| *** tpb has joined #yosys | 00:00 | |
| *** Degi_ has joined #yosys | 00:50 | |
| *** Degi has quit IRC | 00:50 | |
| *** Degi_ is now known as Degi | 00:50 | |
| agg | gatecat: in https://github.com/YosysHQ/yosys/pull/677/commits/b65932edc4a00860cafc04f962f2a08dd782b075 a bunch of seemingly-valid parameters for mult18x18d and alu54b were removed, can you remember why? as far as I can tell they're all accepted by diamond for synthesis (and are described in the documentation) | 00:56 |
|---|---|---|
| agg | did it just turn out they don't have any effect on synthesis or something? could it still make sense to accept them so that code which targets diamond and specifies them continues to work on yosys? | 00:57 |
| *** danvet has joined #yosys | 04:52 | |
| *** srk has joined #yosys | 05:55 | |
| *** m4ssi has joined #yosys | 07:24 | |
| *** m4ssi has quit IRC | 07:31 | |
| gatecat | agg: iirc they might not have actually changed something? | 07:50 |
| *** srk has quit IRC | 10:07 | |
| agg | gatecat: as far as I can tell most of them do _something_, e.g. MULT_BYPASS on the MULT18 errors if the associated ALU MUX bits in the opcode aren't set to the right input and otherwise it changes a bunch of CIB_DSP bits, most of the register bits also change some number of CE/RST bits but not in quite the same way they do for some other registers, that sort of thing | 10:20 |
| agg | a lot of them seem to depend on other attributes, including between connected mult+alu, or they error which might have screwed up fuzzing? | 10:20 |
| agg | i couldn't hand-on-heart promise they all do anything meaningful yet though :p | 10:21 |
| agg | except maybe REG_OPCODEOP1_1_CLK which I think was just missing all along | 10:21 |
| *** vidbina has joined #yosys | 10:24 | |
| *** forrestv has quit IRC | 11:37 | |
| *** forrestv has joined #yosys | 11:39 | |
| *** adjtm_ has joined #yosys | 11:41 | |
| *** adjtm has quit IRC | 11:42 | |
| *** m4ssi has joined #yosys | 11:44 | |
| *** X-Scale` has quit IRC | 11:52 | |
| *** AdamHorden has quit IRC | 12:03 | |
| *** AdamHorden has joined #yosys | 12:13 | |
| *** X-Scale has joined #yosys | 12:28 | |
| *** m4ssi has quit IRC | 12:28 | |
| *** forrestv has quit IRC | 13:00 | |
| *** vidbina has quit IRC | 13:37 | |
| *** vidbina has joined #yosys | 14:08 | |
| *** Raito_Bezarius has quit IRC | 15:03 | |
| *** Raito_Bezarius has joined #yosys | 15:15 | |
| *** vup has quit IRC | 15:20 | |
| *** vup has joined #yosys | 15:20 | |
| *** peepsalot has joined #yosys | 15:54 | |
| *** peepsalot has quit IRC | 16:58 | |
| agg | gatecat: do you think it would be worth re-adding them to yosys, at least so it matches the docs, even if nextpnr was just set to error on non-default settings? at the least i want to add opcodeop1_1_clk, but could restore the rest, and it seems like eventually some of them could be supported in nextpnr too | 17:23 |
| agg | happy to pr, i have a branch ready either way :p | 17:23 |
| gatecat | Yes, that seems reasonable to me | 17:24 |
| agg | thanks, will pr today | 17:26 |
| agg | btw, placement of mult+alu seems to be all working in every situation now | 17:26 |
| agg | but I still need to figure out a bunch of cib_dsp bits to get it to work properly | 17:27 |
| agg | (well, except if you manually constrain the alu but not the mult heap fails, or if you manually constrain the mult but not the alu, it places the alu itself and moves the mult, ignoring the constraint, so I added a specific error for that, but doesn't seem like a problem) | 17:27 |
| *** m4ssi has joined #yosys | 18:08 | |
| *** vidbina has quit IRC | 19:27 | |
| *** ZipCPU has quit IRC | 20:00 | |
| *** ZipCPU has joined #yosys | 20:06 | |
| *** danvet has quit IRC | 20:33 | |
| *** peepsalot has joined #yosys | 20:53 | |
| *** m4ssi has quit IRC | 21:38 | |
| *** lf has quit IRC | 23:27 | |
| *** lf has joined #yosys | 23:31 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!