#milkymist IRC log for Tuesday, 2013-03-26

GitHub55[migen] sbourdeauducq pushed 1 new commit to master: http://git.io/6jPy7A11:06
GitHub55migen/master 574becc Sebastien Bourdeauducq: fhdl/specials: clean up clock domain handling11:06
GitHub73[mibuild] sbourdeauducq pushed 1 new commit to master: http://git.io/S6APqA16:58
GitHub73mibuild/master 74cc4d2 Sebastien Bourdeauducq: generic_platform: remove obj in request + add lookup_request16:58
GitHub29[milkymist-ng] sbourdeauducq pushed 1 new commit to master: http://git.io/DyVp4w16:58
GitHub29milkymist-ng/master 1e860c7 Sebastien Bourdeauducq: Use new Mibuild generic_platform API16:58
GitHub33[mibuild] sbourdeauducq pushed 1 new commit to master: http://git.io/AZGzUA18:19
GitHub33mibuild/master 3b19dfc Sebastien Bourdeauducq: Support for platform info18:19
_florent_hi21:08
_florent_it seems some small change are missing in the last mibuild commit21:09
_florent_http://git.io/MgMhuA21:09
_florent_*changes21:09
lekernelah yes22:05
lekernelthanks22:05
GitHub173[mibuild] sbourdeauducq pushed 2 new commits to master: http://git.io/bYmrHQ22:10
GitHub173mibuild/master 38e92eb Sebastien Bourdeauducq: altera_quartus: fix clock domain name22:10
GitHub173mibuild/master 8cf7c96 Sebastien Bourdeauducq: crg: use new platform.request22:10
aerislekernel do you have a web site presenting migen ?22:10
lekernelhttp://milkymist.org/3/migen.html22:10
_florent_about removing obj parameter in request method, what was the motivation?22:15
lekernelplatform info structures22:15
_florent_(I don't find it very practial in fact...)22:16
lekerneleg when you have differential signals and some are inverted to make PCB layout easier, you can add information in the resource about which signals were inverted22:16
_florent_since we are now doing distinction between modules that have IOs and others that don't have IOs22:16
lekernelalso you don't need to duplicate information such as the number of NOR flash address bit - the norflash core gets that from the object it gets from mibuild22:18
lekernelbtw the number of lines of code of both mibuild and milkymist-ng has decreased since this change ...22:19
lekernelmaybe I can put the DRAM timing info into the platform description too, will have to think of that22:21
_florent_but about simulation22:21
_florent_suppose I want to simulate the new uart with pads parameters22:21
_florent_I will have to provide a objects with all the rx, tx signals22:21
_florent_the uart is maybe not a good example22:22
_florent_since it only has 2 ios...22:22
lekernelwell yes of course, you need to provide a model for the device you are connecting to the core under test :)22:22
lekerneland that model can easily provide such signals22:23
lekernelyou could even pass the model object as 'pads' parameter22:23
_florent_yes of course22:23
_florent_but if you get the uart core alone without it's environment and you want to simulate it you will have to look in the code where pads is used22:25
lekernelthat can be solved with a comment22:26
lekernelthe main idea of this change really is to make cores adapt to slight platform variations, such as inverted signals, number of address/data bits, etc.22:26
lekernelotherwise you need to specify this info twice22:26
_florent_ok I see now the advantage on this point22:27
_florent_I will try to adapt my de0-nano to see if I change my mind ;)22:28
lekernelcool22:33
lekernelsorry, migen/mibuild is one big experimental construction site and I break stuff regularly...22:34
lekernelbtw have you made up your mind about icalepcs?22:34
_florent__no problem for the experimental construction22:42
_florent__just that I don't have enough time my designs...22:42
_florent__*to adapt22:43
_florent__for icaleps I don't know yet sorry22:43
--- Wed Mar 27 201300:00

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