweakoz has quit IRC09:32
*** OmniMancer has joined #symbiflow09:51
*** Maya-sama is now known as Miyu09:53
*** lopsided98 has joined #symbiflow10:15
*** citypw has quit IRC10:37
*** OmniMancer has quit IRC12:46
*** Bertl_zZ is now known as Bertl13:56
*** Vonter has quit IRC14:17
*** adjtm has quit IRC14:33
sf-slack2<butta> Are others seeing that check rr_graph now takes a very long time with the newest symbiflow-arch-defs?  Last week check_rr_graph was taking 5 seconds and its now taking 95 seconds.14:34
*** alexhw has quit IRC14:39
*** alexhw has joined #symbiflow14:41
*** adjtm has joined #symbiflow14:42
sf-slack2<acomodi> @butta Could it be due to this merged PR? https://github.com/SymbiFlow/symbiflow-arch-defs/pull/85814:48
tpbTitle: cmake: update VTR and add new command line options by acomodi · Pull Request #858 · SymbiFlow/symbiflow-arch-defs · GitHub (at github.com)14:48
sf-slack2<acomodi> Try to use this commit 3948971c48cd3ba1f1ba850bd444df48dbc371a814:49
sf-slack2<butta> @acomodi It's not an issue for us since were commenting out the rr_graph check in source built vpr.  Just wondering if it was a known issue.14:50
duck2does it take 95 seconds _with_ suppressed warnings?14:52
sf-slack2<butta> Yes that's with suppressed warnings14:53
sf-slack2<acomodi> @duck2 I was thinking that it might be the cause14:53
sf-slack2<butta> Although I'm writing to a log file on a harddisk, could try dumping to /dev/null14:53
sf-slack2<acomodi> @butta Could you try to launch VPR without the `--suppress_warnings` line?14:54
duck2acomodi: it used to take 140 seconds with the warnings on. I wonder if it's due to printing/logging or due to the execution path being broken so many times.14:54
sf-slack2<butta> @duck2 interesting, with warnings on it only took 5-10 seconds for us14:56
sf-slack2<acomodi> @butta So it's confirmed that `suppress_warnings` is indeed drastically increasing run-time?14:58
*** proteusguy has quit IRC14:59
duck2acomodi: yes i guess. I took another look at http://duck2.lt/warnings.txt and noticed that check_rr_graph completed in normal time when warnings were directed at a log file. 140 seconds was when it was printing to a terminal.15:01
sf-slack2<acomodi> Ok, IMO, being the warning suppression optional, this is not a huge issue15:06
sf-slack2<acomodi> I will check if there is a way to bring warning suppression to acceptable run-time though15:06
sf-slack2<butta> Just ran some tests.  Writing log file to our HDD array took 85 seconds to create the device.  Writing the log file to a local SSD took 10 seconds to create the device. Printing warning to console took 12 seconds to create device.15:13
sf-slack2<acomodi> @butta Well, that is good news. Thanks for the information. I guess then that if you usually use the HDD instead of the SSD it would be better not to write the suppressed warnings log-file15:17
*** alexhw has quit IRC15:43
*** alexhw has joined #symbiflow15:48
*** acomodi has joined #symbiflow15:51
*** kraiskil has joined #symbiflow16:30
*** proteusguy has joined #symbiflow16:51
litghostacomodi: I think it makes sense to make the suppressed warning file optional (e.g. don't write them at all)16:57
sf-slack2<acomodi> litghost: makes sense17:06
*** kraiskil has quit IRC17:33
sf-slack2<butta> I agree with litghost, we would appreciate having the option to not write to file17:45
*** acomodi has quit IRC18:00
*** tweakoz has joined #symbiflow19:38
*** Miyu has quit IRC19:39
*** proteusguy is now known as proteusdude19:41
*** space_zealot has joined #symbiflow21:03
*** space_zealot has quit IRC21:18
mithroduck2: How's that rr_graph.xml parser going?21:30
duck2mithro: I'm looking more into updating the schema for arch.xml so that I fully know which features could be needed. There are many features of XSD which can aid in fully validating the file. For instance, segment wire_switches have to correspond to switch names. XSD>1.0 has a <xs:keyref> tag for that. I'll just add comments about these, then let's21:35
duck2see if I'll be able to implement them21:35
mithroduck2: As you come across these ideas, can you log an issue about them?21:38
duck2mithro: ok21:40
mithroduck2: Also need to log those github issues...21:44
duck2mithro: right. I looked at the schemas first so that I can see if we need more features than what is used in the current ones.21:48
mithroduck2: I added some issues to your repository...21:53
duck2mithro: checking them now22:00
*** alexhw has quit IRC23:32

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