Saturday, 2021-04-03

*** tpb has joined #yosys00:00
*** lutsabound has joined #yosys00:43
*** Degi_ has joined #yosys01:13
*** Degi has quit IRC01:15
*** Degi_ is now known as Degi01:15
*** cr1901_modern has quit IRC02:33
*** cr1901_modern has joined #yosys02:34
promach3yosys does not support $ceil    ?02:48
promach3yosys has the following error for https://github.com/promach/DDR/blob/main/ddr3_memory_controller.v#L106102:52
promach3ERROR: Found error in internal cell \ddr3_memory_controller.$ge$ddr3_memory_controller.sv:1061$73 ($ge) at kernel/rtlil.cc:991:02:52
*** aquijoule_ has joined #yosys03:30
*** richbridger has quit IRC03:32
*** lutsabound has quit IRC03:56
*** m4ssi has joined #yosys06:02
*** kraiskil has joined #yosys06:07
*** emeb_mac has quit IRC07:21
*** kraiskil has quit IRC08:29
*** kraiskil has joined #yosys09:51
mwkpromach3: well you found a bug10:17
mwkhttps://github.com/YosysHQ/yosys/issues/271810:17
mwkas a workaround, I think assigning the constant to an intermediate const-valued wire would help10:18
*** proteusguy has quit IRC10:31
*** proteusguy has joined #yosys10:45
promach3mwk: what do you mean by **intermediate** const-valued wire  ?11:10
*** kraiskil has quit IRC11:22
mwkpromach3: actually now that I look at it, I think the comparison isn't synthesizable verilog in the first place11:49
* promach3 < https://matrix.org/_matrix/media/r0/download/matrix.org/bvPlxqpuksYtjrBbBfnxTjad/message.txt >11:50
mwkbecause it involves a floating point comparison11:50
promach3yup, your suggested countermeasure also does not work11:50
promach3<mwk "because it involves a floating p"> I suppose this is also a bug ?11:51
mwkand the proper fix would be to write "localparam [31:0] TIME_TRFC = $ceil(...);"11:51
mwkso that it'll be cast to integer11:51
mwkno, this is not a bug11:51
mwkit's just that floating point operations are not synthesizable11:51
mwkbecause they don't map to logic circuits without major effort11:52
promach3you mean add `[31:0]` ?11:52
mwkyes11:52
promach3let me try11:52
mwkand btw by intermediate wire I meant something like "wire [31:0] tmp = 2.0; assign O = A < tmp;"11:54
mwkbut adding the type directly on localparam is superior11:54
promach3ok, cool thanks mwk11:56
promach3mwk:  ok, now I could pass yosys check without errors.  How do I pass yosys synthesis result to Xilinx ISE for place-and-route for Spartan-6 ?12:02
mwkthere's an example in yosys tree12:03
mwkhttps://github.com/YosysHQ/yosys/tree/master/examples/mimas212:03
mwkyou'll need to write your own UCF of course, but the same synthesis script should work12:04
mwk(well, that and fix the part number)12:04
promach3ok, thanks mwk12:17
promach3mwk: are you familiar with DDR ?  May I ask some advice on setting constraints for `DQ` and `DQS` signals ?12:25
promach3they are both `inout` signals12:26
*** srk has quit IRC13:35
*** Kamilion has quit IRC14:06
*** Kamilion has joined #yosys14:07
*** kraiskil has joined #yosys14:09
*** unkraut has quit IRC14:30
*** unkraut has joined #yosys14:33
*** AdamHorden has quit IRC14:42
*** emeb has joined #yosys14:43
promach3ok, found https://www.cnblogs.com/lelin/p/12652460.html14:45
tpbTitle: DDR接口时序实例 - 春风一郎 - 博客园 (at www.cnblogs.com)14:45
promach3mwk: Why the example you had given just now does not include SDC constraint file ?16:08
mwkthis is xilinx ISE, you generally use UCF constraints16:15
mwkor at least they were enough for my little example16:15
*** cr1901_modern has quit IRC16:35
*** cr1901_modern has joined #yosys16:35
*** emeb_mac has joined #yosys18:03
*** futarisIRCcloud has joined #yosys18:20
*** FabM has joined #yosys18:38
*** FabM has quit IRC19:15
*** show1 has quit IRC19:35
*** m4ssi has quit IRC19:43
*** kraiskil has quit IRC19:52
*** lf has quit IRC23:40
*** lf has joined #yosys23:40

Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!