| *** tpb has joined #yosys | 00:00 | |
| *** mwk has quit IRC | 01:21 | |
| *** mwk has joined #yosys | 01:21 | |
| *** mwk has quit IRC | 01:25 | |
| *** dys has joined #yosys | 01:31 | |
| *** mwk has joined #yosys | 01:32 | |
| *** dys has quit IRC | 01:49 | |
| *** mwk has quit IRC | 01:53 | |
| *** FL4SHK has quit IRC | 02:07 | |
| *** mwk has joined #yosys | 02:18 | |
| *** FL4SHK has joined #yosys | 02:21 | |
| *** mwk has quit IRC | 02:23 | |
| *** PyroPeter has quit IRC | 02:35 | |
| *** PyroPeter has joined #yosys | 02:47 | |
| *** futarisIRCcloud has joined #yosys | 03:14 | |
| *** gsi_ has quit IRC | 03:14 | |
| *** gsi_ has joined #yosys | 03:15 | |
| *** emeb_mac has joined #yosys | 03:23 | |
| *** gsi__ has joined #yosys | 03:44 | |
| *** gsi_ has quit IRC | 03:47 | |
| *** analognoise has joined #yosys | 03:58 | |
| *** mwk has joined #yosys | 04:23 | |
| *** mwk has quit IRC | 04:28 | |
| *** mwk has joined #yosys | 05:00 | |
| *** mwk has quit IRC | 05:05 | |
| *** mwk has joined #yosys | 05:08 | |
| *** _whitelogger has quit IRC | 06:04 | |
| *** _whitelogger has joined #yosys | 06:06 | |
| *** emeb_mac has quit IRC | 06:40 | |
| *** rohitksingh has joined #yosys | 07:17 | |
| *** futarisIRCcloud has quit IRC | 07:24 | |
| *** _whitelogger has quit IRC | 07:28 | |
| *** _whitelogger has joined #yosys | 07:30 | |
| *** analognoise has quit IRC | 07:33 | |
| *** futarisIRCcloud has joined #yosys | 08:01 | |
| *** rohitksingh has quit IRC | 08:04 | |
| *** dys has joined #yosys | 08:24 | |
| *** rohitksingh has joined #yosys | 08:30 | |
| *** _whitelogger has quit IRC | 08:46 | |
| *** _whitelogger has joined #yosys | 08:48 | |
| *** rohitksingh has quit IRC | 08:57 | |
| *** rohitksingh has joined #yosys | 09:10 | |
| *** _whitelogger has quit IRC | 09:19 | |
| *** _whitelogger has joined #yosys | 09:21 | |
| *** rohitksingh has quit IRC | 09:38 | |
| *** rohitksingh has joined #yosys | 09:38 | |
| *** togo has joined #yosys | 09:43 | |
| *** futarisIRCcloud has quit IRC | 10:14 | |
| *** futarisIRCcloud has joined #yosys | 13:41 | |
| *** gsi__ is now known as gsi_ | 14:52 | |
| *** emeb has joined #yosys | 14:56 | |
| *** _whitelogger has quit IRC | 15:28 | |
| *** _whitelogger has joined #yosys | 15:30 | |
| rrika | hi, I just want to confirm something, I had a verilog array of size "[3:0] name [0:2047]" with one write (name[a] <= ...), and two reads ({name[b], name[b+1]}) which didn't end up using my ecp5's ram | 15:47 |
|---|---|---|
| rrika | it did however use them when doing only one read ({name[b], name[b]}) | 15:47 |
| rrika | so is that because of the number of read ports? | 15:47 |
| daveshah | Yosys can create more read ports by duplicating the BRAM | 15:48 |
| rrika | automatically or on explicit instruction, in this case splitting suffices | 15:49 |
| daveshah | However, a separate <= for each read port is probably required | 15:49 |
| rrika | ah, I wrote b and b+1 but really it was {b, 1'b0} and {b, 1'b1}, so it's really clear they never alias | 15:49 |
| rrika | daveshah, an assign is not good enough? | 15:49 |
| daveshah | I think the concatenation might be the problem | 15:49 |
| daveshah | Something like x[3:0] <= name[b]; x[7:4] <= name[b+1]; should work | 15:50 |
| daveshah | Yosys needs to merge a DFF into the read port to map sync BRAM | 15:51 |
| daveshah | With the concatenation, a single 8-bit DFF is created and I don't know if it can split that. With the two assigns, two 4-bit DFFs will be created that should merge fine | 15:52 |
| *** citypw has quit IRC | 15:53 | |
| rrika | daveshah, that would still mean though that the contents are duplicated, right? | 15:56 |
| *** citypw has joined #yosys | 15:57 | |
| daveshah | Yes, Yosys doesn't support mixed read/write widths which would be the most efficient solution | 15:58 |
| daveshah | The other possible solution would be to use one of the two ports of the BRAM as both read and write if read and write are exclusive, but Yosys doesn't support this either | 15:59 |
| *** futarisIRCcloud has quit IRC | 16:01 | |
| *** citypw has quit IRC | 16:04 | |
| *** dkozel has joined #yosys | 16:40 | |
| tnt | mmm ... having a long flatten wire x[(32*N)-1:0] representing N wires of 32 bits, how would you 'OR' them all together ? (for any N) | 16:51 |
| emeb | wire flat = |x; ? | 16:53 |
| tnt | well ... no ... I want a vector of 32 bits at the end. | 16:54 |
| tnt | every i-th bit of each of the N vector ORed together. | 16:55 |
| emeb | ah - or in only one dimension | 16:55 |
| emeb | sounds like a good excuse to use a for loop | 16:56 |
| tnt | yeah, but my attempts to do so, haven't resulted in anything I can build :/ | 16:57 |
| tnt | err ... I swapped +: and :+ :/ | 17:00 |
| emeb | pebkac :) | 17:04 |
| *** sxpert has quit IRC | 17:30 | |
| *** sxpert has joined #yosys | 17:39 | |
| *** maikmerten has joined #yosys | 18:04 | |
| *** sxpert has quit IRC | 18:06 | |
| *** analognoise has joined #yosys | 18:15 | |
| *** MoeIcenowy has quit IRC | 18:23 | |
| *** MoeIcenowy has joined #yosys | 18:23 | |
| *** sxpert has joined #yosys | 19:52 | |
| *** rohitksingh has quit IRC | 20:03 | |
| *** sxpert has quit IRC | 20:09 | |
| *** sxpert has joined #yosys | 20:15 | |
| *** sxpert has quit IRC | 20:22 | |
| *** sxpert has joined #yosys | 20:24 | |
| *** maikmerten has quit IRC | 21:17 | |
| *** kraiskil has joined #yosys | 21:20 | |
| *** _whitelogger has quit IRC | 21:34 | |
| *** _whitelogger has joined #yosys | 21:36 | |
| *** kraiskil has quit IRC | 21:43 | |
| *** togo has quit IRC | 21:53 | |
| *** emeb_mac has joined #yosys | 22:16 | |
| *** analognoise has quit IRC | 22:53 | |
| *** danieljabailey has joined #yosys | 23:08 | |
Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!