#milkymist IRC log for Sunday, 2012-08-26

--- Sun Aug 26 201200:00
sh4rm4wolfspra1l, did you know that you can push multiple commits at once ? you dont have to push every single commit you make...11:47
sh4rm4just saying since my github startpage is now entirely cluttered by your fpga-tools work :)11:47
sh4rm4additionally if you dont push immediately you can still rebase stuff, ie. commit1, commit2, commit3 which fixes commit1, so 3 and 1 can be rebased into a single commit11:49
sh4rm4if you already pushed, you can't do this without -f11:49
sh4rm4http://i.imgur.com/LdZ3I.png11:55
larscbut it doesn't make sense to horde commits localy11:57
larsci mean they are several hous appart11:57
sh4rm4i have a very long passphrase for my ssh key11:59
sh4rm4for this reason alone i let the commits accumulate before i push...11:59
sh4rm4and it's not like anybody's waiting eagerly for your commits to show up :)12:00
wolfspra1ljust unfollow12:00
wolfspra1l:-)12:00
sh4rm4:)12:00
wolfspra1lI like to push things out because my hdd is making strange clicking sounds, quite simple12:00
larscI eargerly await any of wolfspra1l's commits ;)12:01
wolfspra1lthen I feel good about every line that is in safety12:01
sh4rm4ugh, i see...12:01
wolfspra1lI tried to reinstall a new fedora on a new hdd, but that failed due whatever new kernel issues, so I couldn't continue to waste time with that and just went back to the hdd that includes the nerve wracking clicking12:01
wolfspra1lactually I like it, creates a sense of urgency12:01
wolfspra1ldo not expect my push frequency to go down12:02
wolfspra1l:-)12:02
sh4rm4:P12:02
wolfspra1lunless my hdd breaks and I'm busy with a reinstall for 2 days, or I finally switch to the new one ;-)12:02
wolfspra1lit will slow down soon since my daughter returns12:03
wolfspra1land I am still dancing around the switch boxes/routing which are quite difficult to understand, not difficult as in moon landing, but difficult as in 'damned, why can't they just publish some papers on how it's wired up..."12:03
wolfspra1lsh4rm4: btw without hard feelings, that's a hilarious comment to make. push frequency too high, clutters my view...12:07
wolfspra1lI'm not in any extreme here, coding a bit with a few pushs a day at most. that's already too much? are we in a senior home?12:07
sh4rm4well, i just wanted to share the point of view of a github follower12:08
wolfspra1labsolutely, thanks for bringing it up12:08
larscwolfspra1l: how does a switch box work?12:08
wolfspra1lI would have never ever thought anyone cares or notices12:08
sh4rm4i unfollowed other projects for this reason12:08
wolfspra1lit's really the clicking thing here...12:08
wolfspra1lnot sure, lots of multiplexers I think, mostly 4:1 or maybe 5:112:09
wolfspra1lthe sets of multiplexers are sort of understand, but there's still stuff missing for me to make sense of the bits12:09
wolfspra1lI think 80% or more of the bits in the bitstream initialize those multiplexers or maybe decoders12:09
wolfspra1lbut who knows I am really just scratching the surface and that becomes clearer as I go, which is not bad12:11
wolfspra1lfor example - even if I can make sense of all bits and the model, that only covers the digital knowledge of the chip. what remains totally invisible to me is any physical knowledge, wire widths, whatever other physical properties/differences between physical devices12:12
wolfspra1lI'm sure there are a lot of them, and they will remain invisible to me for good, unless I work my way through the chip's physical structure12:12
wolfspra1lthat will create a big problem later, as I think you cannot just route every signal over every wire, and group/align them at will. without a deeper understand of the physics a lot of those designs will not work well or at all.12:13
wolfspra1lbut I just ignore all that for now12:13
wolfspra1lmaybe there is a slim chance we can discover some of that in-situ, i.e. by running analysis designs on the chip itself, measuring, etc.12:14
wolfspra1lone by one...12:14
wolfspra1la lot more commits hopefully coming :-)12:14
wpwrakthe main parameters should be fan in/out and delay13:33
wpwrakonce you have a few things working, it should be possible to get some usable estimates for all those things13:33
lekernelwolfspra1l: have you seen fig. 3 of http://www.mn.uio.no/ifi/english/research/projects/cosrecos/publications/paper/fpl10beckhoff.pdf ?13:42
lekernelthe paper also gives a lot of info on how to make sense of routing configuration bits13:43
wpwraknice topic (shorts in fpgas) :)14:37
GitHub131[migen] sbourdeauducq pushed 1 new commit to master: https://github.com/milkymist/migen/commit/4164fb4ac93a77cff1a777a52efe8a728bebec0319:24
GitHub131[migen/master] bus/csr: configurable data width - Sebastien Bourdeauducq19:24
--- Mon Aug 27 201200:00

Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!