| *** tpb has joined #yosys | 00:00 | |
| *** Degi_ has joined #yosys | 01:21 | |
| *** develonepi3 has quit IRC | 01:22 | |
| *** Degi has quit IRC | 01:23 | |
| *** Degi_ is now known as Degi | 01:23 | |
| *** futarisIRCcloud has quit IRC | 01:48 | |
| *** s_frit has quit IRC | 02:02 | |
| *** s_frit has joined #yosys | 02:02 | |
| *** s_frit has quit IRC | 03:23 | |
| *** s_frit has joined #yosys | 03:24 | |
| *** Forty-Bot has joined #yosys | 03:31 | |
| *** m4ssi has joined #yosys | 04:21 | |
| *** futarisIRCcloud has joined #yosys | 04:21 | |
| *** m4ssi has quit IRC | 05:21 | |
| *** vidbina has joined #yosys | 05:51 | |
| *** s_frit has quit IRC | 05:56 | |
| *** kraiskil has joined #yosys | 06:12 | |
| *** kraiskil has quit IRC | 06:48 | |
| *** emeb_mac has quit IRC | 07:04 | |
| *** lansiir has joined #yosys | 07:45 | |
| *** oldtopman has quit IRC | 07:46 | |
| *** cr1901_modern has quit IRC | 08:29 | |
| *** futarisIRCcloud has quit IRC | 10:10 | |
| *** develonepi3 has joined #yosys | 11:05 | |
| *** mmicko has quit IRC | 11:21 | |
| *** mmicko has joined #yosys | 11:21 | |
| *** TD-Linux has quit IRC | 11:22 | |
| *** TD-Linux has joined #yosys | 11:22 | |
| *** vidbina has quit IRC | 11:27 | |
| *** s_frit has joined #yosys | 11:52 | |
| *** vidbina has joined #yosys | 11:59 | |
| *** proteusguy has joined #yosys | 12:22 | |
| *** s_frit has quit IRC | 12:41 | |
| *** s_frit has joined #yosys | 12:42 | |
| *** s_frit has quit IRC | 13:10 | |
| *** s_frit has joined #yosys | 13:11 | |
| *** srk has quit IRC | 13:38 | |
| *** srk has joined #yosys | 13:38 | |
| *** vidbina has quit IRC | 13:40 | |
| *** vidbina has joined #yosys | 13:46 | |
| *** s_frit has quit IRC | 13:52 | |
| *** s_frit has joined #yosys | 13:53 | |
| *** kristianpaul has quit IRC | 14:02 | |
| *** kristianpaul has joined #yosys | 14:05 | |
| agg | gatecat: do you mind if i ask a nextpnr q? i'm back poking ecp5 dsp support and trying to do something about packing, i can locate the mults that feed each alu in pack_dsp and add relative constraints to them which should position them correctly wrt the alu, but then the placer doesn't create an initial placement for them (since they have contr_parent set), and then errors computing hpwl because those cells | 14:21 |
|---|---|---|
| agg | aren't present in cell_locs | 14:21 |
| agg | instead of modelling this as adding constraints to the mult cells, should i instead be thinking of packing the mults into the alu and getting rid of them from the placement operation entirely? | 14:21 |
| gatecat | No, I think constraints is the correct approach here | 14:22 |
| agg | I also tried changing the placer to run update_all_chains after the initial placement, which does create a placement for the mults | 14:22 |
| agg | but doesn't assign them a bel, so something later fails | 14:22 |
| agg | not sure if I should "need" to change placer_heap though, or just feed it something better | 14:22 |
| gatecat | I think it probably needs cellGroups to be set up: https://github.com/daveshah1/nextpnr-xilinx/blob/xilinx-upstream/xilinx/arch.cc#L657-L659 | 14:22 |
| gatecat | (this is an example of what you'd need to do to set one up before going into the placer) | 14:23 |
| gatecat | containing both the mult and alu cell types | 14:23 |
| agg | ah interesting, ok | 14:23 |
| agg | does sa not need them? | 14:23 |
| gatecat | if that doesn't work, point me to the code and I'll have a look | 14:23 |
| agg | will do, thanks! | 14:23 |
| gatecat | no, sa works on a cell by cell basis only and doesn't treat different cell types differently at different times | 14:23 |
| agg | should I expect sa to already work if I've just set the relative constraints, then? I only tried heap | 14:24 |
| gatecat | yeah, that's a good test to check the constraints are being set correctly | 14:24 |
| agg | at least the constraints now check my manual placement is valid, if I change the manual bel assignment the constraint check fails first | 14:25 |
| agg | will give it a go though, cheers | 14:25 |
| *** emeb_mac has joined #yosys | 14:31 | |
| *** cr1901_modern has joined #yosys | 14:45 | |
| *** m4ssi has joined #yosys | 16:34 | |
| *** vidbina has quit IRC | 16:43 | |
| *** kraiskil has joined #yosys | 17:23 | |
| *** vidbina has joined #yosys | 18:53 | |
| *** AkechiShiro has quit IRC | 20:29 | |
| *** kraiskil has quit IRC | 20:41 | |
| *** vidbina has quit IRC | 21:57 | |
| *** X-Scale` has joined #yosys | 22:39 | |
| *** bwidawks has joined #yosys | 22:40 | |
| *** phantomcircuit_ has joined #yosys | 22:40 | |
| *** dys has quit IRC | 22:41 | |
| *** phantomcircuit has quit IRC | 22:41 | |
| *** ZipCPU has quit IRC | 22:41 | |
| *** Ekho has quit IRC | 22:42 | |
| *** jrawson has quit IRC | 22:42 | |
| *** jrawson has joined #yosys | 22:42 | |
| *** m4ssi has quit IRC | 22:42 | |
| *** X-Scale has quit IRC | 22:43 | |
| *** bwidawsk has quit IRC | 22:43 | |
| *** Lofty has quit IRC | 22:43 | |
| *** Max-P has quit IRC | 22:43 | |
| *** bwidawks is now known as bwidawsk | 22:43 | |
| *** X-Scale` is now known as X-Scale | 22:43 | |
| *** Ekho has joined #yosys | 22:51 | |
| *** Max-P has joined #yosys | 23:02 | |
| *** lf has quit IRC | 23:44 | |
| *** lf has joined #yosys | 23:45 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!